Water Front

Flood Recovery, Innovation
and Response IV
WIT Press publishes leading books in Science and Technology.

Visit our website for new and current list of titles.
www.witpress.com
WITeLibrary
Home of the Transactions of the Wessex Institute.
Papers presented at FRIAR 2014 are archived in the WIT eLibrary
in volume 184 of WIT Transactions on Ecology and the Environment (ISSN: 1743-3541).
The WIT eLibrary provides the international scientific community with immediate and
permanent access to individual papers presented at WIT conferences.
Visit the WIT eLibrary at www.witpress.com.
Fourth International Conference on
Flood Recovery, Innovation and Response
FRIAR 2014
Conference Chairmen
D. Proverbs
University of the West of England, UK
C.A. Brebbia
Wessex Institute of Technology, UK
International Scientific Advisory Committee
C. Booth
D. De Wrachien
H. Hashimoto
M. Holicky
S. Mambretti
D. Mioc
M. Mohssen
D. Molinari
Organised by
Sponsored by
WIT Transactions on Ecology and the Environment
International Journal of Safety and Security Engineering
WIT Transactions
Transactions Editor
Carlos Brebbia
Wessex Institute of Technology
Ashurst Lodge, Ashurst
Southampton SO40 7AA, UK
Editorial Board
B Abersek University of Maribor, Slovenia G Belingardi Politecnico di Torino, Italy

Y N Abousleiman University of Oklahoma, R Belmans Katholieke Universiteit
USA Leuven, Belgium
K S Al Jabri Sultan Qaboos University, C D Bertram The University of New
Oman South Wales, Australia
E Alarcon Universidad Politecnica de D E Beskos University of Patras, Greece
Madrid, Spain S K Bhattacharyya Indian Institute of
C Alessandri Universita di Ferrara, Italy Technology, India
D Almorza Gomar University of Cadiz, E Blums Latvian Academy of Sciences,
Spain Latvia
B Alzahabi Kettering University, USA J Boarder Cartref Consulting Systems,
J A C Ambrosio IDMEC, Portugal UK
A M Amer Cairo University, Egypt B Bobee Institut National de la Recherche
S A Anagnostopoulos University of Scientifique, Canada
Patras, Greece H Boileau ESIGEC, France
M Andretta Montecatini, Italy M Bonnet Ecole Polytechnique, France
E Angelino A.R.P.A. Lombardia, Italy C A Borrego University of Aveiro,
H Antes Technische Universitat Portugal
Braunschweig, Germany A R Bretones University of Granada,
M A Atherton South Bank University, UK Spain
A G Atkins University of Reading, UK J A Bryant University of Exeter, UK
D Aubry Ecole Centrale de Paris, France F-G Buchholz Universitat
J Augutis Vytautas Magnus University, Gesanthochschule Paderborn,
Lithuania Germany
H Azegami Toyohashi University of M B Bush The University of Western
Technology, Japan Australia, Australia
A F M Azevedo University of Porto, F Butera Politecnico di Milano, Italy
Portugal W Cantwell Liverpool University, UK
J M Baldasano Universitat Politecnica de D J Cartwright Bucknell University, USA
Catalunya, Spain P G Carydis National Technical University
J G Bartzis Institute of Nuclear of Athens, Greece
Technology, Greece J J Casares Long Universidad de
S Basbas Aristotle University of Santiago de Compostela, Spain
Thessaloniki, Greece M A Celia Princeton University, USA
A Bejan Duke University, USA A Chakrabarti Indian Institute of Science,
M P Bekakos Democritus University of India
Thrace, Greece
J-T Chen National Taiwan Ocean M Domaszewski Universite de
University, Taiwan Technologie de Belfort-Montbeliard,
A H-D Cheng University of Mississippi, France
USA J Dominguez University of Seville, Spain
J Chilton University of Lincoln, UK K Dorow Pacific Northwest National
C-L Chiu University of Pittsburgh, USA Laboratory, USA
H Choi Kangnung National University, W Dover University College London, UK
Korea C Dowlen South Bank University, UK
A Cieslak Technical University of Lodz, J P du Plessis University of Stellenbosch,
Poland South Africa
S Clement Transport System Centre, R Duffell University of Hertfordshire, UK
Australia N A Dumont PUC-Rio, Brazil
M W Collins Brunel University, UK A Ebel University of Cologne, Germany
J J Connor Massachusetts Institute of G K Egan Monash University, Australia
Technology, USA K M Elawadly Alexandria University,
M C Constantinou State University of Egypt
New York at Buffalo, USA K-H Elmer Universitat Hannover, Germany
D E Cormack University of Toronto, D Elms University of Canterbury, New
Canada Zealand
D F Cutler Royal Botanic Gardens, UK M E M El-Sayed Kettering University, USA
W Czyczula Krakow University of D M Elsom Oxford Brookes University, UK
Technology, Poland F Erdogan Lehigh University, USA
M da Conceicao Cunha University of D J Evans Nottingham Trent University,
Coimbra, Portugal UK
L Dávid Károly Róbert College, Hungary J W Everett Rowan University, USA
A Davies University of Hertfordshire, UK M Faghri University of Rhode Island, USA
M Davis Temple University, USA R A Falconer Cardiff University, UK
A B de Almeida Instituto Superior M N Fardis University of Patras, Greece
Tecnico, Portugal P Fedelinski Silesian Technical University,
E R de Arantes e Oliveira Instituto Poland
Superior Tecnico, Portugal H J S Fernando Arizona State University,
L De Biase University of Milan, Italy USA
R de Borst Delft University of Technology, S Finger Carnegie Mellon University, USA
Netherlands E M M Fonseca Instituto Politécnico de
G De Mey University of Ghent, Belgium Bragança, Portugal
A De Montis Universita di Cagliari, Italy J I Frankel University of Tennessee, USA
A De Naeyer Universiteit Ghent, Belgium D M Fraser University of Cape Town,
P De Wilde Vrije Universiteit Brussel, South Africa
Belgium M J Fritzler University of Calgary, Canada
D De Wrachien State University of Milan, T Futagami Hiroshima Institute of
Italy Technology, Japan
L Debnath University of Texas-Pan U Gabbert Otto-von-Guericke Universitat
American, USA Magdeburg, Germany
G Degrande Katholieke Universiteit G Gambolati Universita di Padova, Italy
Leuven, Belgium C J Gantes National Technical University
S del Giudice University of Udine, Italy of Athens, Greece
G Deplano Universita di Cagliari, Italy L Gaul Universitat Stuttgart, Germany
I Doltsinis University of Stuttgart, A Genco University of Palermo, Italy
Germany N Georgantzis Universitat Jaume I, Spain
P Giudici Universita di Pavia, Italy
L M C Godinho University of Coimbra, N Ishikawa National Defence Academy,
Portugal Japan
F Gomez Universidad Politecnica de J Jaafar UiTm, Malaysia
Valencia, Spain W Jager Technical University of Dresden,
R Gomez Martin University of Granada, Germany
Spain Y Jaluria Rutgers University, USA
D Goulias University of Maryland, USA C M Jefferson University of the West of
K G Goulias Pennsylvania State England, UK
University, USA P R Johnston Griffith University, Australia
F Grandori Politecnico di Milano, Italy D R H Jones University of Cambridge, UK
W E Grant Texas A & M University, USA N Jones University of Liverpool, UK
S Grilli University of Rhode Island, USA N Jovanovic CSIR, South Africa
R H J Grimshaw Loughborough University, D Kaliampakos National Technical
UK University of Athens, Greece
D Gross Technische Hochschule N Kamiya Nagoya University, Japan
Darmstadt, Germany D L Karabalis University of Patras, Greece
R Grundmann Technische Universitat A Karageorghis University of Cyprus
Dresden, Germany M Karlsson Linkoping University, Sweden
A Gualtierotti IDHEAP, Switzerland T Katayama Doshisha University, Japan
O T Gudmestad University of Stavanger, K L Katsifarakis Aristotle University of
Norway Thessaloniki, Greece
R C Gupta National University of J T Katsikadelis National Technical
Singapore, Singapore University of Athens, Greece
J M Hale University of Newcastle, UK E Kausel Massachusetts Institute of
K Hameyer Katholieke Universiteit Leuven, Technology, USA
Belgium H Kawashima The University of Tokyo,
C Hanke Danish Technical University, Japan
Denmark B A Kazimee Washington State University,
K Hayami University of Tokyo, Japan USA
Y Hayashi Nagoya University, Japan S Kim University of Wisconsin-Madison,
L Haydock Newage International Limited, USA
UK D Kirkland Nicholas Grimshaw & Partners
A H Hendrickx Free University of Brussels, Ltd, UK
Belgium E Kita Nagoya University, Japan
C Herman John Hopkins University, USA A S Kobayashi University of Washington,
I Hideaki Nagoya University, Japan USA
D A Hills University of Oxford, UK T Kobayashi University of Tokyo, Japan
W F Huebner Southwest Research D Koga Saga University, Japan
Institute, USA S Kotake University of Tokyo, Japan
J A C Humphrey Bucknell University, USA A N Kounadis National Technical
M Y Hussaini Florida State University, University of Athens, Greece
USA W B Kratzig Ruhr Universitat Bochum,
W Hutchinson Edith Cowan University, Germany
Australia T Krauthammer Penn State University,
T H Hyde University of Nottingham, UK USA
M Iguchi Science University of Tokyo, C-H Lai University of Greenwich, UK
Japan M Langseth Norwegian University of
D B Ingham University of Leeds, UK Science and Technology, Norway
L Int Panis VITO Expertisecentrum IMS, B S Larsen Technical University of
Belgium Denmark, Denmark
F Lattarulo Politecnico di Bari, Italy K McManis University of New Orleans,
A Lebedev Moscow State University, USA
Russia A C Mendes Universidade de Beira
L J Leon University of Montreal, Canada Interior, Portugal
D Lesnic University of Leeds, UK R A Meric Research Institute for Basic
D Lewis Mississippi State University, USA Sciences, Turkey
S lghobashi University of California Irvine, J Mikielewicz Polish Academy of
USA Sciences, Poland
K-C Lin University of New Brunswick, N Milic-Frayling Microsoft Research Ltd,
Canada UK
A A Liolios Democritus University of R A W Mines University of Liverpool, UK
Thrace, Greece C A Mitchell University of Sydney,
S Lomov Katholieke Universiteit Leuven, Australia
Belgium K Miura Kajima Corporation, Japan
J W S Longhurst University of the West A Miyamoto Yamaguchi University, Japan
of England, UK T Miyoshi Kobe University, Japan
G Loo The University of Auckland, New G Molinari University of Genoa, Italy
Zealand T B Moodie University of Alberta, Canada
J Lourenco Universidade do Minho, D B Murray Trinity College Dublin, Ireland
Portugal G Nakhaeizadeh DaimlerChrysler AG,
J E Luco University of California at San Germany
Diego, USA M B Neace Mercer University, USA
H Lui State Seismological Bureau Harbin, D Necsulescu University of Ottawa,
China Canada
C J Lumsden University of Toronto, F Neumann University of Vienna, Austria
Canada S-I Nishida Saga University, Japan
L Lundqvist Division of Transport and H Nisitani Kyushu Sangyo University,
Location Analysis, Sweden Japan
T Lyons Murdoch University, Australia B Notaros University of Massachusetts,
Y-W Mai University of Sydney, Australia USA
M Majowiecki University of Bologna, Italy P O’Donoghue University College Dublin,
D Malerba Università degli Studi di Bari, Ireland
Italy R O O’Neill Oak Ridge National
G Manara University of Pisa, Italy Laboratory, USA
S Mambretti Politecnico di Milano, Italy M Ohkusu Kyushu University, Japan
B N Mandal Indian Statistical Institute, G Oliveto Universitá di Catania, Italy
India R Olsen Camp Dresser & McKee Inc.,
Ü Mander University of Tartu, Estonia USA
H A Mang Technische Universitat Wien, E Oñate Universitat Politecnica de
Austria Catalunya, Spain
G D Manolis Aristotle University of K Onishi Ibaraki University, Japan
Thessaloniki, Greece P H Oosthuizen Queens University,
W J Mansur COPPE/UFRJ, Brazil Canada
N Marchettini University of Siena, Italy E L Ortiz Imperial College London, UK
J D M Marsh Griffith University, Australia E Outa Waseda University, Japan
J F Martin-Duque Universidad A S Papageorgiou Rensselaer Polytechnic
Complutense, Spain Institute, USA
T Matsui Nagoya University, Japan J Park Seoul National University, Korea
G Mattrisch DaimlerChrysler AG, Germany G Passerini Universita delle Marche, Italy
F M Mazzolani University of Naples F Patania University of Catania, Italy
“Federico II”, Italy B C Patten University of Georgia, USA
G Pelosi University of Florence, Italy W Roetzel Universitaet der Bundeswehr
G G Penelis Aristotle University of Hamburg, Germany
Thessaloniki, Greece V Roje University of Split, Croatia
W Perrie Bedford Institute of R Rosset Laboratoire d’Aerologie, France
Oceanography, Canada J L Rubio Centro de Investigaciones
R Pietrabissa Politecnico di Milano, Italy sobre Desertificacion, Spain
H Pina Instituto Superior Tecnico, Portugal T J Rudolphi Iowa State University, USA
M F Platzer Naval Postgraduate School, S Russenchuck Magnet Group,
USA Switzerland
D Poljak University of Split, Croatia H Ryssel Fraunhofer Institut Integrierte
H Power University of Nottingham, UK Schaltungen, Germany
D Prandle Proudman Oceanographic S G Saad American University in Cairo,
Laboratory, UK Egypt
M Predeleanu University Paris VI, France M Saiidi University of Nevada-Reno, USA
I S Putra Institute of Technology Bandung, R San Jose Technical University of
Indonesia Madrid, Spain
Y A Pykh Russian Academy of Sciences, F J Sanchez-Sesma Instituto Mexicano
Russia del Petroleo, Mexico
F Rachidi EMC Group, Switzerland B Sarler Nova Gorica Polytechnic,
M Rahman Dalhousie University, Canada Slovenia
K R Rajagopal Texas A & M University, S A Savidis Technische Universitat Berlin,
USA Germany
T Rang Tallinn Technical University, A Savini Universita de Pavia, Italy
Estonia G Schmid Ruhr-Universitat Bochum,
J Rao Case Western Reserve University, Germany
USA R Schmidt RWTH Aachen, Germany
J Ravnik University of Maribor, Slovenia B Scholtes Universitaet of Kassel,
A M Reinhorn State University of New Germany
York at Buffalo, USA W Schreiber University of Alabama, USA
G Reniers Universiteit Antwerpen, Belgium A P S Selvadurai McGill University,
A D Rey McGill University, Canada Canada
D N Riahi University of Illinois at Urbana- J J Sendra University of Seville, Spain
Champaign, USA J J Sharp Memorial University of
B Ribas Spanish National Centre for Newfoundland, Canada
Environmental Health, Spain Q Shen Massachusetts Institute of
K Richter Graz University of Technology, Technology, USA
Austria X Shixiong Fudan University, China
S Rinaldi Politecnico di Milano, Italy G C Sih Lehigh University, USA
F Robuste Universitat Politecnica de L C Simoes University of Coimbra,
Catalunya, Spain Portugal
J Roddick Flinders University, Australia A C Singhal Arizona State University,
A C Rodrigues Universidade Nova de USA
Lisboa, Portugal P Skerget University of Maribor, Slovenia
F Rodrigues Poly Institute of Porto, J Sladek Slovak Academy of Sciences,
Portugal Slovakia
G R Rodríguez Universidad de Las Palmas V Sladek Slovak Academy of Sciences,
de Gran Canaria, Spain Slovakia
C W Roeder University of Washington, A C M Sousa University of New
USA Brunswick, Canada
J M Roesset Texas A & M University, H Sozer Illinois Institute of Technology,
USA USA
D B Spalding CHAM, UK E Van den Bulck Katholieke Universiteit
P D Spanos Rice University, USA Leuven, Belgium
T Speck Albert-Ludwigs-Universitaet D Van den Poel Ghent University, Belgium
Freiburg, Germany R van der Heijden Radboud University,
C C Spyrakos National Technical Netherlands
University of Athens, Greece R van Duin Delft University of
I V Stangeeva St Petersburg University, Technology, Netherlands
Russia P Vas University of Aberdeen, UK
J Stasiek Technical University of Gdansk, R Verhoeven Ghent University, Belgium
Poland A Viguri Universitat Jaume I, Spain
G E Swaters University of Alberta, Canada Y Villacampa Esteve Universidad de
S Syngellakis Wessex Institute of Alicante, Spain
Technology, UK F F V Vincent University of Bath, UK
J Szmyd University of Mining and S Walker Imperial College, UK
Metallurgy, Poland G Walters University of Exeter, UK
S T Tadano Hokkaido University, Japan B Weiss University of Vienna, Austria
H Takemiya Okayama University, Japan H Westphal University of Magdeburg,
I Takewaki Kyoto University, Japan Germany
C-L Tan Carleton University, Canada J R Whiteman Brunel University, UK
E Taniguchi Kyoto University, Japan T W Wu University of Kentucky, USA
S Tanimura Aichi University of Z-Y Yan Peking University, China
Technology, Japan S Yanniotis Agricultural University of
J L Tassoulas University of Texas at Athens, Greece
Austin, USA A Yeh University of Hong Kong, China
M A P Taylor University of South B W Yeigh SUNY Institute of Technology,
Australia, Australia USA
A Terranova Politecnico di Milano, Italy J Yoon Old Dominion University, USA
A G Tijhuis Technische Universiteit K Yoshizato Hiroshima University, Japan
Eindhoven, Netherlands T X Yu Hong Kong University of Science
T Tirabassi Institute FISBAT-CNR, Italy & Technology, Hong Kong
S Tkachenko Otto-von-Guericke- M Zador Technical University of Budapest,
University, Germany Hungary
N Tosaka Nihon University, Japan K Zakrzewski Politechnika Lodzka, Poland
T Tran-Cong University of Southern M Zamir University of Western Ontario,
Queensland, Australia Canada
R Tremblay Ecole Polytechnique, Canada G Zappalà CNR-IAMC, Italy
I Tsukrov University of New Hampshire, R Zarnic University of Ljubljana, Slovenia
USA G Zharkova Institute of Theoretical and
R Turra CINECA Interuniversity Computing Applied Mechanics, Russia
Centre, Italy N Zhong Maebashi Institute of
S G Tushinski Moscow State University, Technology, Japan
Russia H G Zimmermann Siemens AG, Germany
J-L Uso Universitat Jaume I, Spain R Zainal Abidin Infrastructure University
Kuala Lumpur(IUKL), Malaysia
Flood Recovery, Innovation
and Response IV
Editors
D. Proverbs
C.A. Brebbia
Editors:
D. Proverbs
C.A. Brebbia
Published by
WIT Press
Ashurst Lodge, Ashurst, Southampton, SO40 7AA, UK
Tel: 44 (0) 238 029 3223; Fax: 44 (0) 238 029 2853
E-Mail: witpress@witpress.com
http://www.witpress.com
For USA, Canada and Mexico
Computational Mechanics Inc

25 Bridge Street, Billerica, MA 01821, USA
Tel: 978 667 5841; Fax: 978 667 7582
E-Mail: infousa@witpress.com
http://www.witpress.com
British Library Cataloguing-in-Publication Data
A Catalogue record for this book is available

from the British Library
ISBN: 978-1-84564-784-1
eISBN: 978-1-84564-785-8
ISSN: 1746-448X (print)
ISSN: 1743-3541 (on-line)
The texts of the papers in this volume were set individually by the authors or under their
supervision. Only minor corrections to the text may have been carried out by the publisher.
No responsibility is assumed by the Publisher, the Editors and Authors for any injury and/or
damage to persons or property as a matter of products liability, negligence or otherwise, or
from any use or operation of any methods, products, instructions or ideas contained in the
material herein. The Publisher does not necessarily endorse the ideas held, or views expressed
by the Editors or Authors of the material contained in its publications.
© WIT Press 2014

Printed in Great Britain by Lightning Source, UK.
All rights reserved. No part of this publication may be reproduced, stored in a retrieval
system, or transmitted in any form or by any means, electronic, mechanical, photocopying,
recording, or otherwise, without the prior written permission of the Publisher.
Preface
The present volume contains papers presented at the Fourth International

Conference on Flood Recovery Innovation and Response (FRIAR) held in Poznan,
Poland. The conference is jointly organised by the Wessex Institute of Technology,
UK, and the University of the West of England, Bristol, UK; sponsored by WIT
Transactions on Ecology and the Environment, and the International Journal of
Safety and Security Engineering.
FRIAR 2014 is the fourth Conference of this successful series. The conference
started at the Institute of Civil Engineers in London 2008 and was reconvened at
the Lombardy Region Headquarters in Milano in 2010 and in Dubrovnik in 2012.
Flooding is a global phenomenon that claims numerous lives worldwide each

year. This winter many parts of Europe have been affected by serious flooding
including several Italian cities such as Pisa, Florence and Rome and others in
Southern France. The UK has been very severely affected by an exceptional run of
winter storms, culminating in serious coastal damage and widespread, persistent
flooding. This record-breaking weather and flooding, has been exceptional in
its duration, and led to the wettest December to January period in the UK since
records began. Heavy rains combined with strong winds and high waves led
to widespread flooding and coastal damage, causing significant disruption to
individuals, businesses and infrastructure.
The damage caused by the flooding over the winter period is estimated to be
£1.1bn in the UK alone; but of course this does not reflect the longer term impacts
to lives and communities and businesses, who will be affected for many months
beyond the flooding itself. For some home owners and businesses, insurers will
assist in the recovery process by providing the necessary funding and services
to restore properties back to a habitable state. For others including those without
insurance, the recovery process will be very challenging indeed and it is likely that
many businesses will simply collapse as a consequence.
Research has shown that in the aftermath of the summer 2007 floods in the UK,
the vast majority of flood affected properties were reinstated to their previous
condition, leaving them equally vulnerable to future flood events. This goes against
the principles of climate change adaptation and represents a missed opportunity
to build back better and improve the resilience of homes and businesses that were
affected. Hopefully, the financial support now being made available to businesses
and homes in the UK will help to ensure resilient measures are installed during the
recovery process.
We know that it is impossible to entirely eliminate the risk from flooding and
that there is considerable uncertainty about future extreme weather patterns .
Clearly, further research is needed to improve our understanding of the challenges
associated with making our rural and urban environments and the communities
that exist within them, more resilient to the effects of flooding. This includes the
development of new innovative solutions as part of an integrated approach to flood
risk management at the community level. The complexity of these challenges
means that we need to work across disciplines and draw on a range of expertise,
recognising the use of both structural and non-structural measures towards arriving
at novel solutions to suit local circumstances.
The conference provided a forum for researchers, academics and practitioners

actively involved in improving our understanding of flood events and new
approaches to response, recovery and resilience. The meeting brought together
social scientists, surveyors, engineers, scientists, and other professionals from
many countries involved in research and development activities in a wide range of
technical and managerial topics related to flooding and its impacts on communities,
property and people. The conference drew together a wide range of experts
from across a range of disciplines and provided a very fertile platform for the
development of new ideas and solutions.
WIT Press, the publishing arm of the Wessex Institute has produced this volume
which is distributed around the world by its own offices in Europe and the USA
and an extensive distribution network. The book is produced in hard copy and
digital format to reach as many colleagues as possible. Furthermore, all conference
papers have been archived online in the Institute eLibrary (http://library.witpress.
com) where they are immediately and permanently available to the international
community.
The Editors are grateful to the authors for the quality of the papers published in
this book and particularly indebted to the members of the International Scientific
Advisory Committee and other colleagues who helped to select them, in this
manner ensuring their names the quality of this volume.
The Editors
Poznan
2014
Contents
Section 1: Flood modelling
A new approach for flood forecasting of river flows

M. Mohssen.......................................................................................................... 3
Agent-based modelling and inundation prediction to enable

the identification of businesses affected by flooding
G. Coates, G. I. Hawe, N. G. Wright & S. Ahilan .............................................. 13
A novel simple method for measuring the velocity of dam-break flow

P. B. Adegoke, W. Atherton & R. M. Al Khaddar .............................................. 23
Numerical simulation of the inundation area for landslide-induced

debris flow: a case study of the Sha-Xinkai gully in southern Taiwan
J.-C. Chen, J.-S. Wang, M.-R. Chuang & C.-J. Jeng ......................................... 35
Section 2: Risk assessment
A practical approach to floodplain mapping for large-scale

catastrophe models
I. Carnacina & A. Jemberie ............................................................................... 49
Vulnerability to flood risks in Japanese urban areas: crisis management

and emergency response for efficient evacuation management
M. Thomas & T. Tsujimoto ................................................................................ 61
Section 3: Flood management
Community-based flood risk management: lessons learned from the

2011 flood in central Thailand
N. Jukrkorn, H. Sachdev & O. Panya ................................................................ 75
Reservoir system operation using a diversion tunnel
J. Ji, H. Kim, M. Yu, C. Choi, J. Yi & J. Kang ................................................... 87
Section 4: Considering ‘Blue-Green’ approaches to

Flood Risk Management
(Special session organised by J. Lamond)
A conceptual framework for understanding behaviours and attitudes

around ‘Blue-Green’ approaches to Flood-Risk Management
G. Everett & J. Lamond ................................................................................... 101
Delivering and evaluating the multiple flood risk benefits in

Blue-Green Cities: an interdisciplinary approach
E. Lawson, C. Thorne, S. Ahilan, D. Allen, S. Arthur, G. Everett,
R. Fenner, V. Glenis, D. Guan, L. Hoang, C. Kilsby, J. Lamond,
J. Mant, S. Maskrey, N. Mount, A. Sleigh, L. Smith & N. Wright .................... 113
Modelling a green roof retrofit in the Melbourne Central Business District

S. J. Wilkinson, C. Rose, V. Glenis & J. Lamond............................................. 125
Section 5: Property-level flooding and health consequences

(Special session organised by C. A. Booth)
Improving the uptake of flood risk adaptation measures for

domestic properties in an insurance regime under transition
D. Cameron & D. Proverbs ............................................................................. 139
Waterproofing basement apartments: technical insights of a

new flood protection solution
D. W. Beddoes & C. A. Booth .......................................................................... 151
An investigation of patterns of response and recovery

among flood-affected businesses in the UK:
a case study in Sheffield and Wakefield
N. Bhattacharya-Mis & J. Lamond.................................................................. 163
Resilient reinstatement: what can we learn from

the 2007 flooding in England?
R. Joseph, D. Proverbs & J. Lamond .............................................................. 175
The role of flood memory in the impact of repeat flooding

on mental health
J. Lamond ........................................................................................................ 187
The long-term health impacts of repeated flood events
J. Stephenson, M. Vaganay, R. Cameron & P. Joseph .................................... 201
Section 6: State-of-the-art flooding-damage survey and assessment

(Special session organised by D. Molinari)
Implementing tools to meet the Floods Directive requirements:

a “procedure” to collect, store and manage damage data in the
aftermath of flood events
D. Molinari, M. Mazuran, C. Arias, G. Minucci, F. Atun
& D. Ardagna .................................................................................................. 215
Flood damage survey after a major flood in Norway 2013:

cooperation between the insurance business and a government agency
H. Berg, M. Ebeltoft & J. Nielsen .................................................................... 227
Section 7: Emergency preparedness and response
An overview of the applications for early warning and mapping

of the flood events in New Brunswick
D. Mioc, E. McGillivray, F. Anton, M. Mezouaghi, L. Mofford
& P. Tang ........................................................................................................ 239
Risk management and emergency response for a 300 km2 sub-sea level area
with a million citizens against extreme storm surge and flood due to the
“Super Ise-Bay Typhoon”
T. Tsujimoto, M. Igarashi & K. Kobayashi...................................................... 251
Multi-robot system for disaster area exploration

F. Burian, L. Zalud, P. Kocmanova, T. Jilek & L. Kopecny ............................ 263
Section 8: Adaptation to flood risk
Floating houses: an adaptation strategy for flood preparedness

in times of global change
P. Strangfeld & H. Stopp ................................................................................. 277
Design as a negotiation platform: new deals and spatial adaptation in

flood-prone areas
F. Rossano & L. Hobeica ................................................................................ 287
Author index .................................................................................................. 299

This page intentionally left blank
Section 1
Flood modelling
Flood Recovery, Innovation and Reponse IV 3
A new approach for flood forecasting of

river flows
M. Mohssen
Department of Environmental Management,
Lincoln University, New Zealand
Abstract
Flood warning mainly depends on reliable flood forecast models. Literature is
rich in flood modelling techniques, but failures of these models, especially on the
very short scale such as hourly flows, do often cause devastating impacts on
the communities affected by these floods, and on many occasions result in loss of
lives. This paper presents a new approach for flood forecasting of river flows
based on the projection theorem in Hilbert space.
The new modelling process obtains the projection of hourly flow rates on
hourly rainfalls over the catchment at previous hours to the projected flow rate. A
total of 25 flow events observed for the Leith River in Dunedin, New Zealand,
along with their corresponding observed rainfalls at two sites in the catchment
have been identified and applied to calibrate and validate the derived model. The
proposed modelling technique was capable of simulating the flow process for
the Leith River, and is a promising tool for flood forecast when other models
fail. The proposed model is easy to apply, doesn’t imply a lot of assumptions or
parameters, as other models usually require, and can be used for long term forecast
based on forecasted hourly rain one day or more before the event, or real time
forecast during the event itself based on rainfall which has been already gauged.
However, for real time (short term) forecast, the forecast time can be a few hours
based on the catchment area and its topography which can lead to a fast flow to
the outlet.
Keywords: flood forecast, flood modelling, rainfall-runoff, projection in Hilbert
Space.
WIT Transactions on Ecology and The Environment, Vol 184, © 2014 WIT Press
www.witpress.com, ISSN 1743-3541 (on-line)
doi:10.2495/FRIAR140011
4 Flood Recovery, Innovation and Response IV
1 Introduction
Natural disasters cause devastating damages to all types of lives on earth, and their
negative impacts can last for long periods with a huge cost to mitigate. Floods are
the most common natural disasters, and unlike other forms of natural disasters
which usually occur in specific regions such as earthquakes, volcanoes,
hurricanes, or tornadoes, floods occur almost everywhere, and no community is
immune from their devastating damages. Flood warning can be quite effective in
mitigating the impacts of a coming event, simply by getting prepared. Even with
the existence of flood protection schemes, there is usually the potential for a bigger
flooding event than what the scheme was designed for. The Environment Agency
of UK and the strategic plan for the US National Weather Service indicated the
urgent need for major investment to develop new forecast models for flood
warning [1, 2].
Flood forecasting is the corner stone for an efficient flood warning system.
New technology and the use of satellite and radar data have significantly improved
our capability of forecasting rainfall, even on an hourly basis, for short term
periods such as the next few hours or longer forecast such as the next 24/48 hours.
However, due to the complexity of this natural event, and the high spatial and
temporal variability of rain, the main driving force for flooding, in addition to the
complex hydrological aspects and characteristics of the catchment area, it is
usually hard to accurately forecast the coming flood event [1]. Many of the
available forecast models in the literature, especially those based on watershed
modelling and hydraulic/hydrologic routing, require a lot of data and include a lot
of assumptions for solving the concerned equations, which adds to their
complexity and applicability. There are many reports in the media and anger in
the communities over failure of their governments/authorities to provide proper
flood warning [3–7].
Time series analysis and modelling, such as ARIMA models, have been applied
in the literature for simulating streamflows. However, these models work more
for longer time periods where stationarity conditions can be assumed, or achieved
by removing apparent cycles or trends [8]. For hourly flows during a significant
event, the series is quite non-stationary, and flow rates react directly and are highly
related to the rainfall intensity during the period preceding this flow. ANN has
been recently applied for flood forecasting, and several techniques have been
suggested for their applications to hourly time steps [9, 10].
In New Zealand, floods are the most costly natural disaster. About 935
devastating floods occurred during the period 1920 to 1983 in New Zealand [11].
Dunedin is the second largest city in the South Island of New Zealand, with a
population of about 120,000 (Statistics New Zealand, 2013). The Leith River,
which drains about 45 km2 of mainly hilly areas around Dunedin, goes through the
city and passes by the prestigious University of Otago. Most of the northern part
of Dunedin lies within the flood plain for the Leith River. A big flooding event
for the Leith River can cause significant damage and loss to Dunedin in particular,
and the whole Otago Region in general. The Leith River has history of flooding,
and Dunedin experienced extensive damage and inundation during the 1877, 1923
and 1929 flood events. The Otago Regional Council has recently conducted
studies for flood protection schemes for the Leith River, and has applied for
consents to carry out the needed work.
2 Flood modelling of the Leith River

The Leith catchment has an area of about 45 km2, and extends on the north/north
west of Dunedin, with Lindsay Creek joining the main Leith in the northern side
of Dunedin. There are two sites for rainfall gauging, the first one is at Sullivans
Dam near the northern boundary of the catchment, while the second site is in the
northern Pinehill suburb of Dunedin. The flow site is located in the southern reach
of the river, near the University of Otago. The Leith River, after passing the
University of Otago, finds its way to the Otago harbour. Figure 1 shows the
catchment area with the locations of rainfall and flow sites. The Leith River has
an average flow of 0.694 m3/s, while its “observed” maximum flow is 114 m3/s,
recorded on 18 February 1991.
2.1 Model development and formulation
The catchment area of the Leith River is not big, and this usually results in a
significant component of the runoff contributing to its high flow hydrograph,
compared to the base flow component which is usually very small (as shown in
Fig. 2). Thus, the straight line approach for separating the base flow has been
applied to estimate the runoff hydrograph due to the rainfall event over the
catchment [12]. This approach should result in good estimates of the runoff
hydrograph, as any error in estimating this very small base flow will not have
effect on the much bigger runoff component. The runoff hydrograph is obtained
by simply subtracting the estimated base flow from the flow hydrograph.
The basic concept of this model is based on the projection in Hilbert Space [8]
of the hourly river flows on the span of hourly rainfall data preceding these flows.
This model represents an extension of the models developed by [13] and [14]
for the univariate and multivariate flood forecast of lake levels.
Thus, the flow rate at time t, Qt, is projected on the span of rainfalls at
antecedent times: Rt-j, j = L1 to L2, where L1 and L2 represent lag-1 and Lag-2
hours before time t. Thus:
∑ (1)
For to have the minimum distance “difference” from Qt, Qt - should be

orthogonal to all elements of the span of the vector R (Rt-j, j = L1 to L2). In Hilbert
space, this yields the following equation:
< Qt - , R >=0, where <X,Y> = E [XY] in Hilbert space (2)
Thus,
< Qt - ∑ , R > =0, j = L1, L1+1, …, L2 (3)
Figure 1: Locations of rainfall and flow sites in the Leith catchment.
Equation (3) produces a system of (L2 – L1 +1) linear equations, which can be
solved simultaneously to obtain the parameters , j = L1 to L2. The projection
theorem guarantees that the produced solution is the unique mapping of Qt onto R.
The projection theorem guarantees that the model provided by (3) will produce
coefficients of (Rt-j) for the best forecasts of Qt. It is assumed in this research that
the relationship between Qt and Rt-j is linear, which might not be the best choice.
However, based on the model application which is shown later, this proved to be
satisfactory. More research is recommended to consider alternative relationships.
60
50
Flow (m3/s)
40
30 Base Flow
20
10
0
11/02/05
12/02/05
13/02/05
14/02/05
15/02/05
16/02/05
17/02/05
Date
Figure 2: A high flow event for the Leith River showing the base flow.
2.2 Model calibration
A total of 25 high flow events have been selected from the available record during
the period March 2000 until November 2013. Twenty three events have been used
for model calibration and two events have been utilised to test the validation of the
developed model. Rainfall over the whole catchment was estimated by applying
Thiessen polygon method to calculate the weight for each rainfall site, and in turn
obtain the average rainfall over the whole catchment area. Thus, one time series
of average hourly rainfalls has been estimated and used in this case study. For the
calibration process, hourly lagged rainfalls for all the events were joined together
in one input file to the model so that the estimated parameters , j = L1 to L2 are
based on all the 23 events, and not only on one event. Figure 3 presents lagged
correlations between runoff flow rates and observed rainfalls at lags 0 to 10 hours
prior to the flow rate. The figure indicates that lags 3 and 4 are the highest, and it
is a must to include these rainfalls for the flood forecast of the Leith River. If L1
equals 3, then this will produce a 3 hours warning before this flow rate for a real
time forecast during the rainfall event. However, if this forecast is based on
rainfalls during the next day, the warning time would be much longer.
Figure 4 shows the observed versus the “forecasted” flows for the combined 23
events which were used in the calibration process.
In general, the model simulated “satisfactorily” the underlying hourly runoff
process, but underestimated some of the significantly high events, and also
overestimated others. However, it has to be stated that it is usually very hard for
any model on an hourly basis to simulate accurately the underlying hydrologic
process. Add to this, that this newly developed model does not “explicitly”
0.9
0.8
Correlation Coefficient
0.7
0.6
0.5
0.4
0.3
0.2
0.1
0
0 2 4 6 8 10 12
Lag (hrs)
Figure 3: Lagged cross correlations between runoff flow rates and rainfalls.
100
80
Flow (m3/s)
60
40 Observed
20 Forecasted
0
117
175
233
291
349
407
465
523
581
639
697
755
1
59
Time (hrs)
Figure 4: Results of the calibration process.
account for hydrologic abstractions, or losses from rainfall before it becomes

runoff. However, this is imbedded in the estimated parameters to obtain the best
match between the observed and the forecasted flows. Still, the model performed
reasonably well. The overall value for Filliben correlation coefficient, which is a
measure of how good are the forecasted flows compared to the observed ones,
is 0.9.
3 Model testing
Validation of the fitted model was carried out by applying the model to rainfall
events which were not included in its calibration process. Thus, these estimated
parameters are not “biased” toward these events. These two high flow events
occurred during the periods 30 July to 1 August 2008 and 15 to 20 June 2013.
Table 1 shows the forecasted peak flows versus the observed ones for the two
events, while figures 5 and 6 show the simulation of the fitted model to forecast
hourly flows for the two events. The table shows that the forecasted peak flows
were within 12% to 23% of the observed peaks, with determination coefficients
(R2) and Filliben correlation coefficients (FC) higher than 90%. It should be noted
that each event has two peaks, and the model was capable of capturing this
behaviour for the second event, but was not able to “properly” simulate the second
Table 1: Observed vs. forecasted Leith River peak flows.
Event Peak Observed Forecast R2 FC %

Date Error
July 1 26.4 23.2 -12.1
2008 2 28.4 21.9 0.97 0.98 -22.9
June 1 49.2 37.8 -23.2
2013 2 18.5 15.7 0.91 0.96 -15.3
30
25
20
Flow (m3/s)
15
Observed
Forecasted
10
0
1/08/08
1/08/08
30/07/08
31/07/08
31/07/08
31/07/08
Date
Figure 5: Observed vs. forecasted runoff hydrographs for the rainfall event
July 2008.
60
50
40
Flow (m3/s)
30
Observed
Forecasted
20
10
0
15/06/13
16/06/13
17/06/13
18/06/13
19/06/13
20/06/13
21/06/13
Date
Figure 6: Observed vs. forecasted runoff hydrograph for the rainfall event
June 2013.
peak of the first event. There is only one determination coefficient and one
Filliben correlation coefficient for each event, as shown in the table. The figures
confirm the conclusion that the model is capable of forecasting the Leith River
high flows, and responded well to the rising limb and the recession of the two
events.
4 Conclusions
A newly derived approach to forecast river flows based on the projection theorem
in Hilbert space has been presented and applied for the Leith River in Dunedin,
New Zealand. The model, once derived and calibrated, is easy to apply and can
be used for forecasting during a rainfall event with a lead time of 3 hours, or can
be used for a much longer time if forecasted rainfall is used. The model required
only hourly rainfall and flow data for its calibration, and only hourly rainfall data
for its application for flood forecast. Despite the fact that the model, in its current
form, does not “explicitly” has a function to account for hydrologic abstractions
from rainfall, still it produced satisfactorily results with its implicit inclusion of
rainfall losses during the projection process.
References
[1] Bye, P. & M. Horner, Easter 1998 Floods Report by the Independent
Review Team to the Board of the Environmental Agency, vol 1,
Environmental Agency, Bristol, 1998.
[2] Demeritt D., H. Cloke, F. Pappenberger, J. Thielen, J. Bartholmes & Maria-
Helena Ramoset, Ensemble predictions and perceptions of risk, uncertainty,
and error in flood forecasting, Environmental Hazards, vol 7, pp. 115-127,
2007.
[3] McClure, M & T. Howell, Forecast failure: how flood warning came too
late for southern Albertans, Calgary Herald, December 31, 1991.
[4] Datta, S., CWC failed to forecast, alert about floods, DNA, New Delhi, June
26, 2013.
[5] Socialist Equality Party (Australia), Australia’s floods: a failure of
government and the profit system, WSWB World socialist web site, January
29, 2011.
[6] Daily Express newspaper, Anger over flood warning failure, Daily Express
paper, London, July 9, 2012.
[7] Handmer, J., Are Flood Warnings Futile? Risk communication in
emergencies, the Australian Journal of Disaster and Trauma Studies,
2000–2.
[8] Brockwell, P. J. & R. A. Davis., Time Series: Theory and Methods,
Springer-Verlag New York Inc., pp. 46-51, 1991.
[9] Tiwari, M. K., Chatterjee, C., Development of an accurate and reliable
hourly flood forecasting model using wavelet–bootstrap–ANN (WBANN)
hybrid approach, J. of Hydrology 394, pp. 458-470, 2010.
[10] Chen-ShenHsien; Lin-YongHuang; Chang-LiChiu; Chang-FiJohn, The
strategy of building a flood forecast model by neuro-fuzzy network. Journal
of Hydrological Processes, 20(7), pp. 1525-1540, 2006.
[11] McSaveney, E., Floods – New Zealand’s number one hazard, Te Ara – the
Encyclopedia of New Zealand, updated 2-Mar-09.
[12] Chow, V.T., D. R. Maidment & L. W. Mays, Applied Hydrology McGraw-
Hill, 1988.
[13] Mohssen, M. and Goldsmith, M., Flood Forecasting of Lake Levels: A New
Concept. Int. J. of Safety and Security Eng., 1(4), pp. 363-375, 2011.
[14] Mohssen, M., A Multivariate Model for Flood Forecasting of Lake Levels.
Int. J. of Safety and Security Eng., 3(2), pp. 141-152, 2013.
Agent-based modelling and inundation

prediction to enable the identification of
businesses affected by flooding
G. Coates1, G. I. Hawe2, N. G. Wright3 & S. Ahilan3
1
Durham University, UK
2
University of Ulster, UK
3
University of Leeds, UK
Abstract
Flooding continues to cause significant disruption to individuals, organisations
and communities in many parts of the world. In terms of the impact on
businesses in the United Kingdom (UK), flooding is responsible for the loss of
millions of pounds to the economy. As part of a UK Engineering and Physical
Sciences Research Council funded project on flood risk management, SESAME,
research is being carried out with the aim of improving business response to and
preparedness for flood events. To achieve this aim, one strand of the research is
focused on establishing how agent-based modelling and simulation can be used
to evaluate and improve business continuity. This paper reports on the
development of the virtual geographic environment (VGE) component of an
agent-based model and how this has been combined with inundation prediction
to enable the identification of businesses affected by flooding in any urban area
of the UK. The VGE has been developed to use layers from Ordnance Survey’s
MasterMap, namely the Topography Layer, Integrated Transport Network
Layer and Address Layer 2. Coupling the VGE with inundation prediction
provides credibility in modelling flood events in any area of the UK. An initial
case study is presented focusing on the Lower Don Valley region of Sheffield
leading to the identification of businesses impacted by flooding based on a
predicted inundation. Further work will focus on the development of agents to
model and simulate businesses during and in the aftermath of flood events such
that changes in their behaviours can be investigated leading to improved
operational response and business continuity.
Keywords: floods, businesses, agent-based modelling and inundation prediction.
doi:10.2495/FRIAR140021
1 Introduction
In recent years, many parts of the world have experienced and suffered from
severe flooding which continues to cause significant disruption to individuals,
organisations and communities. In terms of the impact on UK businesses,
research conducted by the Environment Agency (EA) has estimated the financial
cost of floods in 2012 as being nearly £600 million [1]. Further, the EA has
indicated flooding cost an average of £60,000 for every business affected. Such
significant economic loss has led to flood risk management becoming high on
the political agenda. Consequently, means of reducing the economic impact of
interruptions attributable to flooding at the business level, and thus more widely,
are receiving growing attention. Effective business continuity management
(BCM) is recognised as one means of reducing the effect of flooding on business
operations and enabling a more rapid return to normality. Indeed, BCM is
viewed as an important tool for business survival in the face of a range of
disruptive events [2, 3], including flooding, and a key part of any successful
flood response [4]. The International Organization for Standardization’s ISO
22301, which is related to BCM, is described as the requirements which will help
organisations to be better prepared and handle disruptions of any type [5].
Despite the existence of such standards, in the UK, organisational engagement
with business continuity remains low with less take-up by Small and Medium
Enterprises (SMEs) relative to larger businesses and public sector bodies [6, 7].
The Engineering and Physical Sciences Research Council funded SESAME
project is related to organisational operational response and strategic decision
making for long term flood preparedness in urban areas [8]. The project aims to
create a unified framework of academic knowledge that can be used to influence
the behaviours of businesses, particularly SMEs, faced with flooding and flood
risk. This framework will assist businesses in understanding how they might
reduce the disruption and economic loss associated with flood events thus
strengthening their resilience to flooding and that of the wider economy. To
realise this aim, four interdisciplinary research objectives are being pursued: (i)
achieve a better understanding of how businesses behave in the immediate and
longer term aftermath of flood events; (ii) establish how agent-based modelling
and simulation can be used to assess the behaviours of different types of
businesses at risk of flooding; (iii) assess the impacts of flooding on economic
systems both within and beyond the immediately affected urban area and explore
how changes in businesses’ behaviour could influence these impacts; (iv)
develop and evaluate approaches promoting organisational behaviour change and
adaptive learning throughout the flood cycle. In order to achieve these research
objectives, the SESAME project brings together the academic fields of business
continuity management, agent-based modelling and simulation, flood modelling,
economic modelling and the social/behavioural sciences. This paper focuses on
the development of the virtual geographic environment (VGE) component of an
agent-based model along with how this has been brought together with
inundation prediction, via flood modelling, to identify businesses affected by
flooding in any urban area of the UK.
2 Related work
In the disaster management domain, research in the area of agent-based
modelling and simulation has focused on emergency response to major natural
and manmade events with agents representing emergency responders and/or
members of the public [9 –12]. In relation to flood risk management, agent-based
models have usefully been employed to model evacuation strategies involving
agents representing members of the public [13, 14]. However, despite the
concept of using agent-based modelling in business and organisation problems
[15–22], there is lack of research in the context of modelling businesses faced
with the challenges of ensuring business continuity when subjected to flooding.
This current dearth of research in agent-based modelling and simulation in the
context of business response to flooding offers scope for significant
contributions to knowledge to be made in relation to (i) identifying the specific
businesses affected by a flood event in a particular geographical area which can
then be modelled as agents, (ii) modelling these business agents’ actions and
interactions when responding to flood events based on field data gathered
through interviews with businesses at risk of flooding and/or which have
experience of flooding, and (iii) performing what-if analysis via agent-based
simulations of businesses’ responses to flood events in order to establish the
effect of changes in their behaviour and different approaches taken such as
adhering to flood plans.
3 Agent-based modelling and flood modelling

Agent-based modelling and simulation, coupled with inundation prediction via
flood modelling, is being used to enable the investigation of the organisational
behaviour of businesses when faced with flood events. An overview of the
modelling and simulation framework is presented in Figure 1.
The aim of this framework is to establish how agent-based modelling and
simulation can be used to improve organisational business continuity of different
types of UK businesses when responding to flooding by means of representing
their attributes and simulating their actions, interactions and dynamic behaviours.
Stage 1 of the framework involves developing the agent-based model’s VGE,
which is able to combine Ordnance Survey (OS) information with flood model
output, in Stage 2, in order to identify the businesses affected by flooding. Flood
model output can be static in the sense of providing a single-shot footprint of the
flood water in a geographical area, or dynamic in that the flood inundation varies
with time thus bringing a temporal aspect to simulations performed in Stage 5.
Stage 3 relates to developing agents to model businesses in terms of their
attributes, behaviour, actions and interactions in response to flood events.
Stage 4 involves setting-up an agent interaction framework to enable simulations
to be performed, in stage 5, thus informing businesses how they might change
their behaviour to better prepare for and respond to future flood events.
Agent-based modelling
1 3 4
Virtual Geographic Agent Interaction

Business Agents
Environment Framework
identifies implemented in
Static
2
Dynamic 5
Flood modelling Agent-based simulation
Figure 1: Overview of modelling and simulation framework.
3.1 Virtual geographic environment

Software has been developed to model a VGE of any region of the UK thus
providing the flexibility to credibly model flood events in any urban area and to
identify the businesses affected in that area. To achieve this aim, three layers of
OS MasterMap have been used [23]. Prior to indicating the information within
these layers relevant to the application of this work, it is appropriate to present
Figure 2, which shows the VGE for a case study area currently under
consideration, namely the Lower Don Valley region of Sheffield.
This region was identified as a suitable case study due to its high
concentration of SMEs from a range of sectors allied with their experience of
Figure 2: VGE for the Lower Don Valley region of Sheffield.
significant flooding. In June 2007, approximately 100mm of rainfall fell in 24

hours in the Lower Don Valley region having a devastating impact on more than
1000 businesses including key manufacturing companies with one suffering over
£15 million worth of damage [24].
In order to construct a VGE, use is made of OS MasterMap’s Topography
Layer, Integrated Transport Network Layer and Address Layer 2 to define and
populate the area under consideration with the relevant geographical information.
For any given geographical area: the Topography Layer is used to provide
information on individual buildings; the Integrated Transport Network (ITN)
Layer provides information on the road network; the Address Layer 2 provides
information on commercial properties including the precise location and the
identification of the associated building in the Topography Layer and road link
in the ITN Layer. Information in the Topography Layer and ITN Layer from
EDINA Digimap is freely available to academic institutions; however research
agreements with OS are required to obtain and use Address Layer 2 information.
3.2 Coupling the virtual geographic environment with flood modelling
For flood modelling, OS data sets are also required for the geographical area
under consideration, in this case the Lower Don Valley region of Sheffield. The
data sets obtained covered rivers, building features and a Digital Terrain Model.
From these data sets, the River Don’s centre line was delineated, the adjacent
floodplains were identified and a Geographic Information System (GIS) model
was built. Further, data from three hydrometric gauges in the River Don, near
the study region, were used to provide maximum water levels, which were then
interpolated across the region identified using the GIS model. These maximum
water depths at each (x, y) location within the area under consideration were
input to the VGE, which is shown in Figure 3 for Sheffield’s Lower Don Valley
region with the flood extent included.
Figure 3: VGE for the Lower Don Valley region of Sheffield with flood extent.
In Figure 3, the flood extent is represented coarsely due to being based on a

simplistic inundation prediction analysis. However, this representation is
adequate for the purpose of the initial case study. Future work will involve the
method used to determine maximum water depths being improved and model
data from the Lower Don Valley region being obtained from the EA.
3.3 Business agents
3.3.1 Identification of businesses affected by flooding

Once the geographical information from the three layers of OS MasterMap for
a specific area has been inputted to the software to create the VGE, along with
associated flood model output, a database is created holding information
associated with businesses in the flood affected area, which can be interrogated.
For example, businesses identified can be filtered according to industry type, and
then defined as the businesses to be modelled as agents. In addition to showing
the extent of flood water displayed in the VGE for the case study area, Figure 4
indicates the location of each business using a circle symbol, which is colour-
coded according to whether their associated building is flooded (red), or their
building is not flooded but their associated road link is flooded (orange), or
neither their associated building nor road link is flooded (green).
Figure 4: Identification of organisations affected by flooding.
Using the database created for the Lower Don Valley region of Sheffield,
4037 businesses were identified as being within the bounded area considered
with 531 of those businesses affected directly by flooding based on the
inundation prediction. In this context, “affected directly by flooding” signifies
that both the building, and the road link associated with the building, of a
particular business were under a depth of water greater than or equal to 1 mm.
Based on the inundation prediction, the depth of flood water can be established
for every business. For the Lower Don Valley region of Sheffield, Table 1
presents a profile of the number of businesses’ buildings and road links affected
by flooding in relation to depth of flood water.
Table 1: Number of businesses’ building and road link affected by flooding.
Water depth, Number of businesses’ Number of businesses’

d building affected by road link affected by
flooding flooding
d < 1m 64 40
1m  d < 2m 150 141
2m  d < 3m 108 138
d ≥ 3m 209 212
Using OS MasterMap Address Layer 2’s Valuation Office Agency Non-

domestic Rates Special Category (SCat) code, of which there are 360 categories
for classifying businesses, the most prominent types of businesses affected by
flooding in Sheffield’s Lower Don Valley region are listed in Table 2. While the
SCat classification is broad, importantly, Address Layer 2 provides the name of
each individual business allowing an accurate profile to be constructed in terms
of specific businesses affected by flooding. For example, under the category
‘Factories, Workshops and Warehouses’, heavy engineering businesses dominate
with a variety of companies manufacturing a range of products such as specialist
metals, wire meshes and weighing equipment. Also, in the ‘Offices’ category,
businesses include recruitment firms, property agents and event organisers.
Table 2: Classification of businesses affected by flooding.
SCat description SCat Number of

code businesses
Factories, Workshops and Warehouses 96 146
Offices 203 97
Shops 249 46
In addition to the business categories indicated in Table 2, 45 other business

types were affected by flooding such as vehicle repair workshops, food stores,
scrap metal yards, garages, car showrooms, cafes and takeaway food outlets.
It is recognised that as well as the SCat code in Address Layer 2, other
business classifications exist including Base Function (1500 functions available),
National Land Use Database Code (41 groups available) and Valuation Office
Agency Non-domestic Rates Primary Description (PDesc) Code (8 divisions and
108 sub-divisions available). Further, other strands of the SESAME project
related to business continuity and economic modelling employ the United
Nations Statistics Division’s International Standard Industrial Classification
(ISIC) of All Economic Activities, Rev.4, (21 high level activities and 99 sub-
activities) and the Cambridge Econometrics’ Multisectoral Dynamic Model
(MDM-E3) of the UK economy (46 industry types) respectively. Mapping
between these different classifications will be required once agent models are
developed in preparation for agent-based simulations to be performed.
3.3.2 Preliminary work on modelling business agents

Within the SESAME project, modelling business agents is being driven by
information extracted from transcripts of interviews with businesses at risk of
flooding and/or which have experienced flooding. To date, the focus has been on
businesses in the Lower Don Valley area of Sheffield given that in this
geographical area a significant number of businesses experienced flooding in
2007. A series of interviews has commenced involving businesses from a range
of sectors such that the attributes and behaviours to be defined for agents,
including their actions and interactions in response to flood events, are
representative of those organisations. Information extracted relates to attributes
such as business function, property, customers and suppliers. In terms of
behaviour, information extracted is based on experience of major disruptions, in
particular flooding during and post event, plans for business continuity both pre
and post event, and impact on business operations. Initial interviews have
revealed that SMEs appear not to rely on formal structures or have flood plans in
place should such a disruptive event occur. Rather, these businesses deal with
emergency situations, such as flooding, through improvisation.
4 Conclusions and future work

Flooding is the most common and widespread type of natural disaster in the UK.
For businesses, flooding poses a significant threat which can result in
interruption to operations and financial losses, as well as damage to property.
Thus, when faced with flooding and its effects, a business must know how to
ensure it is able to continue performing critical activities and maintain the
resources required to deliver its products and services in addition to protecting
staff and premises, and maintaining stock. By doing so, business disruption can
be reduced and recovery can be brought about more quickly.
The main aim of the SESAME project is to improve business response to and
preparedness for flood events. Research carried out to date on one strand of the
project has led to the capability to model any geographical urban area in the UK
and, based on inundation prediction via flood modelling, identify businesses
affected by flooding. An initial case study of the Lower Don Valley region of
Sheffield has enabled an accurate profile of businesses affected by flooding to be
constructed, with each business identified to be modelled as an agent. Future
work will focus on the design and development of business agents and an agent
interaction framework to enable simulations to be performed of businesses’
responses to flood events. More interviews will be held with a variety of
businesses in other geographical areas of the UK designated at risk of flooding
and/or which have experience of different types of flooding (fluvial, pluvial and
coastal), thus informing the attributes and behaviour of business agents. Also,
these interviews will be used to elicit types of potential behaviour changes a
business could make in terms of how it prepares for and/or responds to flood
events, which could subsequently be investigated using agent-based simulation.
Depending on the type of business being modelled, examples of such changes
could relate to: flood insurance; registering to receive EA flood warnings;
relocating key operations; training staff; installing flood barriers; making
premises more resilient to flooding using water resistant materials; developing a
flood plan; backing-up customer databases and electronic files; moving storage
areas out of reach of flood waters; being able to quickly move equipment,
computers, furniture, paper files, electrical items above ground level or to an
upper level of the building if possible; identifying alternative supply and
distribution routes; developing relationships with service and supply companies
in advance of flooding such that essential work can be undertaken rapidly to
quicken recovery and reduce business interruption.
Acknowledgement
The authors gratefully acknowledge the funding provided by the UK’s EPSRC
under grant EP/K012770/1.
References
[1] http://www.environment-agency.gov.uk
[2] Elliott, D., Herbane, B. & Swartz, E., Business Continuity Management,
Routledge: London, 2001.
[3] Herbane, B., The evolution of business continuity management: A
historical review of practices and drivers, Business History, 52(6), pp.
978–1002, 2010.
[4] Pitt, M., The Pitt Review: Lessons learned from the 2007 floods, Cabinet
Office, 2008.
[5] http://www.iso.org
[6] Musgrave, B. & Woodman, P., Weathering the Storm: The 2013 Business
Continuity Management Survey, Chartered Management Institute: London,
2013.
[7] Herbane, B., Small business research: Time for a crisis-based view,
International Small Business Journal, 28(1), pp. 43–64, 2010.
[8] Coates, G., Hawe, G.I., McGuinness, M., Wright, N.G., Guan, D., Harries,
T. & McEwen, L., A framework for organisational operational response
and strategic decision making for long term flood preparedness in urban
areas, Proceedings of the 3rd International Conference on Disaster
Management, 2013.
[9] Kitano, H. & Tadokoro, S., RoboCup Rescue: A Grand Challenge for
Multiagent and Intelligent Systems, Artificial Intelligence Magazine,
22(1), pp. 39–52, 2001.
[10] Mysore, V., Narzisi, G. & Mishra, B., Agent Modeling of a Sarin Attack
in Manhattan, Proceedings of the 1st International Workshop on Agent
Technology for Disaster Management in the 5th International Conference
on Autonomous Agents and Multi-Agent Systems, pp. 108–115, 2006.
[11] Bellamine-Ben Saoud, N., Ben Mena, T., Dugdale, J., Pavard, B. & Ben
Ahmed, M., Assessing large scale emergency rescue plans: an agent based
approach, International Journal of Intelligent Control and Systems:
Special Issue on Emergency Management Systems, 11(4), pp. 260–271,
2006.
[12] Hawe, G.I., Wilson, D.T., Coates, G. & Crouch, R.S., Investigating the
Effect of Overtriage on Hospital Arrival Times of Critically Injured
Casualties during a Major Incident using Agent-Based Simulation,
Proceedings of the 6th International Conference on Soft Computing and
Intelligent Systems and the 13th International Symposium on Advanced
Intelligent Systems, 2012.
[13] Liu, Y., Okada, N., Shen, D. & Li, S., Agent based flood evacuation
simulation of life-threatening conditions using Vitae system model,
Journal of Natural Disaster Science, 31(2), pp. 33–41, 2009.
[14] Dawson, R., Peppe, R. & Wang, M., An agent based model for risk-based
flood incident management, Natural Hazards, 59(1), pp. 167–189, 2011.
[15] Nagendra Prasad, M.V. & Chartier, D.A., Modeling Organizations using
Agent-based Simulations, Proceedings of the Workshop on Agent
simulation: Application, Models & Tools, pp. 54–66, 1999.
[16] Gilbert, N. & Terna, P., How to Build and Use Agent-based Models in
Social Sciences, Mind & Society, 1, pp. 57–72, 2000.
[17] Bonabeau, E., Agent-based modeling: Methods and techniques for
simulating human systems, Proceedings of the National Academy of
Sciences of the United States of America, 99(3), pp. 7280–7287, 2002.
[18] North, M. J. & Macal, C.M., Managing business complexity: discovering
strategic solutions with agent-based modeling and simulation, Oxford
University Press: Oxford, 2007.
[19] Gilbert, N., Agent-based models, In: UNSPECIFIED Quantitative
Applications in the Social Sciences, Sage Publications Inc., 2007.
[20] Dignum, V. & Tick, C., Agent-based Analysis of Organizations:
Formalization and Simulation, Proceedings of the IEEE/WIC/ACM
International Conference on Intelligent Agent Technology, pp. 244–247,
2007.
[21] Wilkinson, I.F., Marks, R.E. & Young, L., Toward Agent-based Models of
the Development and Evolution of Business Relations and Networks, In
Unifying Themes in Complex Systems - Volume VI: Proceedings of the 6th
International Conference on Complex Systems, ed. Ali A. Minai, Dan
Braha & Yaneer Bar-Yam, Springer, New York, pp. 414–421, 2010.
[22] Fioretti, G., Agent-Based Simulation Models in Organization Science,
Organizational Research Methods, 16(2), pp. 227–242, 2013.
[23] http://www.ordnancesurvey.co.uk
[24] Environment Agency, Review of 2007 summer floods, December 2007.
A novel simple method for measuring the

velocity of dam-break flow
P. B. Adegoke, W. Atherton & R. M. Al Khaddar
Built Environment, Liverpool John Moores University, UK
Abstract
The study of dam-break waves (DBW) is extremely important in providing the
information needed for risk assessment and management of coastal and riverine
areas. Adequate and acceptable preparedness for such an event to allow
mitigation of adverse impacts requires modelling of the flood as well as accurate
estimation of potential flood depths, flow velocities, and timing of the flood
arrival. This study investigated the effect of floodwater waves on various wall
surfaces and wall slopes in a 4.7m long wave tank by modelling a dam-break
phenomenon. The paper reports the novel simple methods (the Imaging System
(IS) and the sensor Signal Capture (SSC) technique) used for the estimation of
wave front propagation velocity which are the adaptations of the commonly used
Particle Image Velocimetry (PIV). The two techniques demonstrated good
agreement with the dam break wave theory as well as agreement between each
other. However, the SSC method with wave probes at a shorter separation
distance (0.41m apart) appears better and more in line with the results obtained
by previous investigators. The development represents a useful laboratory
scheme that is well suited for educational and initial research studies.
Keywords: dam-break, flow velocity, particle image velocimetry, flood waves.
1 Introduction
The concept of traditional flood protection is increasingly being replaced by
comprehensive risk management, which includes structural and non-structural
measures [1]. Hazard and risk maps are of particular importance for planning
purposes, risk awareness campaigns and the encouragement of private preventive
measures. Flood hazard risks are characterised by flood impact parameters such
as water depth and flow velocity. However, there has been a strong focus on
doi:10.2495/FRIAR140031
inundation depth as the main determinant for flood damage probably due to
limited information about other parameters characterising the flood, e.g. flow
velocity.
A systematic review of flood impacts on buildings and structures by Kelman
and Spence [2] revealed various damage mechanisms including hydrodynamic
actions related to waves and velocity as a result of turbulence. Dam-breaks have
been known for destroying buildings and infrastructures and also being
responsible for numerous losses of life in coastal and riverine areas. They
generally result in flash flood runoff in rivers and streams, debris flow surges
and tsunami run-up on dry coastal plains. In all these cases, the surge front is a
sudden discontinuity characterized by extremely rapid variations of flow depth
and velocity.
Flow velocity is generally presumed to influence flood damage. According to
Kreibich et al. [3] a significant influence of flow velocity on structural damage
could be shown in contrast to a minor influence on monetary losses and business
interruption. Forecasts of structural damage to road infrastructure is determined
to be based on flow velocity alone while the energy head is suggested as a
suitable flood impact parameter for reliable forecasting of structural damage to
residential buildings [3]. However, it is generally accepted that the higher the
flow velocity of the floodwater, the greater the probability (and extent) of
structural damage [4].
USACE [5] states that velocity is a major factor that could aggravate
structural and content damage during flooding events. High velocities limit the
time available for emergency measures and evacuation. The additional force of
high velocities creates greater danger of foundation collapse and forceful
destruction of contents [5]. For instance, Smith [6] states that a velocity of 3m/s
acting over a 1m depth will produce a force sufficient to exceed the design
capacity of a typical residential wall. The study shows further critical
combinations of water depth and flow velocities for building failure for three
different residential building types. These range from above 0.5m water depth
and 4m/s flow velocities to above 3m water depth with no flow velocity for
single storey weatherboard buildings [6].
The study of dam-break flow is important in providing vital information
needed for risk assessment and management of river valleys and coastal plains.
Such information may include useful data on dam-break flow variables such as
initial dam conditions, water depth downstream, flow velocity etc.
Moreover, physical modelling of dam-break waves is relatively limited. Most
predictions on dam-break waves are often based upon numerical predictions,
validated by limited data sets. According to Chanson [7] current knowledge of
dam-break waves in dry channels remains rudimentary despite a few available
studies.
In this paper, an experimental study of a dam-break flow is presented. Most
existing studies about dam-break flows are focused on variables such as
measurements of velocity profile and the water level using a Particle Tracking
Velocimetry (PTV) algorithm and/or Particle Image Velocimetry (PIV)
algorithm. However, in this study, an indirect way of measuring the flow
velocity was applied. The present study developed simple methods for estimating
instantaneous dam-break floodwater front velocity over the whole flow depth in
a dry channel using image acquisition techniques. The main feature of this
development is its simplicity that is well-suited to initial investigations.
2 Dam-break velocity and Imaging System: an overview

Ritter in 1892 was the first to investigate the dam break problem analytically [7].
His results have been used often for comparison of experimental and numerical
data. Ritter derived the velocity of the positive wave front based on shallow
water theory being twice the wave’s celerity co as:
2 2 (1)
where Ho is the initial height of the reservoir.

Lauber and Hager [8] as well as Stansby et al. [9] have been known for
carrying out recent experiments in the field of dam-break waves using digital
image processing. Stansby et al. [9] compared their experimental data to Stoker’s
analytical solution. It was found that solving Stoker’s equations for the positive
wave front of dam-break waves on dry horizontal beds leads to the same constant
front velocity as found by Ritter in eqn (1). Besides these investigations there
had been very few experimental works on dam-break waves in smooth horizontal
channels. This might be the result of high demand on measuring techniques
which has to be provided for extremely unsteady and speedy flow.
Recently, a range of novel experimental methods based on signal and image
analysis system have been developed for measuring flow velocities which are
particularly useful in unsteady flows such as those generated in dam-break
conditions. The data obtained can be used for the validation of numerical
computations. The techniques involve the flow field being illuminated with a
thin light sheet from a powerful source and might be filmed photographically or
digitally.
According to Adrian [10] once several particles appear in the illuminated
area, then the velocity vectors can be obtained for this area using tracking
algorithm techniques based on auto-correlation, cross-correlation or Young’s
fringe method. A typical set-up represented a flow seeded with particles which
could be imaged from above or through a transparent side-wall. The particles are
roughly identical and should appear brighter than the surrounding fluid on the
digital images.
The flow could be imaged from a single camera or from two cameras in a
stereoscopic arrangement. When the imaged scene is immersed in a liquid and
seen from the outside through a transparent wall, the image formation can be
strongly influenced by refractive effects. Each interface separating materials of
different refractive indexes will bend light rays according to Snell’s law [11].
Using imaging systems to obtain quantitative velocity flow field information
from particle movements encompasses a number of different methods depending
on the form of the capture image and the analysis technique employed. The
particle velocities are obtained as inter-frame displacements from the particle

positions using various methods. Such methods include Particle Tracking
Velocimetry (PTV), Particle Image Velocimetry (PIV) and Particle Streak
Velocimetry (PSV). In all the techniques, the displacement of the particles within
a field of view over a known period yields information about the velocity vector
field simultaneously over the whole plane.
PTV requires individual particles to be located in an image and successive
images to be recorded on successive frames and analyses pairs of single exposed
digital images to produce whole field maps of velocity vectors. The distance
travelled by an individual particle is then calculated and the velocity found
knowing the time interval between images. Various correlation algorithms to
allow the tracking of particles from frame to frame were described by Chegini et
al. [12] and Liem and Kongeter [13].
Also, the application of particle streak in fluid mechanics are often used for
qualitative flow visualisation as illustrated by Van Dyke [14] but the images
produced can be digitised for development into the quantitative measuring
technique known as PSV. This method is often used when the medium fluid has
a seeding particle concentration less than that for PTV and does not require
individual streak images to be overlapped and distinguished from each other. As
the individual streak lengths are determined and the exposure time is known, the
velocity associated with the particle streak can be obtained.
The PIV system consists of different optical components. Particles in the fluid
are illuminated in a plane by a light source. The light scattered by the particles is
recorded by a camera on a sequence of frames. In PIV, the average velocity
vectors are obtained for a cloud of particles based on image cross-correlation
techniques whereas for PTV the individual particle motions are resolved and full
sets of particle trajectories can be reconstructed by following the same particle
over many successive frames [15]. Many investigators that have used PIV or its
adaptations have employed the use of coloured droplets having specific gravities
close to unity (e.g. a mixture of carbon tetrachloride, xylene and zinc oxide). The
movements of these particles are then recorded on a cine film as waves pass
down the channel. Frame-by-frame analysis of the motions of the particles
allows the water-particle kinematics to be estimated. For further details on PIV,
literature such as Raffel et al. [16] or Chegini [17] may be reviewed.
However, in this study, water-particle velocity measurements were made
using a different adaptation of PIV. The time variations of the horizontal
components of the front edge of the floodwater were traced and located at
various positions from which the propagation velocities of the floodwater wave
were obtained using appropriate combination of Newton’s Equations of Motion.
3 Experimental work
This work was carried out in the Materials and Hydraulic Laboratory of the
School of the Built Environment, Liverpool John Moores University. A Low
Cost Wave Tank (LCWT) was primarily designed and constructed to simulate
dam failure in order to generate floodwater waves. The main aim was to
investigate the dissipation of energy of the floodwater waves in terms of impact

pressures on newly designed seawall models. However, the preliminary
experiments were conducted with a focus on the estimation of floodwater front
velocity as well as the characterisation of the flow in the channel. The floodwater
flow velocity is largely related to the impact pressures.
The test facility and the detailed laboratory arrangements are as shown in
Figure 1 and Figure 2 respectively. A series of tests were performed in a 4.70m
long, 0.40m wide, and 0.50m deep wave tank. The length of the reservoir was
1.0m while the propagating distance of the floodwater wave was 2.7 m (see
Figure 2). One side of the channel as well as its base was made from plywood
while the other side of the channel was made of clear Perspex which enabled
optical measuring video footage of the whole process (see Figure 1). RCD
protected lights were used in the process to improve visual observation and the
quality of video footage. The flow was imaged by a strategically positioned
camera through the side of the channel made of clear Perspex. A JVC TK –
1085E high-speed digital camera was used, acquiring grey-scale images at a rate
of 40 frames per second, with a resolution of 256 by 256 pixels.
Figure 1: Instrumented Low Cost Wave Tank (ILCWT).
Figure 2: Experimental set-up.
With the gate initially in position to create a dam, an upward impulse is

generated by releasing/pulling the rope through the pulley system. Tests were
conducted with the downstream channel completely dry prior to
experimentation, essentially modelling the dry beach common at urban
waterfronts [7]. Experiments were also conducted with a wet-bed downstream at
various ratios of upstream-downstream depths (Hds/Hus). Positive dam-break

waves downstream were exclusively considered throughout the study. The initial
levels of the water body within the reservoir were varied between 0.15m ≤ d o ≤
0.55m.
For wet-bed downstream experiments, downstream depths (Hds) of 0.05m,
0.10m and 0.15m were investigated. Within the available experimental facilities
any downstream depth higher than 0.15m did not give appreciable outcomes. At
the dead end, only the smooth surface wall model in vertical angle was chosen
for the trial experiments and each run of initial reservoir depth was repeated five
times to analyse the spread of data in terms of the time taken for the wave front
to impact the wall.
Flow patterns of the floodwater in the channel were visualized and video
footage recorded. The camera was strategically positioned to cover the entire
flow area of interest. The flow period between the two locations of interest
within the channel was obtained from the digitized image analysis. The
movement of front water within this field of view was then analysed. However,
the present study assumed the case by which the leading edge of floodwater is
captured rather than the seeding method. Two different approaches were then
used for the leading edge image capturing.
The first approach was by using the video system comparable to the PIV
method referred to as the Imaging System (IS). The second approach involved
using two suitably positioned wave probes (sensors) within the channel hence
termed Sensor Signal Capture (SSC) technique (see Figure 2). For the SSC
technique the two wave probes were placed at two different distances apart
(0.41m and 2.2m apart) to compare the results with that of the Imaging System.
The time at which each wave probe received signal of the leading flow was
deduced. Knowing the distance between the two wave probes, average front
water flow velocity was calculated using an appropriate combination of
equations.
For the IS, when the reservoir water depth do = 0.15m, propagation time
obtained was, t = 2.08s, the propagation distance is a constant value and is given
as, S = 2.7m (see Figure 2). Hence, the rate of acceleration of wave front a, as
well as the average floodwater front velocity v was then computed using
appropriate Newton’s equations of motion.
4 Results and discussions

Using this approach interesting results were obtained in terms of the wave front
velocity for dry-bed and wet-bed downstream conditions at varying reservoir
depths. Figure 3 depicts the variation of the obtained floodwater front velocity
against reservoir depth using the IS. The figure shows that the front velocity of
the floodwater increases with increased reservoir depth. The correlation
coefficient (R2) is 0.9811, indicating a strong relationship exists between the
velocity and the initial depth of water in the reservoir section. This linear
variation is expected from the analytical solution of one-dimensional frictionless
and horizontal dam-break flow problem developed by Ritter in 1892 [7].
The results of the front velocity with dry-bed and wet-bed downstream
conditions are shown in Table 1. Initial downstream water depths of 0.05m,
0.10m and 0.15m were investigated with varying initial reservoir depths and
compared (Table 1). Figure 4 emphasizes that the velocity decreases as the
downstream initial water depth increases. A dry-bed downstream gave some
unexpected results in this case. Figure 4 as well as visual and video analysis also
indicated that higher values and complexity of flow characteristics were obtained
for the lower downstream water depth case than for the case with higher
downstream water depths. Figure 4 also shows that the initial slope of the
velocity variation decreases as the downstream initial water depth increases. For
all depth ratios, the velocity profiles eventually became quite stable after the bore
developed downstream which is considered to be satisfactory for the downstream
subcritical flow region.
Figure 3: Variation of front water velocity with varying initial reservoir

depths for a dry-bed downstream.
Table 1: Computed front water velocity for dry-bed and wet-bed downstream
at varying reservoir depths.
Depth of
water in the Velocity, v Velocity, v Velocity, v Velocity, v
reservoir, do (dry-bed) (Hds=0.05m) (Hds=0.1m) (Hds=0.15m)
(m) (m/s) (m/s) (m/s) (m/s)
0.15 2.5962 2.70 2.17 2.06
0.25 3.3751 3.53 1.56 1.70
0.35 3.6000 4.50 1.23 1.47
0.45 4.1222 4.50 1.23 1.29
0.55 4.4628 5.19 1.07 1.23
Figure 4: Comparison of the front water velocities for dry- bed and wet-bed
at various initial water depths downstream.
The accuracy of the flow velocity relies on several factors. In the present
experiment the flow velocity is mainly associated with the precision of the time
interval between image pairs and the exactness of the displacement
measurement. Thus, floodwater front velocity was again computed using the
SSC technique described earlier to validate the reliability and accuracy of the IS.
Figure 5 compares the results of the two methods. The results obtained using the
two techniques indicated a good agreement with the dam-break wave theory
however; it was observed that the velocities obtained using the SSC method with
wave probes at a shorter distance away from each other (0.41m apart) appeared
to be closer in agreement to that obtained using IS (Figure 5). This result follows
Chegini [17] concept that this distance needs to be small enough to maintain a
degree of correlation and accuracy in the measurements of floodwater
propagation velocity.
Figure 5: Comparison of the computed front water velocity using various

methods for dry-bed downstream conditions.
Some previous investigators interchanged wave celerity with front water flow
velocity. This concept is verified in the present study. An approximation of wave
celerity was obtained from the shallow water relationship taken as:
(2)
where C = wave celerity, g = gravity acceleration and d = initial reservoir water

depth.
The solitary wave theory gives celerity for the steep waves as:
1 (3)
where H = water depth in the channel and d = water surface elevation from Still
Water Level (SWL).
However, considering the dry-bed downstream condition, eqn (3) is
simplified to the form of shallow water relationship equivalent to eqn (2). This
allows the use of eqn (3) to compute floodwater wave celerity for the present
experiments. Also, using the empirical and analytical equations for the flow
velocities proposed by various previous investigators, the velocities obtained
from their models are compared with the celerity of the present study. It can be
seen from Figure 6 that the flow celerity of the present study is in close
agreement with the front water velocity of Lauber and Hager [8] while other
investigators appeared to overestimate the front water velocity in relation to
celerity of the flow.
Similarly, Figure 7 compares the front water flow velocity of previous
investigators with the floodwater front velocity obtained in the present study. It
Figure 6: Comparison of floodwater front velocity of existing theories with the

flow celerity of the present study.
Figure 7: Comparison of the computed front water velocities for this study
with various existing theories.
can again be seen from the figure that the front water velocity computed using
SSC method with wave probes 0.41m apart and that of IS are in close agreement
with Liem and Kongeter’s theory [13] as well as with Ritter’s predictions [7].
Hunt’s theory has fair agreement with front water velocity computed using SSC
method with wave probes of 2.2m apart. It should be noted that a comparison
with Hunt’s theory may be incorrect at the upstream end of the channel since
Hunt’s equation is said to be valid only once the wave front has travelled a
distance of more than 4 times the reservoir length [18].
5 Conclusions
The use of digital imaging for qualitative and quantitative characterisation of
fluid flows is not new. In recent years however, with the rapid development of
powerful digital cameras at affordable prices and the advances in robust and fast
image processing techniques, this tool has become very popular.
In the present study, propagation velocities of floodwater flow have been
computed in an idealized dam-break problem using various adaptations of the
commonly used PIV method. The IS and the SSC methods described in section 3
have been adopted. The results obtained using the two techniques demonstrated
good agreement with the dam-break wave theory. However, it was observed that
the velocities obtained using the SSC method with wave probes at shorter
distance away from each other (0.41m apart) appeared closer in agreement to the
IS than that of the SSC with 2.2m separation. In addition, it is also indicated that
the propagation velocity obtained using the SSC method with wave probes
0.41m apart and that of the IS appeared in close agreement with some previous
researchers, particularly Chegini’s concept [17].
It was also revealed that most previous investigators over-estimated front
water velocity by interchanging it for the wave celerity which implies that
caution should be taken when doing this as it is only applicable in certain
circumstances. Furthermore, the results of comparison of front velocity with
various downstream water levels (DSWL) revealed that higher DSWL reduces
the speed of the bore, which indicates that the water in front of the travelling
bore reduces the speed of the flood wave. More importantly, this development
represents a useful laboratory scheme for analysing hydrodynamics model

studies and is well suited for initial investigations.
References
[1] Sayers, P., Hall, J., Dawson, R., Rosu, C., Chatterton, J. and Deakin, R.,
Risk Assessment of Flood and coastal Defences for strategic Planning
(RASP) – A high level Methodology. DEFRA Conference of Coastal and
River Engineers, Keele University, HR Wallingford, 2002.
[2] Kelman, I. and Spence, R., An overview of flood actions on buildings.
Eng. Geol., 73, pp. 297–309, 2004.
[3] Kreibich, H., Piroth, K., Seifert, H., Maiwald, H., Kunert, U., Schwartz, J.,
Merz, B. and Thieken, A. H., Is flow velocity a significant parameter in
flood damage modelling? Natural Hazards Earth System Science, 9, pp.
1679 –1692, 2009.
[4] Soetanto, R. and Proverbs, D. G., Impact of flood characteristics on
damage caused to UK domestic properties: the perceptions of building
surveyors. Structural Survey, 22 (2), pp. 95–104, 2004.
[5] USACE, Design of Revetments, Seawalls and Bulkheads. EM 1110 – 2 -
1614, 1996.
[6] Smith, D. I., Flood damage estimation – A review of urban stage damage
curves and loss functions. Water SA, 20 (3), pp. 231–238, 1994.
[7] Chanson, H., Applications of the Saint-Venant Equations and method of
Characteristics to the Dam Break Wave Problem. Hydraulic Model
Reports of Department of Civil Engineering, University of Queensland,
Report No. CH55/05, ISBN 1864997966, 2005.
[8] Lauber, G. and Hager, W. H., Experiments to dam-break waves:
Horizontal channel. Journal of Hydraulic Research, 36 (3), pp. 291–307,
1998.
[9] Stansby, P. K., Chegini, A. H. N. and Barnes, T. C. D., The initial stages
of dam-break flow. Journal of Fluid Mechanics, 374, pp. 407–424, 1998.
[10] Adrian, R., Engineering Application of Particle Image Velocimeters. Proc.
of ICALOE, Laser Institute of America, pp. 56–71, 1989.
[11] Douxchamps, D., Spinewine, B., Capart, H., Zech, Y. and Macq, B.,
Particle-Based Imaging Methods for the Characterisation of Complex
Fluid Flows. Proc. of the IEEE Oceans, pp. 20–25, 2004.
[12] Chegini, A. H. N., Pender, G., Slaouti, A. and Tait, S. J., Velocity
measurement in dam-break flow using imaging system. Proc. Of the 2nd
Int. Conf. On Fluvial Hydraulics, IAHR/AIRH, 2 (June), pp. 858–867,
2004.
[13] Liem, R. and Kongeter, J., Application of High-Speed Digital Image
Processing to Experiments on Dam Break Waves. Proc. of Concerted
Action on Dam-Break Modelling (CADAM) Conference, European
Community Workgroup, pp. 399–411, 1999.
[14] Van Dyke, M., An Album of Fluid Motion, Publication of Parabolic Press,
1982.
[15] Capart, H., Young, D. L. and Zech, Y., Voronoi imaging methods for the
measurements of regular flows. Experimental Fluids, 32, pp. 121–135,
2002.
[16] Raffel, M., Willert, C. E. and Kompenhaus, J., Particle Image Velocimetry
– A Practical Guide. Springer-Verlag: Berlin and New York, 1998.
[17] Chegini, A., Fundamental Investigations of Dam-break Flows. Ph.D.
Thesis, Department of Civil Engineering, University of Manchester,
Manchester, UK, 1997.
[18] Hunt, B., Dam-break solution. Journal of Hydraulic Engineering, ASCE,
110 (6), pp. 675– 686, 1984.
Numerical simulation of the inundation area

for landslide-induced debris flow: a case study
of the Sha-Xinkai gully in southern Taiwan
J.-C. Chen1, J.-S. Wang2, M.-R. Chuang1 & C.-J. Jeng1
1
Department of Environmental and Hazards-Resistant Design,
Huafan University, Taiwan
2
Ecological Soil and Water Conservation Research Centre,
National Cheng Kung University, Taiwan
Abstract
Typhoon Morakot struck central and southern Taiwan on August 8, 2009, and
the high rainfall intensity and accumulated rainfall-induced several floods,
landslides, and debris flows. In this study, the destructive debris flow caused by
Typhoon Morakot in the Sha-Xinkai gully of the Liouguei District in southern
Taiwan was selected as a case study for analysis. A two-dimensional model
(FLO-2D software) was used to simulate debris flow. First, hydrological and
geomorphological data were collected on the debris flow event and the
rheological properties of slurry collected from the field were analyzed. Next,
the relationship between debris flow discharge and water flow discharge was
obtained. The simulation results for the deposited area and depth were then
compared to aerial photos taken during a field investigation. Finally, the bulked
coefficient of discharge and the resistant parameters used in the model were
presented. The results showed that the maximum deposited depth in the debris
flow inundated area was over 6 m; the maximum velocity, 6.6 m/s; and the
deposited volume, almost 1,000,000 m3. The simulated deposition depth and
inundation area matched the results from the field investigation reasonably well.
In this study, the parameters and processes needed for the simulation of
landslide-induced debris flows were proposed to provide a reference for hazard
zone mapping and debris flow hazard mitigation.
Keywords: Typhoon Morakot, FLO-2D, rheological property, bulked coefficient.
doi:10.2495/FRIAR140041
1 Introduction
Typhoon Morakot struck central and southern Taiwan on August 8, 2009. The
extreme rainfall (maximum hourly rainfall of 123 mm and 48-h rainfall of
2,361 mm measured at the Alishan rainfall station) associated with the typhoon-
induced several landslides, debris flows, and floods (Chen et al. [1]; Wang et al.
[2]). In the basin of the Raolung River in southern Taiwan, many landslide-
induced debris flow hazards originate from a gully that has a small watershed
area ( A ) (e.g., A smaller than 40 ha) and a high landslide ratio ( RL ) (e.g., RL >
30% where RL is the ratio of landslide area AL to watershed area A or RL =
AL / A ). Gullies with small watershed areas are generally unknown or overlooked
by people, and they are often the cause of serious disasters during extreme
rainfall events. The number of extreme rainfall events in Taiwan has increasing
trend in recent years, which has resulted in a greater number and magnitude of
debris flows during the last decade (Chen et al. [3]). Hence, the development of
techniques that can identify and possibly prevent debris flows in gullies is a very
important research topic for hazard mitigation efforts.
The FLO-2D [4] routing model is software designed for two-dimensional
mathematical modeling of water movement and fast flowing slope processes
including debris flows. The FLO-2D model has been used successfully for debris
flow simulations by many researchers in a variety of countries (Lin et al. [5];
Tecca et al. [6]; Sosio et al. [7]; Stolz et al. [8]; Jakob and Weatherly [9]; Hsu et
al. [10]; Sodnik and Mikos [11]). Data required for model simulations include a
digital terrain model, an inflow hydrograph, rheological properties of the
sediment water mixture, and the Manning roughness coefficient. The results
from debris flow simulations are especially sensitive to the inflow hydrograph
and rheological parameters associated with volumetric sediment concentrations.
The inflow hydrograph may be underestimated for small watershed areas with
high landside ratios. However, previous research has generally focused on debris
flows from large watershed areas. Furthermore, the rheological parameters used
in previous studies are usually determined by the back analysis method or by
comparisons between model simulations and field observations. In contrast, the
rheological parameters used in this study were determined via laboratory
experiments. Two volumetric concentrations were used to simulate landslide-
induced debris flow, and the empirical coefficient (i.e., the discharge bulked
coefficient) that described the relationship between debris flow discharge and
water flow discharge was determined in this study. Results of this study can
provide a basic framework Results of this study can provide a basic framework
for determining debris flow discharges and select rheological parameters in
simulations of landslide-induced debris flows, which is important for hazard
2 Study area
The Sha-Xinkai gully study area is located in the Shinfa Village of the Liouguei
District, Kaoshing city, in southern Taiwan (Figure 1). It has a catchment area of
29.7 ha, a main stream length of 542 m, and an average stream bed slope
of 22.5°.
Liouguei
Kaohsiung Sha-Xinkai gully
Watershed area
N Deposition area on land
Sha-Xinkai watershed
Kaohsiung city
Shinfa rain station
Taiwan
Figure 1: Location of the Sha-Xinkai gully, the Sha-Xinkai watershed, and

the deposition of debris flow material during Typhoon Morakot in
2009.
2.1 Debris flow hazard and rainfall
2.1.1 Debris flow hazard

In 2009, Typhoon Morakot brought intense rainfall to southern Taiwan and
caused many landslides and debris flows in the Shifa village. The Sha-Xinkai
gully, a site of one of the landslide-induced debris flows in the village, was
selected as our study area. The event resulted from a landslide that occurred
upstream and entered the main stream of the gully where it mixed with water to
become a debris flow. The debris flow eroded the sidewalls of the stream, which
entrained additional material that traveled further downstream. In total, the Sha-
Xinkai debris flow produced approximately 1,000,000 m3 of deposited sediments
in downstream areas. The deposited depth was over 6 m in certain areas (SWCB
[12]). The debris flow traveled downstream into the Shifa village and Laolung
River where over 30 houses were buried. Tragically, the debris flow caused the
death of four individuals, and 24 people were reported missing. The maximum
deposition width on land approached 800 m. The landslide area in the Sha-
Xinkai watershed was 12.1 ha, and the landslide ratio was 40.7%.
2.1.2 Rainfall
The hourly and cumulative rainfall data collected from the Shinfa rain gauge
station, which is located approximately 2 km away from the Sha-Xinkai gully,
during Typhoon Morakot is shown in Figure 2. An hourly maximum rainfall

record of 103 mm was recorded at 6:00 PM on August 8, 2009. The 24-h rainfall
maximum of 1200 mm occurred over a period lasting from 3:00 AM on August
8, 2009, to 3:00 AM on August 9, 2009. Debris flows subsequently occurred
within the period of the 24-h rainfall maximum. The initial landslide and small
debris flows began around 7:00 PM on August 8, 2009 at the time that the hourly
rainfall reached its maximum amount. During 8:30 to 9:00 PM on August 8,
2009, the debris flow greatly expanded in size, flowed downstream, and buried
downstream areas in sediment.
200 2400
Hourly rainfall
2200
Cumulative rainfall
Debris flows 2000
160 (19:00-21:00)
1800
Cumulative rainfall, R (mm)

Hourly rainfall, I(mm/hr)
1600
120 1400
1200
Disaster caused by
80 large debris flow 1000
(20:30-21:00)
800
600
40
400
200
0 0
15 18 21 0 3 6 9 12 15 18 21 0 3 6 9 12 15 18 21 0 3 6 9 12 15 18 21 0 3 6 9 12 15 18 21 0 (hr)
Aug. 7 Aug. 8 Aug. 9 Aug. 10 (Month day)
Date
Figure 2: Rainfall data collected from August 7, 2009 to August 10, 2009 at
the Shinfa rain gauge station and the time that a debris flow was
triggered.
2.2 Rheological properties
Rheological properties are very important when modeling debris flows. In the
FLO-2D model, the rheological parameters, including the mixture yield stress
(  y ) and the mixture viscosity (  ), are used to describe the rheological
characteristics of debris flows. The rheological parameters are dependent on the
volumetric concentration ( cV ), and they have a significant effect on debris flow
processes and the final deposition morphology (FLO-2D [4]). To determine the
rheological parameters of debris flow, soil samples with a particle diameter of
less than 1 mm were collected from the flow area of the Hong-Shui-Xian gully,
which is located next to the Sha-Xinkai debris flow. The soil samples were
analyzed in a laboratory experiment using a Brookfield viscometer (type DV-III).
The relationship between the shear stress and shear strain for the soil sample at
various cV values was analyzed. The results showed that the rheological
properties of the debris flow slurries could be described by the Bingham model.
The Bingham model contains two rheological parameters: yield stress (  y ) and
viscosity (  ). The  y (in dynes/cm2 units) and  (in poise units) both
exponentially increased with an increase in volumetric concentration ( C ) V
(Figures 3 and 4), and these terms were defined as:

 y  0.459 e16.43c V
(1)
  0.0485 e14.94 c V
(2)
The results from Eqs. (1) and (2) were consistent with the bounds reported in
previous studies (FLO-2D [4]; Dai et al. [13]; Fei [14]). Because the lithological
characteristics and grain sizes of deposits in the Sha-Xinkai gully were almost
identical to the Hong-Shui-Xian gully, the rheological relationships from
Eqs. (1) and (2) were used to determine the rheological parameters for debris
flow simulations in this study.
3 Methods
3.1 FLO-2D model
A two-dimensional commercial model, FLO-2D, which is physically based and

takes into account the mass and momentum conservation of flows, was used to
analyze the inundation area for landslide-induced debris flow in the Sha-Xinkai
gully. The basic equations used in the FLO-2D model include the continuity
equation and the dynamic equation. The inflow sediment concentration and the
inflow hydrograph provide the input conditions for continuity equation routing,
and the total shear stress involved in the dynamic equation affects the flow
behavior. The parameters related to the total shear stress include the rheological
parameters of yield stress (  y ) and viscosity (  ), the resistant parameter of
laminar flow ( k ), and the Manning roughness coefficient ( n ). These parameters
affect the flow velocity, flow depth, and deposition area. The volumetric
sediment concentration, inflow hydrograph, and the parameters related to flow
resistance (  y ,  , k , and n ) should be determined prior to debris flow
simulations.
3.2 Simulation and analysis procedure
3.2.1 Preparation of topographic and rainfall data and selection

of parameters
Data required for the model simulation included a Digital Elevation Model
(DEM), inflow hydrograph, and various parameters related to flow resistance
such as  y ,  , k , and n . The parameters used in this work are described as
follows:
1. Topographic data: Topographic input data were obtained from a DEM of

the Sha-Xinkai watershed. The data had a resolution of 5 m  5 m.
2. Rainfall data: Rainfall data were collected from the Shinfa rain gauge
station. The maximum hourly rainfall data from this station were used to
determine peak water flow discharges in the Sha-Xinkai gully during
Typhoon Morakot.
3. Parameters for simulation: The relationships for rheological parameters
(Eqs. (1) and (2)) were used to simulate debris flow. In addition to the
rheological parameters, other important parameters included the Manning
roughness coefficient ( n ) and the resistance parameter for laminar flow ( k ).
The n value depends on the land surface, and it can be determined by
referencing the FLO-2D user’s manual [4]. In the Sha-Xinkai gully, the n
value ranged from 0.10 to 0.20. Hence, a n = 0.15 was adopted for use in
this study. The k value can range from 24 to 50,000. For modeling debris
flow, a calibrated k value of 2285 (FLO-2D [4]) was used to simulate the
Sha-Xinkai debris flow event.
3.2.2 Determination of debris flow discharge

In engineering planning, debris flow discharge ( Qdp ) is generally considered to
be directly related to direct runoff ( Qwp ) (Chen et al. [15]) so that Qdp is
proportional to Qwp and can be expressed as:
Qdp  cb Qwp (3)
where cb is the discharge bulked coefficient. The value for Qwp is generally
determined from the rational formula, Qwp  C I A / 360 , where C is the runoff
coefficient, I is the maximum hourly rainfall intensity (mm/h), and A is the
watershed area. In the Sha-Xinkai gully study area, C = 0.8 (SWCB [16]), I =
103 mm/h (i.e., the maximum hourly rainfall observed at the Shinfa rain gauge
station during Typhoon Morakot), and A = 29.7 ha. Hence, Qwp was 6.8 m3/s
according to the rational formula described above. The discharge bulked
coefficient ( cb ) depends on conditions of sediment supplementation. The cb
value can be high when a watershed has a high landslide ratio or when there is
high sediment supplementation. The debris flow discharge in this study was
determined by Eq. (3), and the cb value was calibrated by comparing the results
obtained from numerical simulations to those obtained in the field investigation.
3.2.3 Construction of the inflow hydrograph for debris flow

According to media reports and visits by residents, landslides and small debris
flows began to occur around 7:00 PM on August 8, 2009. This escalated into a
large and rapid debris flow event at approximately 8:30 to 9:00 PM that had
disastrous consequences. Thus, the inflow hydrograph had a duration of
approximately 2 h (7:00–9:00 PM). The duration of the inflow hydrograph was
divided into two stages for this study. Stage one (from 7:00 to 8:30 PM) was the
stage in which landslides gradually transferred to small debris flows and stage
two (from 8:30 to 9:00 PM) was the stage of the large debris flow formation. The
ranges of cV used for the two stages were obtained from reference values in
the FLO-2D user’s manual [4]. Stage one had a cV = 0.55–0.65 for landslides
and stage two had a cV = 0.48–0.55 for debris flows. The inflow hydrograph
used in this study is shown in Figure 3.
Qdp= cb Qwp
Discharge (m3/s)
1 2
1: Landslide: failure with deformation

Cv =0.55-0.65
2: Debis flows
Cv =0.48-0.55
Qwp
PM
4 5 6 7 8 9 10 11 12
Time (h)
Figure 3: The inflow hydrograph used for this study. The hydrograph was
divided into stages 1 and 2 for simulations of debris flow.
3.2.4 Debris flow simulations and parameter calibration

Since debris flow often impact downstream areas where the debris is ultimately
deposited, modeling the depositional area of the debris flow was the primary
interest of this paper. The procedures used for determining the depositional area
of the debris flow and the calibration parameters ( cb and cV ) are described as
follows:
1. Determine the location of the debris flow fan apex such as the mouth of the
valley or the area downstream of the topographic apex. The location of the fan
apex for the debris flow gully was obtained from topographical map and field
investigations.
2. Assume a c b value and a set of cV values for determining the inflow
hydrograph as indicated in Figure 3. Input the inflow hydrograph at the debris
flow fan apex and the various parameters related to flow resistance such as
 y (Eq. (1)),  (Eq. (2)), k (= 2285), and n (= 0.15). The inundation area of
the Sha-Xinkai debris flow was then computed through FLO-2D simulations.
The results of FLO-2D simulations were compared to field conditions in terms
of deposition depth and depositional area. If the simulated results were not in
agreement with field conditions, the inflow conditions (i.e., cb and cV ) were
adjusted until the simulated results were similar to the conditions observed in
the field investigation.
4 Results
4.1 Discharge bulked coefficient
If the water discharge contained in the debris flow discharge is solely contributed
by direct runoff ( Qwp ) (i.e., water flow discharge), the debris flow discharge
( Qdp ) is directly related to Qwp , and it is equivalent to the sum of Qwp and the
sediment discharge ( Qs ) (where Qs = cV Qdp ). The discharge bulked coefficient
( cb ) in Eq. (3) can be expressed as:
cb  (1  cV ) 1 (4)
Similar to Eq. (4), Takahashi [17] derived cb  (1  k c *cV ) 1 for debris flows
generated from gully bed erosion where k c *  c* 1 and c* is the volumetric
concentration of the sediment layer on the gully bed. The maximum cV values
observed ranged up to 0 .9 c * (Takahashi [17]). Based on Takahashi’s research,
the maximum cb = 10 if cV = 0.9c* . This implies that the maximum debris flow
discharge is 10 times that of the water flow discharge. However, in the
relationships for cb  (1  cV ) 1 or cb  (1  kc *cV ) 1 , ground water or the water
contained in the sediment layer was not considered. Hence, the cb value
calculated by cb  (1 cV )1 or cb  (1  kc*cV ) 1 may underestimate the discharge for
debris flows induced by large landslides. The peak water flow discharge ( Qwp )
determined by the rational formula in this case study was 6.8 m3/s. Here, the cV
value was calculated by the relationship of the equilibrium concentration
(Takahashi [17]):
tan 
cV  (5)
(Gs  1)(tan   tan  )
where Gs is the specific gravity;  , the friction angle; and  , the angle of the
gully bed in flow section. Using a Gs = 2.65,  = 35 o , and  = 17 o , the cV
value determined from Eq. (5) was cV = 0.47. Also, the cb = 1.89 according to
Eq. (4). These data imply that the debris flow discharge was 1.89 times that of
the water flow discharge ( Q dp  1.89 Q wp ). The inundation area was modeled using
an inflow hydrograph of debris flow discharge of Qdp = 16.2 m3/s ( 1.89 Qwp ), a
duration of 2 h, and the rheological parameters (i.e.,  y and b ) computed with
Eqs. (1) and (2). Additionally, a cV = 0.47, n = 0.15, and k = 2285 were used
as inputs. Figure 4 shows the inundation area of debris flow from the FLO-2D
simulations. The inundation area and deposition depth from the simulations were
smaller than those determined from the field investigation due to an
underestimation of debris flow discharge. Besides direct runoff, the water flow
that initiated the debris flow likely came from ground water or water contained
in sediments that was brought in by the landslides. Furthermore, water flow
could have been blocked by the sediment brought in by landslides, which would
have rapidly increased water storage in the watershed. When the stored water
combined with sediments burst over a short period of time, this could have led to
a high debris flow discharge. The cb value calculated with Eq. (3) ranged from
15 to 20 when the inflow hydrograph followed the type shown in Figure 3 and
the debris flow volume was estimated at 1,000,000 m3. The cb value in the study
area was calibrated by comparisons of the numerical simulations to field
investigation data.
Sha-Xinkai watershed area

Deposition area on land from investigation
Deposition area from simulation
(without considering water bulked effect)
Figure 4: Comparison of the simulated versus actual debris flow inundated

area. The simulated debris flow was determined using an inflow
hydrograph with a discharge bulked coefficient cb = 1.89. The cb
attributed to landslides was not considered here.
4.2 Parameter calibration
When n = 0.15, cb = 18, and the cV values for stages one and two were 0.64
and 0.50, respectively, in the inflow hydrograph (Figure 3), the depositional area
and deposition depth from the simulations were close to those observed during
the field investigation (Figure 5). The deposited depth in the debris flow
Sha-Xinkai watershed area

Deposition area on land from investigation
Deposition area from simulation
(considering water bulked effect)
Figure 5: The simulated depositional area after adjustments were made

based on field data.
inundated area was over 6 m. The debris flow discharge was 18 times that of the
peak water flow discharge due to the high landslide ratio (40.7%) in the Sha-
Xinkai gully. The simulated results also showed that the debris flow rapidly
inundated the downstream area at 8:30–9:00 PM on August 8, 2009, with a
maximum velocity of 6.6 m/s.
5 Conclusions
The study area in the Sha-Xinkai debris flow gully had a small watershed area
( A ) and a high landslide ratio ( RL ) ( A  29.7 ha and RL = 40.7%). For this type
of the debris flow gully, the peak water discharge computed from the rational
formula was small and it likely underestimated the inflow hydrograph in debris
flow simulations. The depositional area and deposition depth in the simulations
were strongly affected by the inflow hydrograph that was associated with the
discharge bulked coefficient ( cb ) and the volumetric concentration ( cV ). A
method to reasonably determine cb and cV is important for debris flow
simulation research. In this study, the relationship between debris flow discharge
( Q dp ), peak water flow discharge ( Q wp ) (or cb value), and cV values in the
inflow hydrograph were calibrated by comparing the results obtained from the
numerical simulations to data from a field investigation. The debris flow
discharge in the Sha-Xinkai gully had a Q dp = 18 Q wp or cb = 18. Two cV
values for two different stages of the hydrograph were used to evaluate
rheological parameters (yield stress  y and viscosity  ), and these values were
cV = 0.64 and 0.50 for stages one and two, respectively. Calculation results also
indicated that the simulated sediment volume was approximately 1,000,000 m3,
the maximum flow velocity was about 6.6 m/s, and the maximum depth on the
flow was over 6 m. The simulated average depth was close to the depth observed
in the field investigation. These data may useful as a reference for future hazard
References
[1] Chen, T. C., Wu, C. C., Weng, M. C., Hsieh, K. H. & Wang, C.C., Slope
Failure of Lawnon Basin Induced by Typhoon Morakot, Sino-Geotechnics,
122, pp. 13–20, 2009 (in Chinese).
[2] Wang, C.M., Lee, S.P., Li, C.C., Tsang, Y.C. & Shieh, C.L., Disasters
caused by Typhoon Morakot, Journal of the Taiwan Disaster Prevention
Society, 2(1), pp. 27–34, 2010 (in Chinese).
[3] Chen, J.C., Huang, W. S., Jan, C.D. & Yang, Y.H., Recent Changes in the
Number of Rainfall Events Related to Debris-Flow Occurrence in the
Chenyulan Stream Watershed, Taiwan, Nat. Hazards Earth Syst. Sci., 12,
pp. 1539–1549, 2012.
[4] FLO-2D, FLO-2D Users Manual, Ver. 2009. FLO-2D Software Inc,
Nutrioso, AZ, USA, 2009.
[5] Lin, M.L.,Wang, K.L. & Huang, J.J., Debris flow run off simulation and
verification – case study of Chen-You-Lan Watershed, Taiwan., Nat.
Hazards Earth Syst. Sci., 5, pp. 439–445, 2005.
[6] Tecca, P.R., Genevois, R., Deganutti, A.M., & Armento, M.C., Numerical
modelling of two debris flows in the Dolomites (Northeastern Italian
Alps)., Debris-Flow Hazards Mitigation: Mechanics, Prediction, and
Assessment, Chen & Major, eds, Millpress, Netherlands, pp. 179–188,
2007.
[7] Sosio, R., Crosta, G.B. & Frattini, P., Field observations, rheological
testing and numerical modelling of a debris-flow event. Earth Surf.
Process. Landforms, 32, pp. 290–306, 2007.
[8] Stolz, A. & Huggel, C., Debris flows in the Swiss National Park: the
influence of different flow models and varying DEM grid size on
modeling results. Landslides, 5, pp. 311–319, 2008.
[9] Jakob, M. & Weatherly, H., Integrating uncertainty: Canyon Creek
hyperconcentrated flows of November 1989 and 1990., Landslides, 5,
pp. 83–95, 2008.
[10] Hsu, S. M., Chiou, L. B., Lin, G. F., Chao, C. H., Wen, H. Y. & Ku, C. Y.,
Applications of simulation technique on debris-flow hazard zone
delineation: a case study in Hualien County. Taiwan, Nat. Hazards Earth
Syst. Sci., 10, pp. 535–545, 2010.
[11] Sodnik, J. & Mikos, M., Estimation of magnitudes of debris flows in
selected torrential watersheds in Slovenia, Acta geographica Slovenica,
46(1), pp. 93–123, 2006.
[12] SWCB, Disasters caused by Typhoon Morakot in Taiwan, 1999. Soil and
Water Conservation Bureau (SWCB), Taiwan, 2009 (in Chinese)
[13] Dai, J., et al., An experimental study of slurry transport in pipes. Proc.,
Int. Symposium on River Sedimentation, pp. 195–204, 1980.
[14] Fei, X. J., Bingham yield stress of sediment water mixtures with
hyperconcentration, J. Sediment Res., 3, Beijing, China, pp. 19–28, 1981.
[15] Chen, J.C., Jan, C.D., and Lee, M.S., Reliability Analysis of Design
Discharge for Mountainous Gully Flow, Journal of Hydraulic Research,
46(6), pp. 835–838, 2008.
[16] SWCB, Technical Handbook of Soil and Water Conservation, Soil and
Water Conservation Bureau (SWCB), Taiwan, 2005. (in Chinese)
[17] Takahashi, T., Debris Flow. IAHR Monograph. Balkema, Rotterdam,
1999.
Section 2
Risk assessment
A practical approach to floodplain mapping

for large-scale catastrophe models
I. Carnacina & A. Jemberie
Research and Modeling, AIR Worldwide, USA
Abstract
Catastrophe models often cover large geographic areas spanning multiple
countries or, in the case of flood models, entire watersheds. Models must be
sufficiently detailed to accurately account for hydrologic variation, which is
notably challenging when the modeled region is large. This is particularly true for
flood models, which require a highly detailed dataset, usually derived from a
digital terrain model (DTM), for reliable floodplain mapping. For one-dimensional
(1D) hydraulic models, the floodplain mapping approach tends to yield flat
surfaces often resulting in artefacts and inconsistencies near river confluences.
Because flood extent is limited by the length of cross-sectional lines along the
floodplain, these flat surfaces tend to drop sharply when the simulation reaches a
flat delta. The use of a two-dimensional (2D) model avoids these problems, but at
a high computational cost, and requires high quality terrain and bathymetry data.
This paper presents a new methodology for mapping floodplains using water
elevation points along a river network obtained from a 1D hydraulic model and a
DTM. The methodology applies kinematic and diffusion wave equations in
a simplified manner, using water elevation points as internal boundary conditions.
Several parameters control the expansion and smoothing algorithms that generate
realistic flood extent maps for different return periods. This methodology is
particularly suitable for modeling large domains since it produces accurate results
but requires much less computational time than a 2D model. In addition, because
the computation uses several source points per cross section, the flood extent is
not limited by the cross-sectional length, making this methodology appropriate for
levee breaches and in cases where river banks are not well defined and the cross-
sectional geometry is derived from a DTM.
Keywords: risk assessment, flood, flood mapping, large scale model.
doi:10.2495/FRIAR140051
1 Introduction
Flood events in combination with human activity and land use changes threaten
both life and property in much of the world. The human and economic losses
inflicted by flood events have forced communities and governments to adopt new
direct and indirect measures to prevent, assess, and reduce the risk of flooding.
Over the last thirty years, a plethora of 1D, 2D and coupled flood models have
been developed, most of which are commercially available. Risk maps often
require the evaluation of risk for a given non-exceedance probability (often
referred to as return period maps), and thus, steady state models are often
preferred, again for their computational efficiency [1]. Flood hazard maps are
typically produced by governmental agencies, such as FEMA in United States,
ZÜRS in Germany, and many others across Europe. In this context, large scale
catastrophe risk models are catching the attention of more and more researchers in
academia and in industry. These models are used to assess the effect of
catastrophic flood on larger areas as opposed to local and detailed studies.
Despite the large number of such models, the need for nationwide medium-to-
high resolution inundation maps has led to the development of fast numerical
solutions with reduced computational effort.
In this study, we provide a tool for mapping hydraulic model results from a 1D
steady state model using a simplified quasi-physical approach that alternates
between the use of kinematic waves and diffusion waves to interpolate water
elevations between cross sections. This methodology eliminates the presence of
artefacts and drops in flood maps at confluence and reduces the computational
effort required in comparison to 2D models. The same kinematic and diffusion
waves approach is used to predict the flow level through a breach, using, as
boundary conditions, the solution obtained from 1D numerical model and a
volume hydrograph. Finally, the mapping algorithm provides an envelope of the
maximum depth for a given return period.
2 Methodology
One dimensional steady state models can rapidly assess the intensity and the extent
of flooding at specific return periods, provided that the effect of floodplain storage
is negligible [2, 3] and provided that the accuracy of such models is checked at
stream junctions, branches, and lateral inflow [1, 4]. One dimensional model
results, however, are only available locally at predefined model cross sections.
Therefore, 1D model solutions need to be interpolated between cross sections to
fill the gaps between them and can also be used in conjunction with 2D models to
simulate levee overtopping or flow through levee breaches.
Examples of interpolation of 1D hydraulic model results to create a flood extent
map include triangular irregular networks (TIN) generated by HEC-GeoRAS, as
well as chained interpolation between cross sections in MIKE-11.
The industry often requires large scale models (often national or continental) at
medium-to-high resolutions (specifically, 30 m to 90 m). DTMs and land use
datasets provide an essential source of input to generate such large scale models.
Therefore, model cross sections can be automatically extracted from these

datasets. However, given the large spatial coverage of catastrophe models,
automatic cross section generation may lead to issues such as:
1) Over large areas, not every cross section can be checked to assess whether
it covers the whole floodplain, especially at high return periods (typically,
greater than 100 years), thus TIN mapping may create artefacts and
sudden drops of water elevation in flat flood plains;
2) For large scale systems, automatically generated cross sections may not
cover the entire floodplain and may be too short near stream junctions.
Moreover, the presence of flood defences drastically modifies the dynamics of
flow between the main channel and the floodplain. While flood extent maps can
be correctly modeled for relatively small flood defences protecting a small portion
of urban areas, complex systems of levees such as those protecting the Mississippi
or Sacramento rivers create totally disconnected networks, for which the effect of
storage is no longer negligible. While 1D models work efficiently inside the leveed
areas, water flowing through a breach or overtopping this system of protection
needs to be treated separately from the rest of the network.
This methodology uses a quasi-physical raster-based approach to create a water
surface between crosssections. The algorithm uses a wave propagation concept
that avoids artefacts at the end of cross-section cut lines and river junctions that
are yielded by TIN-based interpolation. This methodology is also used to assess
the effect of levee breach or overtopping based on the solution obtained from the
1D model.
2.1 General framework
Free surface wave propagation along rivers is generally approximated using the de
Saint-Venant equation [7, 8]:
/
0 (1)

where Q is the discharge trough of a given cross-section or computational node, A

is the cross-sectional area, g is the gravitational acceleration, y is the depth of flow,
S0 is the bottom slope, and Sf is the friction slope.
Depending on the terms included to approximate the momentum balance (1),
the wave propagation assumes different names: kinematic wave, diffusion wave,
and fully dynamic propagation. In the present work (steady state conditions), only
the kinematic and diffusion waves are considered.
To interpolate the solution of the 1D model (water elevation ) from the cross
section cut lines to the rest of the domain (DTM), a set of points along the
cross section is chosen to serve as source points (subscript S).
The algorithm is then divided in two phases: 1) a first expansion phase, in
which the solution of the 1D source cell is propagated to the rest of the domain,
and 2) a second smoothing phase, in which the water elevations from the
expansion phase are smoothed to reduce the number of artefacts from the elevation
maps (for example, an unrealistic flood elevation between two adjacent cells, i.e.,
wall of water).
For the first iteration, each source point will propagate to the next cell using
either a diffusion or kinematic wave to compute the elevation at the empty
neighbouring cells (subscript C). Figure 1 shows the neighbouring convention
used by this algorithm, in which z is the DTM elevation from the reference datum.
Each source point will loop through the 8 neighbouring cells.
zc , ζ c zc , ζ c zc, ζc zc , ζ c zc, ζc
zc , ζ c zc , ζ c zc, ζc zc, ζc zs, ζs zs, ζs zs, ζs zc, ζc
zc, ζc zs, ζs zc, ζc zc , ζ c zs, ζs zs, ζs zs, ζs zc , ζ c
zc , ζ c zc , ζ c zc, ζc zc, ζc zs, ζs zs, ζs zs, ζs zc , ζ c
zc, ζc zc, ζc
zc , ζ c zc, ζc zc , ζ c
(a) (b)
Figure 1: Neighbouring scheme and expansion steps for: a) 1st iteration and
b) second iteration.
For each successive step, every neighbour cell at the previous step will become
a source point and the original source point will be removed from the set of source
points (Figure 1(b)). The iteration will continue until the ground elevation will not
allow any further expansion from source cell; that is, when the list of source cells
is empty, or when the maximum number of iterations is reached. After each
expansion step, flooded cells’ elevations are added in a smoothing array, while the
source cell evaluated directly from the cross section will not be added, in order to
prevent any alteration of the 1D solution along the cross section.
Thus, for each expansion step, the water elevation of each cell will be
calculated according to eqn. (2):
∆ (2)
where, ∆E is the head drop per expansion step.

Because the kinematic wave propagates its peak without dissipation, and
rearranging the kinematic terms of eqn. (1) and combining it with eqn. (2), the
maximum water depth propagation computed from cell to cell in steady state
conditions will assume the form:
∙ (3)
Note that the source water depth propagates without dispersion in eqn. (3).
In contrast, in the case of a diffusion wave, the wave propagates to its maximum
depth while reducing its peak during wave propagation. The dissipation ∆ can be
calculated by rearranging the diffusion wave in eqn. (1) and evaluating the friction
slope through Manning equation:
∙ ∙
∆ ∆ ∙ ∙ ,∆ (4)
Herein, ∆L=∆X·θ is the distance from the cell centres, ∆X is the cell size, and θ is
a direction factor equal to θ=1 for horizontal and vertical neighbours (θ=1.41 for
diagonal cells), S0= |ζs-ζc| is the DTM slope, nL(T) is the longitudinal (transverse)
Manning coefficient, VL(T) is the longitudinal (transverse) flow velocity, and ψL(T)
is the downhill scale factor. Manning’s n coefficient and velocity are set constant
and do not depend upon the orography of the area. Therefore, ψ and S0 are used to
scale the flow velocity from flat areas to steep areas, and obtain different energy
dissipation values for catchments with different slopes.
The maximum drop ∆Emax is enforced to reduce the energy dissipation in
presence of artefact that may present high slope values.
The selection between kinematic and diffusion wave is based upon the
definition of longitudinal and transversal wave propagation. In case of longitudinal
wave propagation, the diffusion wave assumes velocity and Manning coefficients
that approximate the wave propagation along the river centre line. Conversely, the
transverse wave propagation assumes lower Manning’s n, velocity and thus energy
dissipation, to simulate propagation normal to the river centre line.
For transversal wave propagation, in fact, the free surface slope cannot be
approximated using the ground slope and, thus, the kinematic assumption would
lead to high error in predicting the water surface [7, 9] and cannot be used to
evaluate the water elevation.
In this study, the distinction between longitudinal and transverse wave
propagation is achieved through the location of the expansion cell C originating
from a source point lying on a cross section.
In order to differentiate between the two different wave propagation, it is
necessary to first introduce the Euclidean polygons. These polygons are defined
as the sets of point with a minimum Euclidean distance from the source cross
section. The DTM space is, therefore divided into different Euclidean polygons,
corresponding to each cross section.
Wave propagation of cells within the Euclidean polygon will be assumed to
have transverse wave propagation, while cells outside the Euclidean polygon will
be assumed to have longitudinal wave propagation, and thus, will have higher
energy dissipation to smoothly fill the gap between cross sections.
When the expansion cell lies inside the Euclidean polygon, the algorithm
selects the water elevation to be the maximum of the two elevations computed by
the diffusion or by the kinematic wave. During the longitudinal wave propagation
(thus, outside the Euclidean polygon), the water elevation is evaluated as the
minimum yielded by the kinematic and diffusion equations. Figure 2 shows
the results of the expansion algorithm from the 1st iteration to the final map extent
(Rhone River near Martigny). Figure 2(a) shows the first step of the expansion
algorithm. Here, the green line represents the cross section automatically
generated from the catchment and flow line characteristics (green lines), while the
green dots represent the source point location.
(a) (b)
(c)
Figure 2: Example of modeled flood expansion (Rhone River near Martigny,

CH) at three different iterations: (a) 1st iteration, (b) 3rd iteration,
(c) final extent.
Only the source points with an elevation above the DTM elevation can expand
from the 1st iteration, as shown by the flooded cells (blue) under the source point.
Herein, not all the points have been activated to show the potential of the algorithm
to flood area outside the cross section limit. As the algorithm executes, gaps
between cross sections are filled in both directions (Figure 2(a)). Finally, as the
algorithm proceeds, area outside the cross section extents will be flooded, in both
longitudinal and transversal direction. As shown in the figure (see the lower right
quadrant of Figure 2(c)), the flood extent map does not stop at the end of the cross
section; rather, the algorithm is capable of expanding the modeled flood extent
into those areas that would be left dry using traditional approaches.
2.2 Levee breach and overtopping
The general framework presented here, i.e., kinematic and diffusion wave
propagation, can be used to evaluate the water depth from levee overtopping or
breach to provide a solution that adapts to the morphology of a floodplain. The
simplified 2D mapping algorithm has parameters selected to maximize the
correspondence between the 1D solution and the 2D solution. This is done to
prevent “wall of water” and map discontinuities at junctions observed using TIN-
based flood maps in which cross sections are not long enough to cover the extent
of the floodplain. However, in cases of levee breach or overtop, the longitudinal
dissipation may be either too low or too high; therefore, models of levee failure or
overtopping need to be driven by different physical assumptions, volume carried
by the hydrograph and wave duration.
Several reduced complexity models, or simplified models, have been
developed to assess the risk of levee failure, overtopping, or inundation of urban
areas. Typical approaches include models based on the discretisation of the
diffusive wave equation on Cartesian grids [16], regular and irregular storage cell
models [3, 5, 10–14] and raster based inertial models [15].
These reduced complexity methodologies are based on different assumptions,
but they all aim to decrease the computational cost of assessing levee failure
compared to fully physical 2d shallow water solutions. However, these reduced
complexity methodologies also analyze the evolution of the flood extent within a
certain event, and, thus may need data post processing, which requires additional
analysis and computational overhead, especially over a large domain.
In this paper, we use these types of risk maps to assess the maximum flood
elevation and extents associated with a certain return period in case of levee breach
and overtopping. Velocity within the floodplain will be neglected, due to the
relatively lower velocity in the floodplain compared to the main channel.
Starting from the source points at the end of a cross section, the average depth
can be evaluated using the information from the volume of water available within
a certain time or from a given hydrograph.
To account for the added water volume that may come from a breach, the
expansion parameters must be dynamically changed to adapt the solution to the
average available volume associated with a certain return period.
For each step, the available volume is redistributed in the expanded area such
that the average water depth for the next expansion multiplied by the total
expanded area matches the available volume.
Thus, the expansion volume at the step n, , is calculated as:
̅ (5)
where:
∆
∆
(6)
in which ̅ = average depth at iteration n, ∆ = difference of volume expanded

between step n and n-1, ∆ = difference of extents between step n and n-1,
is total the area flooded at the step n.
The downhill scale parameter ψ is then adjusted according to the normalized
volume error:
/ (7)
where, similarly to the method adopted by Liu and Pender (2010) [5], is the
total maximum volume available from the breach hydrograph at the expansion
step n from the beginning of the simulation, either obtained from a historical event
or from a simulated flood event. In this case, if the error has a positive bias, ψ
increases by a certain amount, in order to increase the dissipation for the next step
and adjust the volume. Since each expansion step depends on the model resolution
and not on a time step, the total volume available at each expansion step needs to
be evaluated from the hydrograph associating a certain time step to each expansion
step. In first approximation, this time step can be associated with typical flood
plain velocity, although more complex formulation can assume a dependency on
the average velocity evaluated from the flow surface gradient. Herein, the
floodplain velocity uexp will be assumed constant and needs to be calibrated to
produce reasonable extents, as discussed later in the validation.
To be more explicit, for a triangular flood hydrograph of which the volume
overtopping or flowing through the flood defence is known, the volume at
the expansion step n will be calculated as:
for n (8)
1 for n (9)

where nTP=TP / (Δx·uexp), nTP is the number of expansion iterations at the
hydrograph peak TP, nTE=TE / (Δx·uexp) is the number of expansion iterations at
the end of the hydrograph, uexp= is the average flood expansion velocity, and Δx is
the cell grid size.
3 Validation and discussion

Modeled flood extents are validated by comparing the Ohio River Federal
Emergency Management Agency (FEMA) 100-year flood maps to the modeled
flood extent for the city of Cincinnati (Ohio), and by comparing gage station
observations and modeled flood extents for the New Madrid 2011 breach [6].
The error between modeled flood elevation and observed flood gage elevation
has been measured in terms of root mean square error (RMSE) between predicted
and measured maximum elevation. The fit between two maps can also be
expressed in terms of fraction of the inundation domain correctly assessed by the

model [17]:
∑ ,
∑ , ∑ , ∑ ,
(10)
where the P is the generic cell in both model (M) and pilot domain (D) either
flooded (subscript 1) or dry (subscript 0). Modeled flood extents have been created
using the U.S. national elevation dataset (NED) with a resolution of 30 m.
Figure 3 compares the FEMA flood extent maps and the modeled flood extent
(AE zones in light blue and X zones protected by levee in yellow) for the Greater
Cincinnati region; the modeled flood extents were determined using the procedure
described in this paper, in which water elevations are propagated from cross
sections.
(a) (b)
Figure 3: 100-year flood extent validation (Cincinnati, OH): (a) modeled

flood extent, and (b) FEMA 100-year AE zone (light blue) and X
protected zone (yellow). NOTE: red and pink colours represent the
urban area extracted from the NLCD 2006 land use land cover data.
The agreement between the FEMA extent map and the modeled map in
Cincinnati urban area considering both AE and X protected zones is F=0.76. The
levee breach model has been validated by comparing the 2011 breach of Cairo,
Illinois and the USACE simulation [6]. The Birds Point, Illinois, breach reached
its maximum volume after 3 days of operation with a maximum volume of around
1.3·109 m3 breach. For reference, USACE provides a model of the 2011 event
Cairo breach showing the floodway evolution after 120 hours (5 days) [6], which
shows an average propagation velocity of 0.2 m/s. The USGS provides a large
dataset of field measurements for this event (around 20 gage stations inside the
floodway, shown by the green dots in Figure 4). The maximum value of each gage
has been selected and used to calibrate the model. A range of different uexp in the
range 0.01 m/s< uexp <0.5 m/s has been selected. According to these data,
the minimum RMSE has been obtained using for uexp=0.2 m/s (RMSE=1.21 m;
F=0.7). Figure 4 compares the observed flood extent obtained from Landsat image
to the modeled flood extent. Water extent from Landsat images have been
extracted using the modified normalized difference water index (MNDWI [18,
19]), using a value of MNDWI>0.3 to detect water features on the image.
Two types of boundary conditions mainly affect the raster model, i.e.,
topography and Manning’s n coefficient correlated with the land use. However,
the results of both the general flood extent model and the levee failure model
presented in this study fit well with observed flood data. Further, the performance
of both models is comparable to that observed by Bates and De Roo [16] at similar
DTM resolutions.
In addition, we noted that a simplified flood model that employs a DTM with
coarser resolution produces modeled flood extent maps that fit less well to
observed flood extents. For example, Bates and De Roo [16], observed that
choosing a 25 meter resolution land use raster rather than 100 m resolution raster
inflicts a 10% loss in fit quality between the modeled and observed flood
extents.” In both Ohio River and New Madrid floodplain, fine resolution DTM
topography accounts for additional and localized levee protection, which cannot
be correctly assessed using the original resolution and need to be manually
surveyed and added on the DTM.
(a) (b)
Figure 4: Comparison between (a) simulated levee failure with uexp=0.2 m/s
(green dots: USGS gage station; pink line: USACE accredited
levees, red line: Mississippi river centre line) and (b) Landsat
extracted water bodies (MNDWI>0.3) for New Madrid floodway.
Other biases may derive from the use of a global land use in both modelling
approaches that may account for localized differences between predicted extents
and satellite images. Moreover, the use of a single uexp used for the levee breach
approach to estimate the amount of volume available for each expansion step may
lead to depth underestimates during the first development phase, where flow
expansion is generally faster due to a steeper free surface. In contrast, depth
overestimates can occur during the last ponding expansion phase of the breach,
where velocity may reach lowest expansion value. Finally, the model does not
account for ponding and backwater effects, which further reduce the
computational performance of this methodology.
4 Conclusion
A new framework to map 1D model results both along river flood plain and due
to overtopping or levee failure has been presented. The model is based on simple
development of kinematic and dynamic wave propagation. The modeled results
are comparable to those produced by other simplified models, while maintaining
a reduced computational cost. Boundary conditions, such as topography and land
use, account for the majority of discrepancy between the FEMA flood map (where
floodplains are manually surveyed) or historical inundation maps. Finally, further
analysis of the sensitivity of the model on the parameters used to reproduce flood
extent maps in both the general and levee framework should be conducted.
References
[1] Mapping the Zone: Improving Flood Map Accuracy; National Research
Council; Washington, DC: The National Academies Press, 2009.
[2] Horrit, M.S. & Bates, P.D., Evaluation of 1D and 2D numerical models for
predicting river flood inundation, Journal of Hydrology, 268, pp. 87–99,
2002.
[3] Krupka, M., Pender, G., Wallis, S., Sayers, P.B. & Mulet-Marti, J., A rapid
flood inundation model, In Proceedings of the 32nd Congress of the
International Association For Hydraulic Research, pp 1–28, 2007.
[4] Büchele, B., Kreibich, H., Kron, A., Thieken, A., Ihringer, J., Oberle, P.,
Merz, B. & Nestmann, F. Flood-risk mapping: contributions towards an
enhanced assessment of extreme events and associated risks, Nat. Hazards
Earth Syst. Sci., 6, pp. 485–503, 2006.
[5] Liu, Y. & Pender, G., A new rapid flood inundation model, In Proceedings
of the First IAHR European Congress, ed. S. Arthur, Edinburgh, UK, 2010.
[6] DeHaan, H., Stamper, J. & Walters, W., Mississippi River and Tributaries
System 2011 Post-Flood Report, USACE, Mississippi Valley Division,
2012.
[7] Ponce, V.M., Li, R.M. & Simons, D.B., Applicability of kinematic and
diffusion-models, Journal of the Hydraulics Division — ASCE, 104 (3),
pp. 353–360, 1978.
[8] Mujumdar, P.P., Flood wave propagation – The Saint Venant Equation,
Resonance, 6 (5), pp. 66–73, 2001.
[9] Singh, V.P. & Aravamuthan, V., Errors of kinematic-wave and diffusion-
wave approximations for steady-state overland flows, Catena, 27 (3–4),
pp. 209–227, 1996.
[10] Cunge, J.A.., Two-dimensional modeling of flood plains. Water Resources
Publications, 17, 705–762, 1975.
[11] Moussa, R. & Bocquillon, C. On the use of the diffusive wave for modelling
extreme flood events with overbank flow in the floodplain, Journal of
Hydrology, 374, pp. 116–135, 2009.
[12] Castellarin, A., Domeneghetti, A. & Brath, A., Identifying robust large-scale
flood risk mitigation strategies: A quasi-2D hydraulic model as a tool for the
Po River, Physics and Chemistry of the Earth, 36, pp. 299–308, 2011.
[13] Gouldby, B., Sayers, P., Mulet-Marti, J., Hassan, M. & Benwell, D.,
A methodology for regional-scale flood risk assessment. Proceedings of the
Institution of Civil Engineers - Water Management, 161, pp. 169–182, 2008.
[14] Falter, D., Vorogushyn, S., Lhomme, J., Apel, H., Gouldby, B. & Merz, B.,
Hydraulic model evaluation for large-scale flood risk assessments, Hydrol.
Process., 27, pp. 1331–1340, 2013.
[15] Bates, P.D., Horritt, M.S. & Fewtrell, T.J., A simple inertial formulation of
the shallow water equations for efficient two-dimensional flood inundation
modelling, Journal of Hydrology, 387, pp. 33–45, 2010.
[16] Bates, P.D. & De Roo, A.P.J., A simple raster-based model for flood
inundation simulation, Journal of Hydrology, 236, pp. 54–77, 2000.
[17] Aronica, G., Bates, P. D. & Horritt, M. S., Assessing the uncertainty in
distributed model predictions using observed binary pattern information
within GLUE, Hydrol. Process., 16, pp. 2001–2016, 2002.
[18] Xu, H., Modification of normalised difference water index (NDWI) to
enhance open water features in remotely sensed imagery, International
Journal of Remote sensing, 27 (14), pp. 3025–3033, 2006.
[19] Ho L.T.K., Umitsu M. & Yamaguchi Y., Flood hazard mapping by satellite
images and SRTM DEM in the Vu Gia – Thu Bon alluvial plain, Central
Vietnam, International Archives of the Photogrammetry, Remote
Sensing and Spatial Information Science, 38 (8), pp. 275–280, 2010.
Vulnerability to flood risks in

Japanese urban areas: crisis management
and emergency response for efficient
evacuation management
M. Thomas & T. Tsujimoto
Nagoya University, Department of Civil Engineering, Japan
Abstract
Today, flood risk in Japan occurs mainly in high density populated areas, as a
consequence of the rapid urban development of the deltaic plains of Japan during
the second half of the 20th century. At the end of the 20th century risk
management began to shift from mainly structural management to a more
“integrated” management. The evacuation process is one of the factors revealing
this shift. In Nagoya the evacuation process enhancement started with the Tokai
flood disaster (September 2000) and continues to this day. The most recent flood
events (urban flood of 2008 and typhoon No. 14 of 2011) highlight, however,
how the crisis management can still be vulnerable regarding evacuation. Our
research intends to assess the vulnerability factors of the crisis management
system, and especially of the evacuation process through interviews and a
questionnaire analysis method, in order to propose an integrated way of dealing
with evacuation in the case of a flood, imputing on GIS geographical as well as
social characteristics and evacuation patterns. Our research shows that the
evacuation process is effective despite low evacuation rate during past flood
event. In that regard improving the evacuation process cannot be separated from
the improvement of informational tools, but it can be seen that the possession of
hazard maps have few impact on evacuation decision. The efficiency of the
evacuation process in the case of a small to moderate flood event could therefore
be enhanced as the large-scale evacuation broadcast tends to target a population
in which more than half of the people do not need to evacuate. In the case of a
small flood event those repeated evacuation demands can increase a relatively
false sense of security and a loss of interest to flooding in general.
Keywords: vulnerability, adaptive capacity, floods, evacuation, GIS, Japan.
doi:10.2495/FRIAR140061
1 Introduction
With its three biggest cities located in deltaic fluvial plains, Japan is at a high
flood risk. In Japanese megacities, although innovative and efficient measures
have been taken, the flood risk remains, and the vulnerability to flood risk is
expected to increase with the combination of natural factors characterizing the
hazards (expected and unexpected effects of climate change), territorial factors
(settlements of stakes in lowlands area, apparition of “new” urban-type floods),
and societal factors (knowledge and acceptance of flood risk, willingness to
evacuate).
As the flood risk is changing, so is its management, and on concentrating our
research on the evacuation process the purpose of this paper is to define small
scale vulnerability of dwellings and its relationship to the evacuation
management at city and prefectural scale. Through the changes in flood risk
management since the disaster of the 11 and 12th September 2000 in Nagoya-city
will be analysed the vulnerability and adaptive capacity concepts to flood risks in
flood risk management, from risk actors standpoint and GIS mapping.
2 Theoretical framework for multi-scale vulnerability

analysis
2.1 Vulnerability and adaptive capacity concepts: towards integrated
system management
The concept of vulnerability is one of the numerous tools that can be used in risk
analysis. It has been described as the flip-side of the resilience concept [1] and of
the robustness concept [1, 3], as it is described in broad terms as “the
susceptibility to be harmed” [4]. The resilience is described as “the ability of a
system to absorb shocks, to avoid crossing a threshold into an alternate and
possibly irreversible new state, and to regenerate after disturbance” [5], and the
robustness concept as “a systems ability to remain functioning under
disturbances” [3]. But if the vulnerability has not always the positive aspect of
the here above two concepts, it consists in a helpful analysis tool when
confronted to the evaluation of a system’s evolution, as it can be considered as
“the potential for a change or transformation of the system when confronted with
a perturbation, rather than the outcome of this confrontation” [2].
The use of the vulnerability concept in risks studies has been used in two
related and complementary approaches. The first, classic (end of 1970s)
approach consisted of measuring the potential exposure of the different stakes in
a system [6]. In this case, the evaluation of the vulnerability is made through the
exposure as an attribute of the relationship between the system and the
perturbation [2]. It corresponds to a technical and effective management of
disasters, which could be summed as the following non mathematical equation:
Hazard × vulnerability (exposure) = risk [7, 8]
The second approach, enriched by research on natural hazards [9, 10], helped
to understand how the vulnerability concept cannot be taken outside of the
system it’s related to. The vulnerability being the propriety of a system, evolving
as the system evolves, and can be revealed during an event, or a disaster.
A key-component to the vulnerability concept is the “adaptive capacity” concept,
“the flexibility of ecosystems, and the ability of social systems to learn in
response to disturbances” [11, 12].
Because the flood risk management in Nagoya these last years knew
noticeable transformations, the vulnerability as propriety of a system increasing
this system to be harmed in case of a perturbation, as well as the concept of
adaptive capacity as formulated hereinabove by Turner et al. [12] will be used to
analyse the ability of the flood risk Japanese management system to evolve
during the past years and the influence of this evolution on the evacuation
process in Nagoya-city.
2.2 Vulnerability and adaptive capacity model
In order to build an efficient risk management, aside from the adaptive capacity
is also needed, and is paired with it, the ability to build efficient risk governance.
It has been made clear in natural hazards and in climate change research that
vulnerability is the propriety of a system. Megacities are a good example of
complexes, human-made, multi-level chain reactions, highly vulnerable type of
systems. Flood risk management in such systems needs the collaboration of very
different actors, with purposes and focuses that may differ. This is why it is
needed to take into account risk governance in the risk management system
vulnerability and adaptive capacity factors.
Building risk governance can be summarized in collecting, analysing and
communicating relevant risk information (through a complex web of actors,
rules, conventions, processes and mechanisms), taking risk management
decisions at the right time, and for those information and decisions to be
understood by the public concerned [13]. Although a difference has to be made
between risk management and risk governance. The definitions for
risk management are scarce, and can have different meanings, from risk
response – the risk management being the management of the crisis to which the
actors are confronted [14] – to the management of the risk at all times of the risk
(mitigation, preparedness, response and recovery [15]). Will be considered here
that risk governance is part of the risk management system, which entails the
different actors, the actions they decide to setup (hardware and software
measures), and the concrete results in the risk system and the consequences those
measures will have. Risk governance would be then the central part of the risk
management.
Building an efficient risk governance, to go further, includes the idea brought
up by the studies on climate change that risks in general have to be thought as
long-term duration processes. Adger [16] made a clear differentiation between
effectiveness and efficiency in adaptability capacity. The purpose of
effectiveness consists in responding to objectives that have been fixed, in
reducing the impacts of hazards and exposure, or to reduce the risk and avoid the
danger (in case of floods, building levees or flood-controlling dams are

considered as effective measures). An effective risk management reduces
therefore the vulnerability to a certain type of flood risk, at a certain time. An
efficient adaptability capability has to take into account, in every measure
considered, not only the effectiveness at a point in time and space, but also on
long-term and at wide-range scale. It consists most of the time in an economic
analysis, but also to the evaluation of the cost-benefit brought by the changes to
come that cannot be calculated, and on the timing on the adaptation action.
Effective risk governance, leading to an effective risk management, will have
immediate visible and invisible results in the adaptation capacity (dike
reinforcement). An efficient one will try to take into account the long-term effect
and the different outcomes of these measures (loss of landscape, oblivion of the
flood risk to riverine population etc.).
For the vulnerability to be durably reduced, and for the adaptive capacity to
be efficient on long-term management, efficient risk governance and therefore
risk management is needed.
Figure 1: Vulnerability assessment model (from Smit and Wandel [17]).
3 Methodology
3.1 Study area
In order to analyse the vulnerability and adaptive capacity of megacities in Japan,

it is easier to compare the risk management at two stages, preferably before and
after a memorable event. The event doesn’t have to be a disaster per say,
however the disastrous events give the opportunity to reveal vulnerability as well
as adaptive capacity, and therefore Nagoya-city and the 2000 Tokai flood
disaster were chosen as the starting point of a multi-scale vulnerability and
adaptive capacity analysis.
Nagoya-city is the 4th largest Japanese city in Japan, with a total population of
2,272,075 [18] on the 1st of January 2014. Floods in Nagoya-city are not an
unknown event, the main water-related disasters that happened during the past
70 years were the Ise-bay typhoon of 1959, the flood of 2000 (called Tokai
flood) and the 2008 urban flood. The Tokai flood occurred on September 11 and
12, and damaged part of the Tokai region, due to heavy rainfall, amounting up to
a total of 567 mm (one third of the average annual rainfall). The Shin River and
Tenpaku River suffered levee breaches, the Shonai River and Yahagi River
flooded by levee overtopping, and the rainfall accumulating near the levees could
not be evacuated due to lack of drainage ability [19, 20]. The total loss for the
Tokai region reached the amount of 978.3 billions of yen, 155 injured people and
10 fatalities for the Tokai region, and 37% of Nagoya-city urban territory
flooded, 45 injured and 4 fatalities.
3.2 Interviews analysis grid
Between the 01/04/2012 to the 01/11/2012 exploratory semi-conductive

interviews to 32 risk managers were conducted (from State actors level to local
disaster manager actors) for Nagoya-city, and 13 individuals either living in
risky areas or having experienced the Tokai flood in September 2000.
As it is shown in table 1, the purpose of those interviews was in a first time to
setup the different factors of vulnerability and adaptive capacity. As for the risk
governance, being part of the risk management system and a condition to the
reduction of the vulnerability, it was set aside in a third category. In a second
time were interrogated the different factors of vulnerability, adaptive capacity
and the risk governance status through the effectiveness and efficiency concepts.
3.3 GIS small scale evacuation vulnerability
The interviews to the risk managers confirmed by a survey analysis realized in

2011 by Aichi prefecture [21] pointed out the actual problem of the evacuation
process in case of floods, (details in next section). The interviews to the
population helped understand the reasons why one would hesitate to evacuate
despite the evacuation recommendation, but did not help to understand how
much and in what measure the residents in risky area are vulnerable. Moreover,
neither of the interview methodology gave a clear understanding of the
small-scale vulnerability to evacuation.
The type of building and their vulnerability factors have been input (number
of floors, housing embankment) on GIS, crossed with information available for
the public, the Nagoya Hazard Maps and the three types of expected floods:
major flood from national class river, medium flood from prefectural class river
and small urban flood.
A total of 961 building were referenced in three city-blocks of the Nishi ward
in Nagoya-city (Komoharachou, Ashiharachou and Nakaotaisanchoume), the
field has been chosen for being in risky area, and having suffered from the Tokai
flood in 2000.
Table 1: Effectiveness and efficiency in risk management for evacuation

interview analysis grid.
INTERVIEWS
To official risk managers To population
Last emergency evacuation
Software measures setup
lived
Vulnerability Willingness to evacuate
factors Hardware measures setup
in the future
Risk knowledge
Changes in the flood risk Enhancements to

protection housing after 2000
Principal changes after
Emergency supplies
Adaptive capacity 2000
factors
Effectiveness Increase of the interest in
Main objectives of the
efficiency risk management
flood risk management
willingness to know
today
more about floods
Access to different data and

Actors in touch with
understanding data
Actors of the risk
Understanding
management
Risk governance communicated data
communication
Personal goals and their
integration to the risk
management system’s
understanding
4 Results
4.1 Risk management and adaptive capacity
It has to be pointed out that despite the amount of physical damages the human
damages during the Tokai flood in 2000 were low. The risk managing system
succeeded in an emergency evacuation of more than 5,500,000 people [19],
despite the absence of official evacuation process at that time. Evacuation was
difficult because it happened as an emergency measure after levees on Shin
River breached and most of the people had to evacuate while the water level was
high, but considered all in all successful. The Tokai flood marked for Nagoya the
starting point of the creation of evacuation process measure. The evacuation
process measure gathered old risk management actors: the national River Bureau
and municipal disaster prevention actors through the implementation of “watch”,

“danger”, and “evacuation” water level thresholds for a quicker and better
management, and new actors: media companies in charge of warning the
population of evacuation recommendation through cellular phones, radio, and
television. The Hazard Maps released in 2001 and revised in 2010, are sent
directly to the population and downloadable from the city mayor office [18].
Local disaster management bureau created maps of damages at the ward level, to
increase their local knowledge of risky areas, and some of them enhanced flood
risk knowledge through reunions and seminars on flood risk. Hardware measures
have been enhanced the year following the disaster, and were finished in 2005.
In 2011 during an event with a similar hazard to the Tokai flood, it has been
demonstrated that the hardware measures setup between 2001 and 2006 were
effectively containing the hazard. However, at that time, communication
misunderstandings between old and new actors of the evacuation management
lead to the evacuation recommendation for one million persons, thus exceeding
by far the actual need for evacuation for this event.
4.2 Evacuation as a vulnerability factor for population?
It has been difficult to find persons willing to talk about flood disaster and
evacuation processes. As a consequence, the data collected during the interviews
to the population were analyzed in comparison with the survey for evacuation in
case of flood realized by Aichi prefecture in December 2011 (3 months after the
2011 flood event).
When interrogated about the risk culture, most of the interviewees answer
knowing the major past flood disasters (Ise-bay typhoon and Tokai flood), but
also the small event of the precedent year. The same findings have been found by
the Aichi prefecture survey, although during the interviews it was obvious that
resident who experienced the Tokai flood were reluctant to refer to it when asked
about floods (flood risk is accepted, but with limits). The risk culture is also
enhanced by the preparation (survival kits, knowledge of safe areas and risky
areas) to more general type of disasters (earthquakes).
The Tokai flood seems to be remembered by the people who lived in flooded
areas in 2000. It is not described as a shock, although the experiences related
clearly showed that the experience was not pleasant: car stuck in the high water
or because of the traffic, and impossibility to move, doubt about what to do and
where to go when the water reached the house… The rare persons talking about
the flood without referring to unpleasant experience were two people who stayed
at home “because I knew it to be safe”. The levee breaches especially seem to
remain a shocking enough event to be considered in the survey [21] a risk more
important than other flooding types. The trust bestowed upon authorities was not
a subject broached during the interviews. Most of the interviewed people seemed
to have high confidence in structural measures realized after 2000, but somehow
did not seem to be interested in evacuation warnings (most of the time “my
house if safe” was the main reason called upon). Comparing these results with
the 2011 Aichi prefecture survey, it seems that the trust in authorities is high
(expectation for government and the local residents association to improve the
disaster prevention effectiveness and the information delivery), and so is the

current medium chosen to deliver information (television, radio). Although if
the distributed hazard maps seems to be for the population, a valid teaching
material in their opinion, only 15% of the surveyed population responded that
they read and understood it.
4.3 Exposure to flood risk and needed evacuation
Comparing the empirical data gathered during on the field and the statistical data
available for housing and households from Nagoya office, an error of 268
apartments (18% of all the 1424 apartments evaluated) have been noticed and
will be taken into account when evaluating the number of households
and persons in need of an evacuation for the three reference model floods. The
number of family members living in one household will also be considered.
The Komoharachou district being more an industrial type of district, the average
family members for a household is 1.7 whereas in the two other districts, it
reaches 2.5.
The repartition between housing and non-housing buildings is close to equal
(46.5% non-housing building for 56.5% building housing). Repartition of
evaluated population by building class is however very differentiated (figures 2
and 3 and table 2).
Depending on the hazard type the rate of persons that need evacuation differs,
knowing that most of the people living in risky areas live in high condominiums
(63%) they are therefore not in grave danger in case of a flood disaster, and a
large amount of persons do not need to evacuate, even if the more vulnerable
private two-stories house type is the most common. For Shonai River model
flooding 41% of the population evaluated would need to evacuate, 39% for a
Shin River model flood, and 0.8% on case of urban flooding.
In order to represent population need of evacuation in case of flood, maps
have been realized, to help understand patterns in housing vulnerabilities. The
repartition of housing is however too homogenous in these districts to notice
patterns.
C C
3% 1%
A B
17% 36%
A
B 63%
80%
Figure 2: Left: Housing type repartition (C = 1 floor, B= 2 floors,

A = 3 floors and more); Right: Apartments repartition by housing
type.
Figure 3: Housing evacuation needed for Shonai River model flood.
Table 2: Evacuation needed evaluation.
Shonai Shin Urban Total

flood flood flood population
evacuation 1030 976 20 2499
no evacuation 1580 1583 2139 2499
5 Discussion
Adaptive capacity is one of the strong points of the risk management in Nagoya-
city. Confronted to a new kind of disaster, the risk management system respond
quickly and effectively (has been seen in 2000, and 2008) to new kind of hazards,
and new kind of disasters. The lessons learned from the past, and different
disasters are also observable (1995 Kobe earthquake), making the flood
risk management system flexible. This adaptive capacity allows the flood risk
management to evolve and to integrate new actors, and new purposes, building
therefore better risk governance.
The interviews to the population revealed less clear adaptive capacity, and
more potential vulnerability to flood in the fact that it is difficult to evaluate the
number of persons willing to evacuate in case of a disaster. From the population
standpoint, it was also difficult to evaluate their perception of the risk
management and how they were integrated in it, as the answer both to the
interviews and the survey [21] were unclear. It can be said, though, that
the acceptance of the flood risk might be high, and the preparedness good,
demonstrating a good adaptive capacity. The acceptance of a flood disaster is
very much less clear, as is the acceptance of an unnecessary evacuation warning.

Therefore it was found the need to evaluate the actual necessity for evacuation in
case of a flood disaster like the Tokai flood.
From the GIS data gathered, the difference of people asked to evacuate in
case of a flood warning and evacuation recommendation might be high. As said
previously, more than one million people were asked to evacuate during the last
event in 2012, due to a lack of understanding between old actors and new actors
of the risk management. But even so, in one city-block and for the worst-case
river flood scenario, 41% of the population should evacuate to be safe, and in the
least dangerous case 0.8%. If broad evacuation recommendation is effective in a
short-term goal, it may be not as efficient as wanted to answer to long-term goals.
6 Conclusion
Japanese flood management system in Nagoya-city has changed during these
past years, demonstrating a high adaptive capacity to new challenges flood risk
poses in megacities today. New flood risks appeared; they are caused by more
intensive hazards or generated by urban shape. The risk management aims and
succeeds in improving structural measures, developing software measures
through the integration of new actors, developing a more integrated management
and therefore better risk governance for flood risks. In that regard, the flood risk
vulnerability can be considered low, and the system aiming for efficiency. The
low evacuation rates for the last flood events seem to better correspond to
effectiveness achievement goal. Efficiency in evacuation procedures would be
better achieved with a clear understanding for the population of the received
information, making it easier for them to make a choice when confronted to a
disaster. On that matter, risk governance still has progress to do, despite the high
adaptive capacity of both risk managers and residents in risky areas.
References
[1] Folke, C., Carpenter, S., Elmqvist, T., Gunderson, L., Holling, C.S.,
Walker, B., Bengtsson, J., Berkes, F., Colding, J., Danell, K., Falkenmark,
M., Gordon, L., Kaspersson, R., Kautsky, N., Kinzig, A., Levin, S.A.,
Maler, K.-G., Moberg, F., Ohlsson, L., Olsson, P., Ostrom, E., Reid, W.,
Rockstro¨ m, J., Savenije, H., Svedin, U., Resilience and sustainable
development: building adaptive capacity in a world of transformations.
Report for the Swedish Environmental Advisory Council 2002:1, ed.
Ministry of the Environment, Stokholm, pp. 437-440, 2002.
[2] Gallopin, C.G., Linkages between vulnerability, resilience, and adaptive
capacity. Global Environmental Change, 16, pp. 293-303, 2006.
[3] Mens, M.J.P., Klijin, F., de Brujin, K. M., van Beek, E., The meaning of
system robustness for flood risk management, Environmental Science &
Policy, 14, pp. 1121-1131, 2011.
[4] Adger, W. N., Vulnerability, Global Environmental Change, 16,
pp. 268-281, 2006.
[5] Miller, F., Osbahr, H., Boyd, E., Thomalla, F., Sukaina, B., Ziervogel, G.,
Walker, B., Birkmann, J., van der Leeuw, S., Rockström, J., Hinkel, J.,
Downing, T., Floke, K., Nelson, D., Resilience and vulnerability:
complementary or conflicting concepts? Ecology and Society 15(3):11,
pp. 1-25, 2010.
[6] Thouret, J. C., D’Ercole, R., Vulnérabilité aux risques naturels en milieu
humains : effets, facteurs et réponses sociales, Cahiers de Sciences
Humaines, 32(2), pp. 407-422, 1996.
[7] Reghezza, M., Réflexions autour de la vulnérabilité métropolitaine : la
métropole parisienne face au risque de crue centennale, doctorate thesis,
Paris Nanterre-University, pp. 58-63, 2006.
[8] Reghezza, M., La vulnérabilité, un concept problématique (chapter 3), la
vulnérabilité des sociétés et des territoires face aux menaces naturelles,
ed. F. Leone, F. Vinet, PUM: Montpellier, pp. 35-41, 2005.
[9] Wisner B., Westgate, K., O’Keefe, P., Taking the Naturalness out of
Natural Disasters , Nature, 260, pp. 566-567, 1976.
[10] Blaikie I. et al. At Risk: Natural Hazards, People’s Vulnerability, eds.
Routledge, London, pp. 88-123, 2003.
[11] Gunderson, L., Holling, C. S., Panarchy, Island: Washington DC,
pp. 103-120.
[12] Turner, B.L., Kasperson, R. E., Matson P. A., McCarthy J. J., Corell R.
W., Christensen L., Eckley N., Kasperson, J. X., Luers, A., Martello, M.
L., Polsky, C., Pulsipher, A., Schiller, A., A framework for vulnerability
analysis in sustainability science, Proceedings of the National Academy
of Sciences of the United States of America, 100(14), pp. 8074-8079, 2003.
[13] Renn, O., Risk Governance. Coping with uncertainty in a complex world.
London, Earthscan, pp. 5-11.
[14] Heitzmann, K., Sudharshan, C., Siegel, P. B., Guidelines for assessing the
sources of risk and vulnerability, Social protection discussion paper series,
0218, pp. 1-60, 2002.
[15] Schelfault, K., Pannermans, B., van der Craats, I., Krywkow J., Mysiak, J.,
Cools J., Bringing flood resilience into practice: the FREEMAN project,
Environmental Science & policy, 14, pp. 825-833, 2005.
[16] Adger, W. N., Successful adaptation to climate change across scales,
Global Environmental Change, 15, pp. 77-86, 2005.
[17] Smit, B., Wandel, J., Adaptation, adaptive capacity, and vulnerability.
Global Environmental Change, 16, pp. 282-292, 2006.
[18] Nagoya-city Mayor office, http://www.city.nagoya.jp/
[19] Tominaga, A., Lessons learned from Tokai heavy rainfall, Journal of
Disaster Research, 2(1), pp. 50-51, 2007.
[20] Zhai, G., Public preference and willingness to pay for flood risk reduction
(chapter 4), Toward resilient society to emerging disaster risks in mega-
cities, eds Ikeda S., Fukuzono T., Sato T., TERRAPUB and NIED,
pp. 57-87, 2006.
[21] Aichi prefecture, Survey concerning evacuation in case of flood, pp. 1-63,
2011 (Japanese).
Section 3
Flood management
Community-based flood risk management:

lessons learned from the 2011 flood in
central Thailand
N. Jukrkorn1, H. Sachdev2 & O. Panya3
1
Research and Development Institute,
Phranakorn Rajbhat University, Thailand
2
Faculty of Environment and Resource Study,
Mahidol University, Thailand
3
Greenpeace South East Asia
Abstract
Thailand mega floods in 2011 highlighted the need for an integrated approach to
a flood risk management approach, combining local level community-based
action and a national strategic policy in preparation and reduction of
vulnerability of a country as a whole. This paper provides fact about a flood
crisis in 2011 and a set of lessons learned of community-based flood risk
management from affected communities scattered around the great flood areas in
central Thailand. Data and insightful information were drawn from a field visit
and a three-day participatory workshop attended by over 50 participants who had
experience of the flood. Included in this were community people, representatives
of local administration organizations and centralized agencies responsible for
dealing with natural disaster and crises.
Lessons learned from the workshop are conceptualized into six knowledge
platforms (KPs), highlighting the community best practices in response to the
situation during and after the crisis. They include 1) structural measures;
2) nonstructural measures; 3) emergency responses; 4) how to cope with the
community financial risk; 5) risk information and decision making; 6) dealing
with floods crisis recovery planning.
Keywords: Thailand, community-based, flood response, flood risk management.
doi:10.2495/FRIAR140071
1 Introduction
The concept of flood risk management (FRM) has been widely embraced over
the past decade. In many instances this conceptual acceptance has resulted in a
change in decision making and practices highlighting risk management as
potentially more complex, but more efficient and effective than a traditional
engineering standard-based approach (Sayers et al. [3]).
The 2011 flooding crisis in Thailand undermined public confidence in the
capacity of government to manage water resources, to guide responsible
development, and to tackle recurrent and unforeseen emergencies. Many agreed
that what was missing from the country’s emergency flood response.
In retrospect, the problem was that the government pursued these measures
with limited engagement of civil society, civil volunteers, the private sector, and
the non-profit sector. As people suffered tremendously, the experience has
propelled communities to take collective action in a manner consistent with
traditional values of self-reliance. In these circumstances, citizens and civil
society organizations began to help themselves by recruiting volunteers to gather
new information on households in affected communities, and confirming the
information and knowledge that they needed to secure proper assistance. Local
communities and authorities, therefore, played a lead role in FRM, both in the
short and long terms. It is also importance for local communities to review the
process of FRM through identifying lessons from their knowledge and
experiences that could be learned from past experiences and make improvements
for future practices.
Data and information were drawn from field visit and a three-day
participatory workshop. Over 50 participants who had had experience of the
flood from both urban and rural community attended. As well, representatives of
local administration organizations and centralize agencies responsible for dealing
with natural disaster and crises. The workshop was manage based on the
dialogue theory that can better motivate people to share experiences and
knowledge (Bohm [2]). Every session of the workshop was taped recorded and
subsequently transcribed into a text form of over 300 pages. Ethnographic
interpretation was undertaken in order to draw upon some measures and future
mechanisms that community could be integrated into future flood risk
management.
2 Facts about the 2011 flood crisis in Thailand: an overview

Thailand had not foreseen an event of flood disaster. In the year 2011, the event
was a high-impact and chaos phenomenon. Event with a low probability of
occurrence, the damage was unprecedented and enormous. The floods hit 65
provinces, including 1052 districts, and 75 main highways were inaccessible.
More than 1.5 million village people had suffered, more than 750 people were
dead, and over 4 million people lives were affected. The country’s loss of
farmland is estimated at 5,110,327 acres of land cultivated, while 60,124 acres of
fish ponds, animals were also affected by the floods. The losses of baht 32.41
billion was estimated, due to the flooding impact over a long period of time.
Major drivers causing the flood include the following.
2.1 The highest record of rainfall and tropical storms and the flow capacity
of rivers
Since 1901, the rainfalls in 2011 were the historical record in Thailand (World
Bank 2012). The heavy rainfalls in the latter period were the consequences of 5
tropical storms, (between the end of June to the beginning of October 2011 – see
also Figure 1). Together, a study of World Bank (2012) states that one of the
main causes of the flood crisis in 2011 was the low-flow capacity of the river
(Lower North and Central Plains of Chao Phraya River and tributaries – see also
Figure 2). Hence, water runoffs from major rivers had caused the overtopping of
river dykes and breaching in any river tributaries.
Figure 1: Average cumulative annual rainfalls – 1960–2011. (Source:

Thailand Integrated Water Resources Management
(www.thaiwater.net).)
2.2 Country unplanned urbanization and land use change
Rapid and unplanned urbanization and unsuitable land use in the flood plain
areas is probably one of the most important factors worsening the floods in 2011.
Ayutthaya province, where industrial and housing estates were located in the
areas, were supposed to be the flood plains many infrastructural facilities had
caused the blockage of the flood way. In Nonthaburi province, especially in
Bangyai district, as semi urban-rural area located at western side of the lower
Chao Praya river basin, flood plains and canals were also blocked by both the
public and private infrastructure and urban sprawls. Many public canals simply
disappeared because of illegal encroachments. Such changes in land use took
away the ability to drain water from the northern part of Bangkok into the canals
and drainage systems, and then to the drainage stations by the sea coast of the
city.
Figure 2: Chao Phraya River and tributaries. (Source: Google Maps.)
2.3 Central government flood mismanagement and political intervention
The floods crisis in 2011 was made worsened by man-made mistakes,

particularly from the Central government mismanagement and political
intervention includes:
2.3.1 The weakness of the flood master plan, action plan, and policy
responses from the central government
Despite the severity of flood and the government quick responses, evidence
showed there has been no concrete studies on the impact of the 2011 flood, by
drafting a flood management master plan and allocating about 330,000 million
baht (USD 11.3 billion) for the flood protection action plan as well as assistance
and compensation for the flood victims. The action plan budget consisted of
immediate flood compensation budget and budget for the flood action plans.
Although the master plan consisted of both the plan on infrastructural
investment, rehabilitation and maintenance, and the non-infrastructural plan, it
does not give much attention to the latter, particularly in term of local
agencies/communities involvement. No concrete policy and measures have been
proposed, specially, inadequate attention to the complex long-term issues of
fragmented water management and required institutional changes of integrated
water management to cope extreme weather conditions, the appropriate
combination of single command authority and decentralization.
2.3.2 Weaknesses of existing operation and of major reservoirs

Especially, the inflexibility on changes from higher authority when they needed
to quickly change the schedule of gate opening in response to the emergency.
Second, there was a lack of effective flood forecasting and early warning
systems. Third, it had a lot to do with inadequate information on change in

cropping patterns which affect the detailed gate operation schedule.
2.3.3 Irrigation facilities maintenance failure

During the flood crisis, at least 13 sluice gates were damaged, 3 of which
collapsed and had caused big floods in some areas. The damages were not only
caused by the big flood but also by the lack of proper maintenance of the flood
protection infrastructure, which was the primary reason for structural failure and
breaches of the flood protection embankment along the Chao Phraya River.
2.3.4 Emergency mismanagement

Slow responses to major sluice gate breakdowns, especially Bang Chom Sri
sluice gate’s collapse, clearly evident. There was too much water flowing into
the entire Lopburi Province, and then and continued downwards to Ayuthaya
district via Lopburi River. Not only because of the slow response, but the repair
of Bang Chom Sri sluice gate was left to the resource-poor local government
instead of professional central authorities.
2.3.5 Political intervention on dam operation and irrigation management

Along the Chao Phraya River, there were several barrages and dams that were
used for regulating water for irrigation and flood management. Anyhow, there
were, as newspaper reports claimed, some influential politicians might have
influenced the decision in controlling the sluice gates and to delay the water
discharge into one of the western provinces for some periods of time to allow the
farmers in their constituency to harvest their rice crop.
3 Key lessons learned from the flood in 2011

Key lessons learned from the workshop delivered a set of lesson learned, sharing
and exchanging among the participants. The discussion and synthesize cover
what worked, what did not, and why in the response to the year 2011 flooding
crisis. The event highlighted the key points in the field of community flood risk
management and disaster response to flood detention areas, are as follows.
3.1 Information and communication management is crucial in emergencies
During the flood crisis, all participants pointed to two common information
problems: i) the lack of real-time information on conditions and on coordination
among parties (that is, on who is doing what); and ii) the loss of critical public
records vital to reconstruction.
With regards to the first point, during the floods crisis the national
government collected information from municipal governments, while additional
information was crowd-sourced and channeled through social media and the
Internet. Many post-disaster situations were made worse by the lack of
communications strategies that make use of appropriate media to deliver critical
messages. Good information enabled individuals and communities to not only
stay safe but also contribute more effectively to relief and recovery. It also
ensured that communities have a realistic set of expectations about relief and
reconstruction. If communication was to help people stay safe and minimize the
disruption to their lives, they must be able to trust the information and its
sources. Together, communication regarding evacuation, temporary shelters, and
emergency food distribution was mismanaged, creating confusion throughout the
crisis phase of the flood.
3.2 New crowd sourced information and the use of social media
Social media, “community” radio stations were extensively used for searches and
rescues. Social media included web-based applications that use the Internet to
connect people (prominent examples are Twitter and Facebook), web sites and
computer applications that enable users to collaborate and create contents, such
as YouTube. Emergency “community” local radio stations also played a crucial
role in the aftermath of the crisis. When the emergency communication systems
in many cities broke down due to power failures and lack of emergency backup
power, community radio stations were able to get useful information out to
residents.
With the relatively high levels of mobile phone penetration in Thailand, social
media could be very useful during disasters, to the extent that they are already
used in normal times. They could also serve to link up with communities outside
the flood-stricken areas to facilitate the acquisition and allocation of aid and
assistance. In many developing countries, lack of physical accessibility to
disaster-affected sites is a key issue. Mobile networks and social media can be
used to collect and share localized information to improve accesses to rescue and
relief efforts. Reliability and trustworthiness of information is an extremely
important factor in the use of social media.
3.3 Spreading a better understanding of risk planning and risk-assessment

technologies need to be understood
At the beginning, the government predicted a low probability of the floods risk
occurring, and underestimated its size and the incoming monsoon risks. The
official risk depicts areas that were small than the area actually affected by the
floods.
In addition, accurate risk assessment and interactive communication systems
which could connect local communities, government agencies, and experts,
made people less vulnerable and more resilient. We have learned that under the
enormous crisis and mismanagement of communication, community members
should not be encouraged to stick to a single scenario. Community networking –
“flood information on land,” (e.g., rise of water levels and flooding areas) would
allow what was happening and what kind of preparation needed to the upstream
and downstream flooding indication. At the same time, people in these
communities all needed as frequent weather forecasts—“information from the
sky,” (weather monitor and forecast) as much as possible.
A better understanding of nature and limitations of risk planning among
communities, local authorities and the population at large would have to improve
their collective and individual decision making, especially in emergencies.

Communication about the unfolding disaster could and should have been more
interactive among local communities, governments, and experts. Distributing
risks plans and issuing early warnings were not enough. In the event, the
magnitude of the floods crisis was underestimated, which may have led people to
delay their evacuation.
3.4 Providing better evacuation centres, considering vulnerable groups and

gender sensitivity
At evacuation centers, the needs of women, kids, cross cultural people and the
disabled were not fully met. The overwhelming majority of the leaders of
community organizations were male. Relief goods delivered to the shelters were
biased in favor of male evacuees. New measures are needed to assure privacy
and security for all vulnerable groups and should be planned in advance. These
measures call for empowering marginalized groups for long-term recovery and
including a gender perspective in planning and managing shelters, which will
require women to be more deeply involved in shelter management. Vulnerable
groups must not only be protected but also engaged in decision making.
Understanding and meeting the challenges of the elderly, children, and women,
both during the emergency and in its aftermath, are priorities for effective post
disaster response. Local cultural knowledge sound solutions that take account of
special needs among segments of the population should be planned in advance in
order to enhance resilience and facilitate recovery and reconstruction.
During the crisis, it was reported that shelters provided for did not give
sufficient privacy for anyone, particularly for women, many of whom did not
have private spaces where they could change their clothes or breast-feed their
babies. At the peak of the relief effort, more than 2,000 people were housed in
one evacuation centre, while some left their communities and stayed with
relatives and friends who lives outside the flooded area. Most facilities, such as,
schools and community centers, were publicly owned and were urgently set up
as evacuation centers.
3.5 Recovery planning on debris and waste management
In the areas affected by the floods crisis, community representatives were

organized on recovery planning committees from the earliest stages. The local
governments outside the disaster-affected area helped affected municipalities
plan their recovery. There was an urgent need to dispose tons of debris left
behind by the floods crisis. The debris was an enormous obstacle impeding
recovery plan. Among the many issues arose were that of the availability and
selection of storage sites, methods of incineration, decisions about recycling, and
waste treatment and disposals.
Because of the fact that maintaining existing sources of income and creating
jobs were crucial during the recovery plan for communities, local and municipal
governments were expected to professionally manage disaster related waste,
select different treatment, different disposal methods in accordance with the

composition of the debris.
3.6 The importance of community participation in flood risk management
During crisis, tradition of community participation in preparedness was a key

factor in minimizing the number of lives damaged and lost. The role of the
community goes far beyond evacuation. Prior measures to crisis should also be
provided. This includes risk planning and warning systems, and ongoing
education, all of which proved essential in the evacuation that followed the
floods crisis. Local governments and communities in affected areas served as
first responders, managers of evacuation centers, and planners of post disaster
rehabilitation.
After the crisis, flood risk reduction activities should be well integrated into
the daily lives of most communities, ensuring that awareness of floods risk
management is never far from their mind. The national and local governments
must recognize and support the involvement of the communities at risks through
laws and regulations that define roles and commitments. This could be defined
as a community-based approach in dealing with disasters. Decision making must
come from this community-based organization involving local governments,
organizations and people’s participation. Although managing evacuation centers
is a municipal responsibility, most municipalities in the disaster-affected areas
suffered staff losses, seriously weakening their capacity to cope with the
emergency as a result. At the beginning, most centers were supported by local
school teachers, volunteers, and other civil society groups. As the evacuation
period lengthened, evacuees themselves started taking initiatives to manage their
communities. Experienced from the crisis, all participants were in agreement that
social safety nets for vulnerable groups are needed in times of emergency and
during recovery as a priority.
3.7 Coordination mechanisms on the ground should be agreed upon before

the fact
During the crisis, coordination among various groups, such as governments

(national and local), civil society organizations (CSOs), and private entities was
often poor – or at least not optimal. Local governments, whose facilities in some
cases were wiped out by the disaster, had little experience working with other
large-scale organizations, As a result, they received insufficient supports from
the central government in managing the new forms of cooperation.
Overall, the coordination system among local governmental organizations with
the communities, central government agencies and relief organizations and
donors was not up to the unprecedented task.
Effective coordination from stakeholders must develop. Although the national
government managed to establish the rescue headquarters very quickly with inter
prefectural emergencies and rescue units and technical forces were deployed in
record time, the mechanisms for formal coordination among the various
stakeholders (government agencies at all levels, CSOs, and private entities) were
inadequate. One weakness of coordination observed on the ground during the

flood demonstrated that coordination mechanisms should have been established
with advanced agreements and clear definitions of responsibility.
3.8 The need for a holistic approach to floods risk-management
Single-sector development planning cannot address the complexity of problems

posed by floods disasters. Faced with complex risks, flood-effected communities
chose to build capacity by investing in preventive structural and nonstructural
measures, by nurturing local culture and learning from past disasters, and by
promoting cooperation among multiple stakeholders, between government
agencies and ministries, between the private sector and the government, and
between multiple levels of government, and from local to national levels. The
essence of the approach is to design and maintain resilient infrastructure capable
of absorbing damages caused by flood and natural disasters to an extent that they
exceed all feasible and affordable measures. In the wake of the floods disaster,
communities also recognized that additional efforts were required to plan and
design measures capable of countering events of low probability but high impact.
4 The guidelines for the future

Echoing the key lessons learned from the workshop are conceptualized as
Knowledge Platforms (KPs), highlighted the community best practices in the
field of flood risk management. These KPs were grouped into six clusters,
including the following.
4.1 Structural measures
Generally, check dams and dikes are both necessary and effective in preventing
ordinary floods, which are relatively frequent, but they are of limited use against
the extreme events that occur less frequently. As the case of Bang Chom Sri dam
showed, construction standards and stability performance under worst-case
scenarios should be further investigated. Structures should be able to withstand
floods that exceed their designed flow, reducing the force of the water before
they collapse and thereby mitigating damages.
The Central Government master plan generally put an emphasis on the
structural flood management and little attention was given to the issues of non-
structural aspects of flood management. Efforts within flood risk management
have to create solutions based on community ownership and consensus. By
preparing and increasing community awareness and capacity of local
governmental authorities to handle flood situations has been recognized as a
focal point for flood risk management. After mega flood, government launch a
mega project for flood protection, however it only focusing on construct dam and
flood way.
4.2 Non-structural measures
During the floods crisis, as occurred in many centers, a self-governing body

emerged, with leaders and members of various committees selected by the
evacuees themselves. Key actors were that of community-based organizations,
who had saved many lives and needs of the victims. When the crisis
management overwhelmed local agencies, local communities were forced to use
their own knowledge and resources to survive on flood crisis chaos, save lives
and assets. Fortunately, throughout the floods areas, communities had been
engaged in floods preparedness. Therefore, knowledge of community-based
flood risk management (CBFRM) is very useful for the FRM strategy in the
future.
4.3 Emergency response
Partnerships needed to facilitate emergency operations: Coordination among

governmental agencies, military forces, and other stakeholders in dealing with
the emergency was an overwhelming challenge. The system for delivery of relief
goods, evacuation centers and temporary housing should be supported from
professional logistics specialists from local government in unaffected areas. The
special needs of cross-cultural and vulnerable groups (including the elderly,
children and the disabled) needed to be included in transition-shelter initiatives.
In addition, the experience points to the importance of bringing in professional
staff to care for the disabled and vulnerable. Considering the difficulties faced by
local governments after the flood crisis, coordination mechanisms should be
established in the central government, or under an umbrella organization.
4.4 Coping with the financial risk
During the 2011 flood crisis, full financial impact (including direct and indirect
impacts) form flood disaster will not be known for some time. The government
must play an important role in alleviating the disaster’s impact on households
and businesses through measures that ensure the stability of the financial system,
timely approvals of supplementary budgets, and provisions for rapid
disbursement disaster assistances, all of which helped citizens jumpstart their
recovery processes. The financial resources for recovery and reconstruction are
being funded by taxes to avoid leaving the cost to future generations. Flood
insurance helps people get back on their feet. Governments can play an
important role in fostering the growth of this kind of infrastructure, thereby
enabling the private insurance industry to offer cost-effective and affordable
insurance solutions.
4.5 Risk information and decision making
Risk information is needed to be understood. Uncertainties associated with

floods risk probabilities should be assessed based on multiple options, taking
into account every conceivable eventuality and utilizing all the tools science has
to offer.
The sharing of information among governments, communities, and experts
left much to be desired. While science-based early-warning systems are
important during a disaster, it was best for the target population if it could
combine with information on the ground through regular sharing of pre-disaster
information at the local level. The sharing should be accompanied – over time
and with the community’s involvement – by disaster drills, community mapping
exercises, and other measures. In recent years, remote-sensing data have been
used around the world to rapidly map the damage resulting from natural
disasters.
4.6 Recovery planning
In the areas affected by the floods crisis, communities were organized on
recovery planning committees from the earliest stages. In general, authorities
should be prepared for disasters by designating temporary storage sites, traffic
routes for transporting waste; including hazardous and toxicity waste. The role of
the private sector in debris management, as well as cooperation with
organizations and government bodies outside the affected areas, should be
explored. The possibility of recycling should be considered. Finally, rice and
vegetable seeds as well as young fruit trees are of most needed by rural
communities, as they could save a great deal of money on food and begin a new
agricultural season without spending money on much of them.
5 Conclusion
Lesson learned from Thailand has pointed that, flood risk management
implemented by communities and local government is crucial. Communities and
local government have the opportunity to design solutions that are adaptable to
the needs of their local communities and are consistent with local policies and
priorities. The measures and possible future mechanisms in addressing
community’s flood risk-management. From the people’s view, the integrated
flood management mechanism does not have to rely entirely on the: “predict-
and-act’ approach, which is conventionally used for the designed structural
measures. The Thai experience showed that success of flood risk management
lay in community involvement. Effective flood risk management requires close
coordination among all affected areas, including all responsible municipalities as
well as their agencies and departments, in order to support all-inclusive and
country relevant solutions. Policy makers and urban/rural development experts –
both structural and non-structural aspects – should be well advised to listen to the
communities and empower them to be the focal part of the solution.
Acknowledgements
The authors wish to thank, Asian Development Bank, for providing the budget
for the workshop, community leader district and representatives from local
government agencies in Thawung and BangYai Sub-district, as well as
representatives from the Department of Disaster Prevention and Mitigation

Minister for insightful information.
References
[1] Asian Development Bank, Community-based Flood Risk Management and
Disaster Responses the report, 2013.
[2] Bohm, David, On Dialogue, London Routledge, 1996.
[3] Paul Sayers et al., Flood Risk Management A Strategic Approach, Paris
UNESCO, 2013.
[4] World Bank, Thai Flood 2011. Rapid Assessment for Resilient Recovery
and Reconstruction Planning, Bangkok, 2012.
Reservoir system operation using a diversion tunnel
J. Ji1, H. Kim2, M. Yu1, C. Choi1, J. Yi3 & J. Kang4

1
Department of Civil and Transportation Engineering,
Ajou University, South Korea
2
Water Supply Business Division,
Korea Water Resources Corporation, South Korea
3
Department of Civil Engineering, Ajou University, South Korea
4
Engineering Research Institute, Ajou University, South Korea
Abstract
Although available water resources are limited, water demand is continuously
increasing due to population increases, economic development, and additional
uses, such as recreational and environmental uses. Constructing new reservoirs
has traditionally been the approach to develop new water resources. However,
such construction can be hampered by negative perceptions, adverse
environmental effects, and opposition from NGOs to dam construction. Although
Andong and Imha reservoirs are located close to each other, and they have
similar hydrological and meteorological characteristics, the storage capacity of
Imha reservoir is only about half that of Andong reservoir. This makes the
operation of both reservoirs inefficient. This paper evaluates the effect of a
diversion tunnel connecting Andong and Imha in the flood season. Water yield
and spillway release reduction capability with 95% reliability were analyzed
using historical daily inflows data for 30 years. By changing the reservoir
operation methods, the reservoir system performance was evaluated. The system
operation of the reservoirs with the diversion tunnel showed better results than
the individual operation.
Keywords: flood control, reservoir system operation, diversion tunnel.
1 Introduction
Demand for water resources has been rapidly increasing because of population
increases and economic development. Recently, demand for water necessary for
recreation and environmental improvements has been also continuously
doi:10.2495/FRIAR140081
increasing. In the case of Korea, difficulties in water resource management have

been growing because approximately 70% of the mean annual precipitation
occurs in the flood season (June–September) and the rainfall concentration in the
flood season has also intensified recently.
The construction of new large-scale storage facilities is the optimum approach
to water resource security and flood management and preparedness. Although
constructing new dams is the best solution, difficulties may arise during the
development of such dams because of the diffusion of negative perceptions of
dam construction and opposition by environmental organizations and
communities.
Multilateral measures are necessary to solve problems arising from the
phenomenon of rainfall concentration. Methods of increasing the capacity to
the level necessary for water supply can be divided into structural methods and
nonstructural methods. Structural methods include dam raising and reservoir
sediment dredging, and nonstructural methods include reservoir reallocation in
flood seasons and multi-reservoir operation. Multipurpose dam operation, which
is a nonstructural method, focuses on water utilization and flood control. It can
be used to secure water resources by reducing the occurrence of floods and
drought and related damage through the efficient distribution of water resources
and operation of reservoirs.
With regard to studies on reservoir operation, the majority of past studies aimed
at reducing flood damage by focusing on the operation of single reservoirs and
flood control.
In contrast, most recent studies have focused on multi-reservoir operations,
utilizing generalized models based on complex analyses and multilateral
approaches.
Kojiri et al. [1] proposed a flood control system to calculate discharge by
applying 3 h of old inflow rate values to fuzzy inference and analyzed the
reservoir operation using four fuzzy sets. Cheng and Chau [2] studied the ability
of a reservoir flood control management system developed using programming
languages, such as FORTRN, C-language, and PowerBuilder, to mitigate flood
damage. Xiang et al. [3] developed a module for controlling restricted flood
water levels considering the uncertainty of inflows and applied it to the Three
Gorges Reservoir in China. In that study, dynamic control of restricted reservoir
flood water levels effectively facilitated hydroelectric power generation and
increased water utilization rates without increasing the risks of flooding. To
improve the rule curves for flood events using folded dynamic programming,
Kumar et al. [4] collected flood data from 1958 to 1995 and applied the data to
Hirakud reservoir in India that has been operated since 1956. Vonk et al. [5]
used the shortage index and mean annual energy production to analyze the
performance of a reservoir using an operation method of multiple purpose
reservoirs. They proposed a connected simulation optimizing method for
adapting to changes in water supply and demand.
In the present study, as a measure to minimize flood season spillway release
with a view to preventing floods and securing water resources that are limited,
the impact of a diversion tunnel between the Andong dam and Imha dam on the
system operation of the reservoirs was investigated. The reservoir operation

results were analyzed using actual discharge data for 30 years, and the effects of
the separate operation of the reservoirs, the connected operation of reservoirs,
and the reservoir system operation were reviewed.
2 Present state of basins and multipurpose dams

2.1 Overview of basins
Andong dam and Imha dam belong to the Nakdong river basin, which occupies
approximately 25% of the territory of Korea. Andong dam is located at the
uppermost stream of the Nakdong River, and Imha dam is located at
the Banbyeon stream, which is the first branch of the Nakdong River. The
Andong dam basin corresponds to approximately 6.8% (1,584 km2) of the entire
Nakdong river basin (23,384 km2), and the total length of flow paths in it is 31
km. The Imha dam basin occupies approximately 5.8% (1,361 km2) of the entire
Nakdong river basin area, and the total length of flow paths in it is 75 km.
Figure 1: Diagram of the Andong dam and Imha dam basins and the diversion
tunnel.
2.2 Specifications of the multipurpose reservoirs
Andong reservoir is a multipurpose reservoir located in the main stream of the

Nakdong River. The dam is 83 m high and 612 m long, and its total
impoundment is approximately 1,248 10 m . The area of the Andong
reservoir basin is 1,584 km2, and the reservoir supplies 926 10 m of water to
Gumi and Daegu annually. This reservoir’s annual power generation is 89 GWh,
and its designed maximum volume of water consumption for generation is
161 m3/s. The construction of Andong reservoir on the Nakdong river system
began in April 1971, and it was completed in October 1976. The aim of the
reservoir was to reduce flood damage in the downstream region and to secure
irrigation water, industrial water, and domestic water.
Imha reservoir was constructed 17.4 km above Andong-si. The dam is 73 m
high and 515 m long. Its total impoundment is 595 10 , and its basin area is
1,361 km2. Imha reservoir supplies 591.6 10 m of water annually, its annual
power generation is 78.7 GWh, and its designed maximum volume of water
consumption for generation is 119 m3/s. It is a multiple-purpose reservoir
constructed as part of a multipurpose water resource development project called
the Master Plan for the Development of Four Major River Basins. Construction
of the reservoir began in December 1984, and it was completed on December 31,
1993.
Although Andong reservoir and Imha reservoir are close to each other and
have similar basin areas, the reservoir storage of Imha reservoir is only
approximately 50% that of Andong reservoir. As Imha reservoir has a small
water bowl, at times of similar rainfall events to those encountered at Andong
reservoir, there are difficulties in flood control and in securing water utilization
capacity when spillway release occurs. Therefore, this paper analyzed the effects
of a diversion tunnel on the operational efficiency of both reservoirs.
2.3 Water supply plan
The reservoirs’ planned monthly water supply volumes were divided into
irrigation water, domestic water, industrial water, and instream flows. Andong
reservoir’s annual water supply is 926 10 m and that of Imha reservoir
is 591.6 10 m .
The annual domestic and industrial water supplies of Andong reservoir and
Imha reservoir are 450 10 m and 363.6 10 m , respectively. The annual
irrigation water supplied by Andong reservoir is 300 10 m , and that supplied
by Imha reservoir is 13 10 m . With regard to irrigation water, irrigation water
consumption in the basin during the busy farming season from April to October
is reflected in the planned monthly water supply volumes. Variations in the
planned monthly water supply volumes are larger in Andong reservoir compared
to Imha reservoir. The annual instream flow of Andong reservoir is 176
10 m and that of Imha reservoir is 215 10 m when the instream flow of
Imha reservoir includes the volume supplied to Yeongcheon raceway (4.8 m3/s).
In this study, normal monthly water supply volumes were selected to analyze
the effects of the connected operation (via the diversion tunnel) of the reservoirs
on water supply volumes and additional discharges.
2.4 Review conditions
Storage capacities were calculated using the reservoir continuity equation based
on the specifications of the Andong reservoir and Imha reservoir to determine
discharges, additional supply volumes, power generation discharges, and
spillway releases. The storage capacity of day t was determined by the storage
capacity and inflow of day t-1. The discharge volume and the volume of the
diversion tunnel on day t and the reservoir discharge volume were obtained by
applying an additional supply rate for day t to the basic planned (normal) supply
volume. To determine the discharge volume that leads to reduced supply or
spillway release, the calculated storage capacity of day t was compared to the
storage capacities that correspond to the full water level and the low water level.
The power generation discharges of the reservoirs were discharged first, and any
discharge volumes that occurred in excess of the maximum power generation
capacity of the two reservoirs were counted as spillway release volumes
(equation (1)).
, (1)
where is the storage volume on day t, is the storage volume on day t-1, I
is the inflow volume on day t, is the discharge volume on day t, and is the
diversion tunnel diversion volume on day t.
The volume of the diversion tunnel was calculated considering the
entrance/exit head losses and the friction head loss. This was considered the
reservoir’s inflow volume. In this case, 0.2 was used as the entrance loss
coefficient, and 1.0 was used as the exit loss coefficient. The friction head loss
was calculated using the Darcy–Weisbach formula, and a roughness coefficient
of 0.015 was assumed. The sum of the diversion volumes of the two reservoirs
through the diversion tunnel was 0, and the diversion volume of each reservoir
increased or decreased according to water movements between the two reservoirs
(from Andong to Imha or from Imha to Andong). The formulas for calculating
the volumes of the diversion tunnel resulting from head loss differences are as
follows:
∆H W (2)
∆
V (3)
∆
D AV (4)
.
f – , (5)
where ∆H is the water level difference between the two reservoirs (m),
is the entrance loss coefficient,
W is the water level (m) of Andong reservoir,
is the exit loss coefficient,
W is the water level (m) of Imha reservoir,
is the friction loss coefficient,
is the entrance head loss (m),
is the length (m) of the diversion tunnel,
is the exit head loss (m),
is the diameter (m) of the diversion tunnel,
is the friction head loss (m),
n is the roughness coefficient,
D is the volume (m /s) of the diversion tunnel diversion,
is the flow velocity (m/s) of the diversion tunnel
A is the cross-sectional area (m ) of the diversion tunnel.
No diversion through the tunnel occurs when the water levels of the two
reservoirs are lower than the height of the diversion tunnel and water can be
moved between the two reservoirs only when the water levels of the two
reservoirs are higher than the diversion tunnel. Therefore, if the water level of
only one reservoir is higher than the diversion tunnel, diversion will occur until
the higher water level of the reservoir goes down to the height of the diversion
tunnel. If both the water levels of the two reservoirs are higher than the diversion
tunnel, diversion will occur from the reservoir with the higher water level to the
reservoir with the lower water level until the water levels of the two reservoirs
become the same.
As both reservoirs are installed with hydroelectric generation facilities and
generate hydroelectric power through the power discharge. Generation is
calculated through the discharge. The power generation, P, was calculated using
equation (6), and a generator efficiency value of 0.86 and a hydraulic turbine
efficiency value of 0.95 were applied to both reservoirs.
P 9.81 , (6)
where P is the power generation (GWh),
γ is the generator efficiency,
γ is the hydraulic turbine efficiency,
Q is the reservoir discharge volume (m3) for time T,
H is the Imha reservoir water level (m),
T is the friction loss coefficient (h).
The head loss difference, H (equation (7)), value was obtained by deducting
the tail water level from the head water level. The low water level, which is the
head water level, was determined by the average of the water levels of the two
reservoirs at time t and the next time t+1. The tail water level is the water level at
which the water was discharged. This is usually determined by the water level of
the regulating reservoir. However, in this paper, the value obtained by deducting
the average value of tail water levels from the average value of forebay water
levels was applied because the water level of the regulating reservoir could not
be considered and the simulation was conducted focusing on main reservoirs.
′ ′
H , (7)
where W is the low water level at time t, W is the low water level at time
t 1, HWL′ is the full water level of the regulating reservoir, and LWL′ is the
low water level of the regulating reservoir
The number of days of water shortage was calculated based on the low water
level. Days on which the water level of the reservoir was the same as the low
water level were counted as days of water shortage. That is, cases where the
water level of the reservoir dropped to the low water level and could not satisfy
the basic supply volume were defined as cases of water shortage, and the
numbers of days of water shortage determined in this way were counted.
2.5 Reservoir system operation
Reservoir system operation is important when two or more reservoirs are

operated simultaneously and when those reservoirs are located in series or in
parallel. In such cases, the operation of one (e.g., inflow volumes and discharges)
affects the operation of the other, with at the same downstream point.
Simultaneous operation of reservoirs allows more efficient use of water
resources than operating individual reservoirs separately.
In the case of connected reservoir operations, two reservoirs are automatically
connected by a diversion tunnel. When the water level of any one reservoir is
higher than the diversion tunnel, the reservoir system operation occurs
automatically. However, if the water levels of the two reservoirs are lower than
the diversion tunnel, the two reservoirs will be operated separately.
The system operation is a nonstructural operation mode intended to secure
maximum water utilization capacity through the control of discharge volumes
between the reservoirs. It can be utilized for various purposes and takes the
characteristics of individual reservoirs into account to control flood peaks and to
optimize flood control. Hirsh et al. [6] advised that integrated reservoir system
operation has synergy effects and illustrated this through numerical experiments.
If the Andong reservoir and Imha reservoir are system operated, when a
downstream target point is considered, the water volume required at this point
can be satisfied when there is a water shortage in one of the reservoirs or when
the other reservoir has surplus water by discharging water at the necessary flow
rate. If these reservoirs are not system operated, if a water shortage occurs (a low
water level is reached) in either of the reservoirs, the water demand of the target
point cannot be satisfied. If the number of days of water shortage in both of the
reservoirs is counted, each occurrence of water shortage in the reservoir will be

counted in the number of days of water shortage.
Therefore, cases where system operation is in place cannot be compared with
cases where it is not in place based on the number of days of water shortage.
For this reason, to compare the number of days of water shortage under
system operation and nonsystem operation, the days were calculated using the
concept of deficit supply. In the deficit supply method, the number of days of
water shortage is not based on the concept of firm yield, which is the defined
supply volume, such as the basic planned discharge volume, guaranteed to be
supplied by each reservoir. Instead, it is based on the water volumes that have to
be supplied by any reservoir with extra storage capacity to downstream regions.
3 Application and results

In this study, the results of separate-system operation when the two reservoirs
were not connected and the results of connected-system operation when the two
reservoirs were connected through a diversion tunnel were analyzed. In the
analysis, additional discharge volumes, the number of times of spillway releases,
and spillway release volumes were reviewed based on the same number of days
(n=1,077) of water shortage. Here, the system operation of the two reservoirs is
discharging the sum of the basic planned water supply volumes of the two
reservoirs by assigning the water supply volume according to the ratios of the
amounts of storage of the two reservoirs (the current effective impoundment of
each reservoir/the sum of the effective storage capacities of the two reservoirs).
The additional supply volumes were divided into two equal parts and each part
was assigned to each of the two reservoirs.
As can be seen in the tables and figures (Table 1, Table 2, Table 3, Fig. 2,
Fig. 3), the largest volume of additional water was supplied during the connected
operation, followed by the separate-system operation and the connected-system
operation in order of precedence. The integrated reliability, which is the average
reliability of Andong reservoir and that of Imha reservoir, was calculated.
According to the results, the integrated reliability of the separate operation was
95.09%, that of the separate-system operation was 95.51%, that of the connected
operation was 95.71%, and that of the connected-system operation was 95.77%.
Therefore, the integrated reliability was improved in the connected-system
operation by 0.68% compared to that during the separate operation. When the
separate-system operation was conducted while the reservoirs were not
connected by the diversion tunnel, 1.47 m3/s of additional discharge was possible.
The number of times of spillway releases decreased by 4 compared to the
separate operation, and the spillway release volume decreased by 566 10 m .
When the connected-system operation was conducted by connecting the two
reservoirs with the diversion tunnel, 1.83 m3/s of additional discharge was
possible. The number of times of spillway releases decreased by 13 in total
compared to the separate operation, and the spillway release volume decreased
by 933 10 m . The annual total power generation increased by 1.9 GWh
during the connected-system operation compared to the separate operation,

although this difference was not considered significant.
Table 1: Additional supply volumes resulting from system operation.
Andong dam Imha dam Possible

annual
Additional Number of
System additional
discharge Number of Number of days of
operated days of % days of % supply
(/s shortage
shortage shortage volume
(10 m )
X - 223 97.97 854 92.21 1,077 -
Separate
O - 395 96.40 503 95.41 898 -
operation
O 1.47 473 95.69 604 94.49 1,077 46.3
X - 251 97.71 690 93.71 941 -
Connected X 1.00 272 97.52 805 92.66 1,077 31.5
operation O - 406 96.30 452 95.88 858 -
O 1.83 502 95.42 574 94.76 1,076 57.8
Table 2: The number of times of spillway releases and spillway release

volumes during system operation.
Number of times of spillway

Additional Spillway release volume
System releases
discharge
operated Andong Andong
(/s Imha dam SUM Imha dam SUM
dam dam
Separate X - 26 45 71 2,008 3,064 5,072
operation O 1.47 32 35 67 2,313 2,193 4,506
Connected X 1.00 31 26 57 2,368 1,810 4,178

operation O 1.83 31 27 58 2,373 1,766 4,139
Table 3: Annual power generation during system operation.
Annual power generation Possible annual

Additional Annual total
System additional
discharge Andong Imha power generation
operated SUM supply volume
( / dam dam (GWh
10 m
Separate X - 26 45 71 173.7 -
operation O 1.47 32 35 67 175.7 46.3
Connected X 1.00 31 26 57 178.1 31.5

operation O 1.83 31 27 58 175.6 57.8
Figure 2: The number of times of spillway releases during system operation.
Figure 3: Spillway release volumes during system operation.
4 Conclusion
In this study, the effects of the connected reservoir operation of Andong
reservoir and Imha reservoir using a diversion tunnel were analyzed to prevent
floods and to ensure an uninterrupted water supply during drought periods by
efficiently managing the water resources that are spillway-released during flood
seasons. Structural methods using a connected reservoir operation and a
nonstructural a system operation method were applied. Using daily discharge
data for 30 years from 1979 to 2008, daily simulations were conducted using the
reservoir continuity equation, and water yields and effects of reducing spillway
release volumes were reviewed based on 95% reliability.
According to the results of the analyses based on the concept of deficit supply,
the number of times of spillway releases and spillway release volumes decreased
the most during the separate operation of the reservoirs, followed by the
reservoir system operation, the connected operation of the reservoirs, and the
connected reservoir system operation in order of precedence. In addition, the
possible annual additional supply volumes increased the most during the separate
operation of the reservoirs, followed by the connected operation of the reservoirs,
reservoir system operation, and the connected reservoirs system operation in
order of precedence.
Although system operation without the diversion tunnel enabled reducing
spillway release volumes and securing additional supply volumes during drought
periods, larger effects were obtained when the diversion tunnel and system
operation were used simultaneously. Given these results, using the diversion
tunnel and system operation together is considered to result in a structurally
stable connected reservoir operation.
In this study, the effects of using the diversion tunnel and the results of
system operation were examined in terms of their flood season spillway release
reducing effects and water supply during water utilization periods. With regard
to connected reservoir operations, a comprehensive analysis of hydroelectric
power generation, water quality improving effects, and economic and
sociological benefits is needed to address.
Acknowledgement
This work was supported by a National Research Foundation of Korea (NRF)
grant funded by the Korean government (MEST) (No 2013-065006).
References
[1] Kojiri, K., Ikebuchi, S. & Yamada, H., Basinwide flood control system by
combining prediction and reservoir operation. Stochastic Hydrology and
Hydraulics, 3, pp. 31–49, 1989.
[2] Cheng, C. & Chau, K.W., Flood control management system for reservoirs.
Environmental Modeling & Software, 19, pp. 1141–1150, 2004.
[3] Xiang, L., Shenglian, G., Pan, L. & Guiya, C., Dynamic control of flood
limited water level for reservoir operation by considering inflow uncertainty.
Journal of Hydrology, 391, pp. 124–132, 2010.
[4] Kumar, N.D., Baliarsingh, F. & Raju, S.K., Optimal Reservoir Operation for
Flood Control Using Folded Dynamic Programming. Water Resour Manage,
24, pp. 1045–1064, 2010.
[5] Vonk, E., Xu, Y.P., Booji, M. J. & Augustijn, D.C., Adapting Multireservoir
Operation to Shifting Patterns of Water Supply and Demand. Water Resour
Manage, 28, pp. 625–643, 2014.
[6] Hirsh, R. M., Cohon, J. L. & ReVelle, C. S., Gains from joint operation of
multiple reservoir systems. Water Resources Research, 13, pp. 239–245,
1997.
Section 4
Considering ‘Blue-Green’
approaches to
Flood Risk Management
(Special session
organised by J. Lamond)
A conceptual framework for understanding

behaviours and attitudes around ‘Blue-Green’
approaches to Flood-Risk Management
G. Everett & J. Lamond
Abstract
This study develops a conceptual framework to inform thinking around the social
research approach adopted to consider the development of ‘Blue-Green’
approaches to Flood Risk Management (BG-FRM) in UK cities. The framework
informs the manner in which research is conducted and data analysed, to
understand current and possible future household and business behaviours as BG-
FRM becomes more established, and so possibly (or not) more ‘normalised’, as
well as the influences upon these behaviours that can potentially be played by key
stakeholders. A conceptual map is drawn up that outlines the key players, their
domains of agency and lines of influence concerning larger-scale (neighbourhood,
city-level) and smaller-scale (household, business) approaches. A conceptual
framework is then developed, thinking about the motivations and barriers that
could encourage or inhibit adoption of blue-green approaches and the behaviour
changes necessary for their sustainability, before surveying research already
conducted in this area. Social Practice Theory (SPT) is suggested as a new manner
of framing research to understand the ways in which behaviour may change, or
fail to change, and the opportunities and barriers to any such changes. SPT, it is
argued, could provide a means by which to consider present behaviours and
attitudes, so that we might more effectively look for opportunities to encourage
progressive behavioural developments that could increase the chances of BG-
FRM’s sustainability.
Keywords: Blue-Green, Flood-Risk Management, sustainable, behaviour, Social
Practice Theory (SPT).
doi:10.2495/FRIAR140091
1 Introduction
With the increased incidence of flood events in recent years and the hypothesis
that flooding may increase (or currently be increasing) as a result of development
pressures and climate change (King [1], Whitmarsh [2]), governments are taking
very seriously the need to deal with the economic and social threats from this
(DEFRA [3, 4], Environment Agency [5, 6], SEPA [7]). Thinking is shifting
away from simple notions of resisting outright inundation towards developing
resilience to flooding – living with water and making space for water gaining
prominence in academic literature and policy ([3], Pitt [8], Bowker [9], McBain
et al. [10]). Thinking has also moved away from erecting structural defences to
establishing softer and more sustainable FRM that retains, filters and makes use
of water-flows. The latter has begun happening for a number of reasons: the
environmental, aesthetic and socio-economic impacts of structural work; the
need to adapt urban areas to cope with a changing climate (using fewer resources
and emitting less waste), and an argued need to rethink our relationship with
water, reintegrating the natural water-cycle with the urban environment,
producing ‘water sensitive cities’ (Brown et al. [11]. Howe and Mitchell [12],
Kazmierczak and Carter [13], BGD [14]). BG-FRM approaches involve
improving green infrastructure, raising water-absorption capacity and promoting
natural channelling rather than containing and culverting (Abott et al. [15]).
A number of authors have begun publishing research around public attitudes
around Blue-Green approaches to Flood Risk Management (BG-FRM) (Bastien
et al. [16], Wright et al. [17], Kenyon [18], Apostolaki and Jefferies [19],
Werritty [20], Johnson and Priest [21]), some drawing conclusions as to how
people will behave around them. Results are apparently quite split between those
who found strong preferences for structural defences [20, 21] and others who
found preference for more sustainable solutions [16, 17]. In Apostolaki’s study,
awareness of SuDS’ flood functions was argued to be low, whilst others have
found it to be quite high (around 75% of respondents [19]).
What has not yet been done is to think about how behaviour might change
over time. Public attitudes may be cynicism and mistrust if people are not
involved in discussions from the outset. All parties will have a lot to contribute,
from scientific-technical assertions about BG-FRM in/efficacy to local
knowledge which could dismiss certain options or illustrate that others were
relevant and likely to work. Dialogic learning will be imperative to thinking
about viable BG-FRM options; for instance, significant differences have been
found to exist between actual and perceived SuDS’ safety levels (cf. McKissock
et al. [22]). Bastien et al. [16] and Apostolaki and Jefferies [19] found small
amounts of litter considered significant ‘pollution’, highlighting the need for
agreed maintenance systems. There could initially be hesitancy in uptake, with
safety fears around poorly lit green spaces (Bixler and Floyd [23]), water-butts
being seen as something for keen ‘productive’ gardeners (Chappells et al. [24]),
and green roofs and permeable paving possibly acceptable or workable only if no
behaviour-change were required (cf. Whitmarsh [25] on the asymmetry of
intentions and actions concerning climate change). There may be reluctance to
accept installation and maintenance costs, although Bastien found that

‘willingness to pay’ for pond amenities could potentially cover installation and
maintenance. Behavioural changes required would include emptying water-butts,
treating permeable with care and not littering ponds.
Research published so far highlights the need for further research to
investigate stakeholder preferences and the potential for change more closely.
Key questions would revolve around the normalisation of BG-FRM, how long
this takes and how it can be encouraged, as habits change and objects come to be
viewed, experienced and used differently. To adapt from Shove and Southerton
[26], ‘the business of becoming normal involves a two-way process in which
[SuDS] respond to their surroundings and at the same time impose something of
their own script’. Improving and maintaining BG-FRM would require changes in
behaviour to ensure functionality (for example, emptying water-butts, not
littering ponds and cleaning permeable paving), and no literature has yet
addressed this issue. Structural solutions work to a ‘fit and forget’ model for
most stakeholders. BG-FRM requires that people ‘live with water’. The more
people interact with BG-FRM, the more they may appreciate, value and want to
take care of it. This could be due to changes in how outdoor space is used
(improved green areas, relaxation and recreation) or shifts in observation and
appreciation of nature as green-cover brought flora and fauna to the city. Another
important factor would be the time it took for BG-FRM to offer services referred
to (flood protection, habitat provision, leisure space); several years’ would be
needed for flora-cover to offer significant water-absorbency. So BG-FRM would
not be a quick fix, but then larger-scale structural work could take as long from
inception to completion. Questions would arise as to how new infrastructure fits
into the routines and domestic lives of people affected, how practices could
change so they fitted more, and how new practices might come to seem normal.
Notions like the necessity of driveways for household parking, preferred
aesthetics for rooftops and ease of rooftop drainage as against water-butts would
all need to shift over time for true acceptance of a blue-green approach.
This paper will develop a conceptual framework for understanding
behaviours around BG-FRM and how these may or may not change over time,
looking to the incentives for and barriers to any such changes. It will argue the
need to look to future behaviour over current as this will change as new
infrastructures become more established and normalised. We argue that new
approaches to researching and understanding behaviour are needed, and suggest
an effective way forward. Section Two looks at the ‘system’ and ‘stakeholders’,
mapping out lines of influence for investigation. Section Three outlines the
theoretical approach in a Conceptual Framework considering motivations and
barriers to adopting BG-FRM. Section Four considers the ‘units’ of analysis at
play and argues for employing a Social Practice Theory (SPT) approach that can
account for the disconnect between potential and actual outcomes from
environmental programmes, and Section Five concludes. Reasons why further
research is needed are outlined over the course of the paper; over the past decade
debate has emerged around public opinions of BG-FRM, but this dialogue has
not yet considered how behaviour might alter with its establishment and
normalisation.
2 Stakeholder map
Understanding preferences and what affects these (and so behaviour) is a
necessary first step in negotiating, and overcoming, barriers and concerns. In
order to gain this understanding we need to draw up a stakeholder map before
outlining a conceptual framework of perceptions around BG-FRM, and thinking
through the motivators and barriers to behavioural changes that could facilitate
SuDS’ functioning. This work could then be used to monitor shifts in behaviour
and perceptions as BG-FRM grew, developed and became more normalised.
The system would need to be bounded to allow for proper analysis and
appraisal. An appropriate framing would be city boundaries, although some
agents will have wider operations and the watershed may stretch beyond, so
factors from outside would need consideration. Nonetheless, framing the
‘system’ at city level makes sense administratively and will be appropriate for
the majority of stakeholders. These would include communities (households,
governmental and non-governmental organisations, public service providers and
businesses and their representatives), the front end of dealing with inundation.
Environmental and wildlife groups will have a strong interest in BG-FRM, as
will landowning and advisory bodies such as farms and Natural England [27].
Water companies would be affected by BG-FRM insofar as water supply and
disposal of wastewaters would be altered by interventions. The Planning,
Development and Building industries would be affected, with new opportunities,
responsibilities and demands placed upon their work (RTPI [28], Scottish
Government [29]).
Key actors with responsibility for larger-scale BG-FRM would be Local
Authority, Environment Agency and national government bodies such as
DEFRA (see Figure 1). However the aggregate of household- and business-level
BG-FRM
could also make a significant contribution to reducing flooding. The Commission
for Architecture and the Built Environment (CABE [30]) found that whilst
increasing green space and tree cover in urban areas by 10% would reduce
surface water run-off by around 5%, adding green roofs to all buildings could
reduce it by 20%. So this, combined with water-butts and replacement of hard-
standing, could impact significantly on potential flooding.
3 Motivations and barriers

A first major block to BG-FRM comes when we look at current practice. Despite
around 5.5 million properties being at risk from flooding in England and Wales
([5, 6], DEFRA/EA [31], Wedawatta and Ingirige [32]), action to install
measures remains low; around 25% for households that have previously
experienced flooding and only 6% for those that have not (Thurston et al. [33],
Harries [34, 35]). A number of barriers face households, succinctly schematized
Figure 1: A stakeholder map for BG-FRM in a major UK city.
by Proverbs and Lamond [36] as Desire (awareness, perception, ownership) and

Ability (knowledge, finance, belief). They draw up a five-point classification of
financial and emotional constraints, informational barriers, aesthetic
considerations and timing issues. Each of these points would hold for BG-FRM;
finance could be overcome with subsidies, and belief, informational barriers and
emotional constraints could be addressed in part through a dialogic approach to
developing solutions.
The ‘stabilising’ factors for BG-FRM would include: hopes of reducing
flood-risk; improvements in access to ‘natural’ spaces; improved recreational
and leisure-use areas; improved biodiversity, air quality and reduced heat-island
effects; reduced housing and living costs (lower water-use and water-disposal,
improved insulation and opportunities for growing food), and improved house-
prices. Some of these points could be limited if hard-cover outside houses is
considered essential for parking and green-roofs are inaccessible. The
‘destabilising’ barriers discouraging take-up of BG-FRM could include:
awareness that such were an option; understanding what was possible and
appropriate for buildings; concern about installation and maintenance costs;
belief that they could work; fear of neighbourhood disapproval (for admitting
flood risk); safety concerns (individual, with children and ponds, and
community, with concerns of antisocial behaviour around green spaces, Bixler
and Floyd [23]); concerns about maintenance, impact on house-prices and
degentrification (as those who were able moved away), and contrastingly, that
improved aesthetics, recreational amenities and flood risk encourage
gentrification. For businesses, BG-FRM should open up opportunities; as

adoption grew, potential returns to capital investment would become more
apparent attracting new entrepreneurs, increasing local economic gain and
employment, embedding BG-FRM into the local economy and social fabric.
These initial lists of suggestions will need investigation with stakeholders, as
there will undoubtedly be motivations and barriers not considered here and the
degree to which each acts on considerations may differ hugely; to explore these
fully would require on-the-ground research in a case study city.
Changing behavioural patterns will be essential to ensuring longer-term BG-
FRM sustainability; we need to consider influences that make people more or
less likely to act. Motivations can be broadly divided into external and internal:
external (extrinsic) motivations will come from outside the agent, as fines or
subsidies, threats of litigation, or rewards, whilst internal (intrinsic) motivations
would depend on self-identity, needs, desires, aspirations and beliefs, the self-
satisfaction or self-worth derived from performing (or not) tasks (Organ [37]).
Organ provides a comprehensive literature review of motivation theories and
tailors these for looking at household energy efficiency refurbishments. From
this, they derive three motivation ‘themes’, economic, social and environmental,
that could broadly be translated to research into BG-FRM and behaviour change.
Economic motivations would include savings on water and heating bills from
green roofs, installation and maintenance costs as against incentives from local
and national government, household income and so spare capital (or lack of) for
undertaking works, and questions of how works might affect property value.
Social motivations include notions of comfort, people’s sense of the role of their
‘home’ (‘a platform for activities, social interactions, a haven, etc.’ [37]), social
norms, what is seen as ‘acceptable behavior’ and fashions and tastes of people’s
social groups, and people’s ‘locus of control’, the extent to which they feel able
to affect change through their own actions [37]. Environmental motivations will
encompass people’s sense of the positive effects stemming from a blue-green
approach and will be tied up with their sense of self (‘social’ motivation). If
people’s ideal self-image is environmentally conscious and responsible, they
may be more likely to adopt blue-green initiatives for the positive environmental
payback offered. Because of the nature of water-flows, installation of BG-FRM
in parts of a city not directly at flood risk may reduce risk faced in areas that are.
Occupiers of buildings in these areas may seemingly have no intrinsic motivation
to install devices, the act of so doing being thought of as solely due to external
motivations, altruism or improving self-image. However if such infrastructure
helps to avoid associated costs and inconveniences of flooding (roads and daily
life disrupted, Council Tax increases to cover damages, business costs with
disruptions to supply and sales-chains), then ‘externals’ could be internalised.
A significant demotivating factor could be the ‘why me?’ or ‘what difference
can I make?’ argument. Blue-Green approaches to flood action, like climate
change, will require large-scale collective action to be effective, and so if
individuals are asked to freely choose to engage of their own accord then we may
end up with a ‘tragedy of the commons’ (Hardin [38], cf. Lorenzoni et al. [39])
where there is no incentive to undertake the changes required if people suspect
their neighbours will free-ride on the benefits. Activities will need to be scaled
up to community level, which in turn necessitates dialogue working towards
consensus on action. Local authority or government regulation may be needed to
ensure all parties act appropriately, but given that all would stand to benefit from
BG-FRM over time then this may be avoidable through dialogue and co-
construction of viable preferred solutions.
4 Social Practice Theory (SPT)

To think through how different actors and groups may respond to BG-FRM, how
they might behave and how behaviour could change over time, we will need to
have an idea of: the ‘units’ under consideration; the de/stabilising influences
affecting behaviour, and a model for how these influences affect behaviour.
Traditionally, much social theory has been very broadly divisible into two camps
or approaches: Atomism (individualistic utilitarianism), wherein rational
individuals are the units of agency, acting to advance their own interests having
assessed costs and benefits (homo economicus); and Structuralism, wherein
‘human behaviour is an ‘effect’ of symbolic structures in the ‘unconscious’
mind’ (Reckwitz [40]), social norms, values and ‘rules’ that determine how
people behave. Under the former, relevant units of agency would be people
within households, businesses, Councils, Government, Environment Agency and
so forth; changing the behaviour of populations would mean simply shifting
perceptions of costs and benefits – providing free home flood surveys and
subsidising the cost of defence measures, for example, as advocated in DEFRA
documents [4]. However as DEFRA’s Resilience Grants Pilot Projects [41]
found, even with free provision of flood assessments, protection devices and
installation, only 83% of households took up the initiative, indicating other
factors were also at play. The structuralist tradition has also been criticised for
being over-determined and not allowing space for changes in practice, both
gradual and revolutionary.
Social theory has thus for some time been seeking to move beyond this
restrictive dualism of rationalistic atomism and deterministic structuralism. One
response has been a ‘family’ of ‘theories of social practice’, influenced by Pierre
Bourdieu, Anthony Giddens, later Michel Foucault, Judith Butler, Bruno Latour
and others. These Social Practice Theory (SPT) approaches adopt a both/and
positioning to overcome the agency/structure dualism; situating their analysis on
the (series of) practices of (groups of) individuals [40]. These sets of practices
are understood as forming ‘shared behavioural routines’ that are argued to be co-
constitutive of individuals. Individual actors retain their agency in a
contextualised fashion, continually re-producing practices and so contributing to
shifts and alterations at each turn, but the set of social practices is the unit of
analysis rather than the individual actor (Spaargaren [42]).
Social practices have been variously defined by those seeking to focus the
discussion. A widely cited quote we might reasonably use is provided by
Reckwitz [40]:
A practice is a routinized type of behaviour which consists of several

elements, interconnected to one another: forms of bodily activities, forms
of mental activities, ‘things’ and their use, a background knowledge in the
form of understanding, know-how, states of emotion and motivational
knowledge. A practice – a way of cooking, of consuming, of working, of
investigating, of taking care of oneself or of others, etc. – forms so to
speak a ‘block’ whose existence necessarily depends on the existence and
specific interconnectedness of these elements, and which cannot be
reduced to any one of these single elements.
So practice refers to an ensemble of factors ‘constitutive of particular
domains of social life’ (farming, business, voting, teaching, recreation, industry,
religion), a ‘set of considerations’ that shape how people act (Schatzki [43]). As
Schatzki elaborates, a practice ‘rules action not by specifying particular actions
to perform, but by offering matters to be taken account of … it qualifies the how
as opposed to the what of actions’. These notions of practice do not support
structuralist determinism, rather they serve quite the contrary. Practices and so
behaviour can change through the development of practices themselves: ‘[t]he
concept of practice inherently combines a capacity to account for both
reproduction and innovation … practices also contain the seeds of constant
change’ (Warde [44]). As practices are re-performed by different agents, certain
parties may hold to older variants, some perform currently dominant types and
others seek to replace conventions with new approaches. In this way, economic,
political and technological developments, cultural and historical influences and
other practices can all affect the development of a practice (Shove et al. [45]).
These new approaches are finding use within environmental-social sciences,
exploring spaces for changes in practice to enable the greening of consumption
and resource-use (Shove and Pantzar [46], Shove [47], Spaargaren and Mol
[48]). Environmental groups long been preoccupied with awareness-raising
around carbon footprint, but research indicates awareness is ‘a weak predictor’ of
actual behaviour, meaning we could usefully look to social practices for
possibilities of change (Spaargaren [42]). Parallels can be drawn with broader
social practice changes that communities will need to display for BG-FRM to be
sustainable over the longer-term. In Shove’s analyses of the freezer’s place in
modern society, for example, she ‘concentrates on the construction and
transformation of collective convention’ (Shove and Southerton [26], Shove
[47]); changing narratives around the freezer’s purpose and function, the
development of the frozen food industry and microwave, perceived increasing
demands upon time and the positioning of the freezer as a solution to this. In this
way, we look beyond individual preferences and habits without entirely
discounting them, to ways in which reasonable, ‘practical’ and desirable options
are framed over time. Shove and Southerton [26] quote Hackett and Lutzenhiser
[49]: ‘[w]hat [objects] are good for is a consequence, not a determinant, of their
use … they have consummatory as well as instrumental meaning’. We can say
the same of many BG-FRM options.
Use of SPT will be shaped by the nature of the investigation. Research would
look to future behaviour from stated current behaviour, preferences and
intentions, rather than being a reinterpretation of historical data as with Shove.

Examples of behaviour change over time could be explored from BG-FRM
installed some years previously, although of course the socio-economic and
cultural context in each case would differ from contemporary proposals. SPT
would be deployed as a heuristic device, ‘a sensitizing ‘framework’ for empirical
research’ rather than an overbearing theoretical structure determining what
should be seen and how it should be interpreted (Reckwitz [40]). Using SPT to
inform a conceptual framework would allow researchers to explore how different
practices could make the sustainability of BG-FRM more or less difficult and to
consider the conditions allowing for shifts in practices to increase sustainable
behaviour (and where responsibility for those conditions lie). The research would
look beyond individual respondents’ stated behaviour to consider ‘the many
institutions involved in structuring possible courses of action’ in the hope of
‘making some very much more likely than others’ (Shove [47]).
5 Conclusion
This paper has outlined a conceptual framework to guide how social research
into practices (and changes in practices) regarding city-level adaptations to
increase BG-FRM could be undertaken. Principal stakeholders have been
provisionally identified and lines of influence between these for BG-FRM
approaches suggested. Introducing BG-FRM to a city will be a complex matter
affecting many different stakeholders, and this is but one reason why the process
will need to be as inclusive as possible from the very beginning, to ensure that all
relevant and concerned voices are being listened to.
The stabilising and destabilising factors and motivations affecting people’s
behaviour have also been outlined and considered. While some research has
looked at a time-slice of attitudes, none has as yet tried to think around why and
how behaviour patterns might change as infrastructure developed and became
more normalised. This is an essential next step in thinking about the viability of
BG-FRM over time, understanding more about people’s current thinking and
how this could develop, to develop a clearer picture of likely outcomes. The
Social Practice lens has identified new avenues for interrogating this behaviour.
Studying this further using an SPT approach will require close work with
communities and other stakeholders at the start and throughout the proposed
changes, to understand how multiple considerations might settle or shift
regarding FRM options under a variety of hypothetical situations over time.
Acknowledgements
This research was performed as part of an interdisciplinary project programme
undertaken by the Blue-Green Cities Research Consortium
(www.bluegreencities.ac.uk). The Consortium is funded by the UK Engineering
and Physical Sciences Research Council under grant EP/K013661/1, with
additional contributions from the Environment Agency, Rivers Agency
(Northern Ireland) and the National Science Foundation.
References
[1] King, D.A., Climate change science: adapt, mitigate, or ignore? Science,
303(5655), pp. 176–177, 2004.
[2] Whitmarsh, L., Are flood victims more concerned about climate change
than other people? The role of direct experience in risk perception and
behavioural response. Journal of Risk Research, 11(3), pp. 351–374, 2008.
[3] DEFRA, Making Space for Water. Defra: London, 2005.
[4] DEFRA, Consultation on policy options for promoting property-level
flood protection and resilience, Defra: London, 2008a.
[5] Environment Agency, Flooding in England: A National Assessment of
Flood Risk, Environment Agency: Bristol, 2009.
[6] Environment Agency, Flooding in Wales: A National Assessment of Flood
Risk, Environment Agency: Bristol, 2009.
[7] Scottish Environment Protection Agency, Flooding in Scotland: A
Consultation on Potentially Vulnerable Areas and Local Plan Districts.
SEPA: Stirling, 2009.
[8] Pitt, M., Learning Lessons from the 2007 Floods: An Independent Review,
Cabinet Office: London, 2008.
[9] Bowker, P., Flood Resistance and Resilience Solutions: An R&D Scoping
Study, Defra: London, 2007.
[10] McBain, W., Wilkes, D. & Retter, M., Flood Resilience and Resistance for
Critical Infrastructure, CIRIA: London, 2010.
[11] Brown, R., Keath, N. & Wong, T., Transitioning to water sensitive cities:
historical, current and future transition states. 11th International
Conference on Urban Drainage, 10, 2008.
[12] Howe, C. & Mitchell, C., Water Sensitive Cities, IWA Publishing:
London, 2012.
[13] Kazmierczak, A. & Carter, J., Adaptation to climate change using green
and blue infrastructure. A database of case studies. University of
Manchester: Manchester, 2010.
[14] Blue-Green Dream website, http://bgd.org.uk, accessed on 10th September
2013, Imperial College London: London, 2013.
[15] Abbot, J., Davies, P., Simkins, P., Morgan, C., Levin, D. & Robinson, P.,
Creating Water Sensitive Places – scoping the potential for Water
Sensitive Urban Design in the UK, 2013.
[16] Bastien, N., Arthur, S. & McLoughlin, M.J., Valuing amenity: public
perceptions of sustainable drainage systems ponds. Water and
Environment Journal, 26(1), pp. 19–29, 2011.
[17] Wright, G.B. Arthur, S., Bowles, G., Bastien, N. & Unwin, D., Urban
creep in Scotland: stakeholder perceptions, quantification and cost
implications of permeable solutions. Water & Environment Journal, 25(4),
pp. 513–521, 2011.
[18] Kenyon, W., Evaluating flood risk management options in Scotland: A
participant-led multi-criteria approach. Ecological Economics, 64(1),
pp. 70–81, 2007.
[19] Apostolaki, S. & Jefferies, C., Social impacts of stormwater management

techniques including river management and SUDS. Final report, SUDS01.
Environment Agency: Bristol, 2005.
[20] Werritty, A., Sustainable flood management: oxymoron or new paradigm?
Area, 38(1), pp. 16–23, 2006.
[21] Johnson, C.L. & Priest, S.J., Flood Risk Management in England: A
Changing Landscape of Risk Responsibility? International Journal of
Water Resources Development, 24(4), pp. 513–525, 2008.
[22] McKissock, G., D’Arcy, B.J., Wild, T.C., Usman, F. & Wright, P.W., An
Evaluation of SuDS Guidance in Scotland, Diffuse Pollution Conference,
Dublin, 2003.
[23] Bixler, R.D. & Floyd, M.F., Nature is Scary, Disgusting, Uncomfortable.
Environment & Behaviour, 29(4), pp. 443–467, 1997.
[24] Chappells, H., Medd, W. & Shove, E., Disruption and change: drought and
the inconspicuous dynamics of garden lives. Social & Cultural
Geography, 12(7), pp. 701–715, 2011.
[25] Whitmarsh, L. Behavioural Responses to Climate Change: Asymmetry of
intentions and impacts. Journal of Environmental Psychology, 29, pp. 13–
23, 2009.
[26] Shove, E. & Southerton, D., Defrosting the Freezer: From Novelty to
Convenience. Journal of Material Culture, 5(3), pp. 301–319, 2000.
[27] Natural England, Green Infrastructure Guidance, Natural England:
Sheffield, 2009.
[28] RTPI, Briefing on Green Infrastructure in the United Kingdom, RTPI:
London, 2013.
[29] Scottish Government, Green Infrastructure: Design and placemaking, The
Scottish Government: Edinburgh, 2011.
[30] CABE, Using Green Infrastructure to Alleviate Flood Risk, Commission
for Architecture & the Built Environment: London, 2011.
[31] DEFRA/Environment Agency, The Appraisal of Human-related
Intangible Impacts of Flooding, DEFRA: London, 2004.
[32] Wedawatta, G. & Ingirige, B., Resilience and adaptation of small and
medium-sized enterprises to flood risk. Disaster Prevention and
Management, 21(4), pp. 474–488, 2012.
[33] Thurston, N., Finlinson, B. & Breakspear, R., Developing the Evidence
Base for Flood Resistance and Resilience: Summary Report, Environment
Agency: Bristol, 2008.
[34] Harries, T., Household Flood Protection Grants: The householder
perspective. Defra and Environment Agency Flood and Coastal Risk
Management Conference, 2010.
[35] Harries, T., The anticipated emotional consequences of adaptive behavior
– impacts on the take-up of household flood-protection measures.
Environmental Planning A, 44(3), pp. 649–668, 2012.
[36] Proverbs, D. & Lamond, J., The Barriers to Resilient Reinstatement of
Flood Damaged Homes. 4th International i-Rec Conference, 2008.
[37] Organ, S., Motivations for energy efficiency refurbishment in owner-

occupied housing. Structural Survey, 31(2), pp. 101–120, 2013.
[38] Hardin, G.J., 1968. The Tragedy of the Commons. Science, 1968.
[39] Lorenzoni, I., Nicholson-Cole, S. & Whitmarsh, L., Barriers perceived to
engaging with climate change among the UK public and their policy
implications. Global Environmental Change, 17(3-4), pp. 445–459, 2007.
[40] Reckwitz, A., Toward a Theory of Social Practices: A Development in
Culturalist Theorizing. European Journal of Social Theory, 5(2), pp. 243–
263, 2002.
[41] DEFRA, Resilience Grants Pilot Projects. DEFRA: London, 2008.
[42] Spaargaren, G., Theories of practices. Global Environmental Change, 21,
pp. 813–822.
[43] Schatzki, T.R., Social Practices: A Wittgenstinian Approach to Human
Activity and the Social. Cambridge University Press: Cambridge, 1996.
[44] Warde, A., Consumption and Theories of Practice. Journal of Consumer
Culture, 5(2), pp. 131–153, 2005.
[45] Shove, E., Pantzar, M. & Watson, M., The Dynamics of Social Practice.
SAGE: London, 2012.
[46] Shove, E. & Pantzar, M., Consumers, Producers and Practices:
Understanding the invention and reinvention of Nordic walking. Journal
of Consumer Culture, 5(1), pp. 43–64, 2005.
[47] Shove, E., Social Theory and Climate Change. Theory Culture and
Society, 27(2-3), pp. 277–288, 2010.
[48] Spaargaren, G. & Mol, A. Greening Global Consumption. Global
Environmental Change, 18(3), pp. 350–359, 2008.
[49] Hackett, B. & Lutzenhiser, L., The Unity of Self and Object. Western
Folklore, 44(4), pp. 317–324, 1985.
Delivering and evaluating the multiple flood

risk benefits in Blue-Green Cities:
an interdisciplinary approach
E. Lawson1, C. Thorne1, S. Ahilan2, D. Allen3, S. Arthur3,
G. Everett4, R. Fenner5, V. Glenis6, D. Guan7, L. Hoang5,
C. Kilsby6, J. Lamond4, J. Mant8, S. Maskrey1, N. Mount1,
A. Sleigh2, L. Smith9, 10 & N. Wright2
1
School of Geography, University of Nottingham, UK
2
School of Civil Engineering, University of Leeds, UK
3
School of the Built Environment, Heriot-Watt University, UK
4
Construction and Property Research Centre,
5
Department of Engineering, Cambridge University, UK
6
School of Civil Engineering and Geosciences, Newcastle University, UK
7
School of Earth and Environment, University of Leeds, UK
8
The River Restoration Centre, Cranfield University, UK
9
Centre for the Analysis of Time Series,
London School of Economics & Political Science, UK
10
Pembroke College, University of Oxford, UK
Abstract
A Blue-Green City aims to recreate a naturally-oriented water cycle while
contributing to the amenity of the city by bringing water management and green
infrastructure together. The Blue-Green approach is more than a stormwater
management strategy aimed at improving water quality and providing flood risk
benefits. It can also provide important ecosystem services and socio-cultural
benefits when the urban system is in a non-flood condition. However, quantitative
evaluation of benefits and the appraisal of the relative significance of each benefit
in a given location are not well understood. The Blue-Green Cities Research
Project aims to develop procedures for the robust evaluation of the multiple
doi:10.2495/FRIAR140101
functionalities of Blue-Green Infrastructure (BGI) components within flood risk

management (FRM) strategies. The salient environmental challenge of FRM cuts
across disciplinary boundaries, hence an interdisciplinary approach aims to avoid
partial framing of the ongoing FRM debate. The Consortium, comprising
academics from eight UK institutions and numerous disciplines, will investigate
linkages between human behaviours and physical processes, and produce an urban
flood model to simulate the movement of water and sediment through Blue-Green
features. Individual and institutional agents will be incorporated into the model to
illustrate how their behavioural changes impact on flooding and vice versa. A
methodological approach for evaluating the interaction of urban FRM components
within the wider urban system will be developed and highlight where, when and
to whom a range of benefits may accrue from BGI and other flood management
interventions under non-flood and flood conditions. Recognition of the compound
uncertainties involved in achieving multiple benefits at scale will be part of the
ongoing robust method of uncertainty evaluation. The deliverables will be applied
to a chosen demonstration case study, Newcastle, UK, in the final year of the
project (2015). This paper will introduce the Blue-Green Cities Research Project
and the novel, interdisciplinary framework that is adopted to investigate multiple
FRM benefits.
Keywords: Blue-Green Cities, flood risk management, multiple benefits,
interdisciplinary, green infrastructure, ecosystem services, pluvial flooding, urban
planning, and agent-based modelling.
1 Introduction
The combined impacts on social, economic and environmental systems make
flooding one of the World’s most serious hazards. Over 2.4 million properties in
England alone are at risk of fluvial or coastal flooding, with a further 2.8 million
properties susceptible to surface water flooding [1]. Increasing frequency and
magnitude of intense precipitation events in future decades are predicted to
increase flooding and damages incurred [2], particularly in cities where the
consequences of flooding are especially severe. Increasing urbanisation, economic
growth, and the concomitant increase in impermeable surfaces will further
exacerbate the urban flood risk. There is thus a demand for new and innovative
research that can help reduce the probability and/or consequences of urban
flooding while helping cities become more resilient and able to adapt to new flood
risks imposed by climate change [3] and economic development.
Non-traditional measures for flood risk management (FRM) aim to reduce the
amount of water entering man-made drainage systems and offer an alternative to
traditional grey infrastructure (e.g. piped drainage and waste water treatment
systems for pollution control). Natural measures are gaining increasing support as
efforts are made to better integrate the water cycle with urban design and
development needs, particularly in light of future climate change and the limited
adaptability of grey infrastructure to events that exceed the design standard. A
move towards urban water management that holistically considers the
environmental, social and economic consequences of different strategies is
illustrated by efforts to adopt water-sensitive urban design (WSUD) and

incorporate this in UK policy [4]. WSUD regards urban surface water runoff as a
resource, rather than a nuisance, diverging from the traditional paradigm of
removing surface water quickly and efficiently to advocating the protection of
urban water resources and generation of multiple benefits from multifunctional
landuse [5]. Such benefits may be achieved at lower costs if water services are
linked with other urban infrastructure systems [6]. WSUD and investment in green
infrastructure in the UK is in its infancy yet advances in Australia [7], other
European countries (including Scotland) [8], and the US [9], provide illustrative
examples of successful incorporation. However, the pace of transition to
connected and adaptive practices in urban water management, which integrate
FRM with new forms of sustainable and socially equitable urban planning and
design, must increase. Research projects, such as the ‘Blue Green Dream’ [10],
are helping advance the paradigm shift away from grey infrastructure yet
widespread implementation requires negotiation of the “Blue-Green” vision by all
representative stakeholders, and subsequent ownership of that vision.
The integration of urban design with various disciplines of engineering and
environmental sciences defines the WSUD process [5] and illustrates the
importance of utilising expertise from multiple disciplines for effective research,
planning and application. Holistic, interdisciplinary approaches are increasingly
endorsed as the most effective way to provide sound science and tackle the
environmental and societal problem of flooding while avoiding partial framing of
the FRM debate [11]. This paper introduces the Blue-Green Cities Research
Project and the novel interdisciplinary framework that places people, society and
their interactions with FRM policy at the heart of the research. Blue-Green Cities
Research is founded on strong internal and external communication networks and
will develop procedures for the robust evaluation of the multiple functionalities of
Blue-Green Infrastructure (BGI) components within FRM strategies. We aim to
generate novel findings on the behaviour and attitudes of individuals and
institutions to changes in the management practices of the urban water system, and
will subsequently apply this in a demonstration case study.
2 The Blue-Green Cities concept

A Blue-Green City aims to recreate a naturally oriented water cycle while
contributing to the amenity of the city by bringing water management and green
infrastructure together [12]. This is achieved by combining and protecting the
hydrological and ecological values of the urban landscape while providing
resilient and adaptive measures to deal with flood events (Fig 1). Key functions
include restoring natural drainage channels, mimicking pre-development
hydrology and improving water quality, reducing imperviousness, and increasing
infiltration, surface storage and the use of water retentive plants [13].
Blue infrastructure includes the ponds, flowing waterways, wet detention
basins and wetlands that exist within the drainage network. Green infrastructure
refers to natural land and plant based ecological treatment systems and processes.
This comprises open spaces, parks, recreation grounds, woodlands, gardens, green
corridors, vegetated ephemeral waterways and planted drainage assets that

undergo a wet/dry cycle due to runoff flow, e.g. green roofs and street trees. BGI
provides a range of services that include; water supply, climate regulation,
pollution control and hazard regulation (blue services/goods), crops, food and
timber, wild species diversity, detoxification, cultural services (physical health,
aesthetics, spiritual), plus abilities to adapt to and mitigate climate change [10].
Such services, and hence the benefits that are directly attributed to them, are often
absent where traditional grey infrastructure is used to manage surface water and
flooding. The Blue-Green concept places value on the connection and interaction
of blue and green assets and proposes a network of interconnected BGI to convey,
treat and manage urban runoff and flooding, while maximising the accrual of
multiple benefits. However, the lack of space in highly urbanised catchments may
restrict the incorporation and retrofitting of BGI, and hence, grey infrastructure
also has a role in the Blue-Green concept, particularly for high magnitude events
with a low probability of occurrence.
Figure 1: Comparison of the hydrologic (water cycle) and environmental

(streetscape) attributes in conventional (upper) and Blue-Green
Cities.
2.1 Multiple benefits of the Blue-Green Infrastructure
Blue-Green Cities may generate a multitude of environmental, ecological, socio-

cultural and economic benefits when the urban system is in both flood and non-
flood states. BGI that perform to the design standard will fulfil the primary goal
of reducing the risk of surface water inundation during a flood event. In addition,
when in the flood state, BGI may reduce water pollution and improve water
quality, help control the water supply and prevent the cascade of negative socio-
economic impacts that generally occur in the aftermath of a flood, e.g. high repair
costs, displacement from homes, damage to health, decline in business and
reduced economic prosperity. Furthermore, construction and maintenance of BGI
is often cheaper than the grey alternative, as illustrated by Portland’s “Green
streets” project to reduce stormwater runoff and the risk of combined sewer
overflow. $250 million in hard infrastructure costs was saved through the design
and landscaping of soil and plants into the urban streetscape to aid infiltration and
reduce peak stormwater flow (at a cost of $8 million) [14].
Blue-Green Cities also offer numerous benefits when the system is in a non-
flood state. Environmental benefits include; reduction in the urban heat island
effect, improved air quality, noise reduction, carbon sequestration and a carbon
emission reduction potential through avoiding highly carbon intensive
alternatives, groundwater recharge, increased biodiversity, habitat enhancement
and related ecosystem services. Socio-cultural benefits include; traffic calming
and road safety, reduction in water demand and water recycling, improved health
and wellbeing, attractive landscape, improved quality of place, crime reduction
and education potential. BGI may also augment the ability of cities to mitigate and
adapt to climate change [14] and is frequently a key component of economic
regeneration projects to improve the liveability of urban environments [15].
3 Interdisciplinary research and the Blue-Green

Cities Project
The potential benefits of the Blue-Green approach span the environmental, socio-
economic and cultural spheres of the urban environment, and hence, require an
interdisciplinary team to fully evaluate. Similarly, issues of FRM do not fit neatly
into a disciplinary boundary and an interdisciplinary approach is particularly
suitable. Interdisciplinary research may also be more responsive to public needs
and concerns and a valid means of generating science policy [11].
‘Interdisciplinarity’ is a highly debated term yet most definitions refer to the
integration of disciplines within a research environment driven by interactions and
joint-working amongst academics motivated by a common problem-solving
purpose [11, 16]. The field of a single discipline is therefore transgressed by
collaboratory working [17]. Similarly, an interdisciplinary approach can help
develop FRM policies that address the issue of future climate change and
resiliency; changes cannot solely be made through technological capabilities but
must also address variability in social expectations and lifestyles [18].
‘Blue-Green Cities’ is a highly interdisciplinary project funded by the
Engineering and Physical Sciences Research Council (EPSRC, February 2013–
January 2016). The Research Consortium comprises academics from eight UK
institutions and numerous disciplines; hydrodynamics, geomorphology, ecology,
physics, social sciences, engineering, and environmental economics. The main
research components (Fig 2) are denoted by Work Packages (WP), held together
by a strong communications package to promote interdisciplinarity and coherent,
integrated results, based on shared conceptual, methodological and theoretical
ideas [19]. A strong communications network, both internally and with external
stakeholders, is central to our goal to investigate the linkages between human
behaviours, physical processes and policy constraints regarding FRM. We aim to
progress from the multi-disciplinary approach where discrete disciplinary work
packages are completed and subsequently combined at the end of the project, with
little cross-discipline engagement during the research process. Rather, we aim for
data exchanges and common epistemological approaches to marry the
interdisciplinary appeal with the disciplinary mastery [20]. This will create
knowledge that is solution oriented and socially robust [21], and transferable to
both scientific and societal practice. Co-evolution of understanding and
knowledge, aided by tight integration within the team, will ensure that the sum of
the whole (in terms of deliverables) exceeds the sum of the parts.
The aim of the Consortium is to develop new urban FRM strategies as part of
wider, integrated planning intended to achieve urban renewal and environmental
enhancement in which multiple benefits of BGI are rigorously evaluated and
understood. Focussing on a common case study (Newcastle) in the third year of
the project (2015) is key to visualising the Consortium aim and converging on
common deliverables, with success relying on the co-production of knowledge and
multi-way exchange within the Research Consortium and wider stakeholders.
Communication is often ineffective and one-way between academia and end-
users, e.g. key stakeholders (including decision makers) and local communities
(those at risk of flooding and directly affected by decisions and hence should take
an active role in decision making regarding FRM [18, 22]). We aim to facilitate
discussion and include these groups from the outset.
3.1 Key deliverables
Research will focus primarily on fluvial and pluvial flooding; the latter typically
caused by extreme local storms and insufficient capacity of subsurface drainage
networks. The Consortium is developing urban flood models that realistically
represent the urban environment (land use and terrain) in its complexity. Coupled
surface/sub-surface hydrodynamic models will produce inundation predictions
across a range of events of different frequencies and lengths, visualised in
probability maps for inundation across an urban area. Flood inundation modelling
is being developed to include the movement of water through Blue-Green features
such as blue and green roofs, retention ponds, permeable paving, green space and
bioswales, to enable a comparison of flow velocity, depth and inundation extent
before and after the adoption of BGI. BGI as a FRM strategy will be assessed by
a set of scenarios including ‘business and usual’ (no additional BGI) and a Blue-
Green future (BGI as preferred assets).
Figure 2: Structure of the Blue-Green Cities Research Project.
Modelling existing flood risks is being linked to semi-quantitative assessments

of sediment and debris dynamics in emerging vegetated and naturalized urban
drainage systems. Fieldwork will fill knowledge gaps in network forms and
functions as part of a source-pathway-receptor analysis. Research is addressing
the movement of sediment and debris from catchment surfaces into and through
BGI, and assessing the potential for debris to block culvert trash screens. This will
develop the understanding of how sediment and debris sources and transportation
dynamics may impact on urban flooding. Sediment mass and volume, total
suspended solids, particle density, organic matter content and tracer techniques,
e.g. rare earth oxides and passive integrated transponder technology, are used to
analyse the performance of drainage networks. Sediment and debris dynamics,
such as entrainment, deposition, re-suspension and blockage potential at choke
and pinch points, are being identified to illustrate the efficiency of the multi-
element urban drainage network to detain or convey sediment and pollutants from
the source (urban surfaces) to receptor (receiving water body). The project will
also complete an impact assessment of Blue-Green vs. grey design on habitats and
biodiversity in open watercourses to advance the understanding of how
morphological and ecological diversity in urban streams may be increased and
ecosystem services accrued.
3.2 Determining agent responses to FRM and BGI
Successful simulation of the movement of water and sediment through the urban
environment will indicate design benefits of select infrastructure components and
generate recommendations to achieve multiple benefits. However, the physical
system cannot be assessed in isolation. Societal perceptions of the costs and
benefits of different FRM approaches play an important role in progressing
research into policy [22]. Interaction and involvement in the evolution of Blue-
Green design by the stakeholder community is essential to the concept of Blue-
Green Cities. Individual and institutional agents will be incorporated into the flood
inundation model to illustrate how behavioural changes impact on flooding and
vice versa. Such knowledge is crucial when making the case that agents need to
be part of the decision-making process for FRM. Fieldwork will be used to identify
and understand the behavioural responses of individuals and institutions to a range
of FRM strategies including Blue-Green. Evidence-based rules are being
developed using stated preference models to represent those behaviours and will
provide the data input to an agent-based model to investigate alternative scenarios
of future Blue-Green FRM strategies under different socio-economic conditions.
We are developing an understanding of how agents respond to stimulus and
change in the physical landscape, and how this may alter the probability of
flooding. We are also interested in how agents behave in a way to reduce the
consequences of flooding. Potential barriers to the implementation of FRM
strategies arise depending on where and to whom the benefits of BGI accrue during
times of no flood. This, and the potential for positive and negative interactions
with wider urban infrastructure, may act as an incentive/disincentive for the
adoption of innovative, non-traditional solutions.
3.3 FRM components, interfaces and uncertainties
Tools and methodologies are being developed to represent FRM and Blue-Green
networks in a single urban environment, as part of a wider complex ‘system of
systems’ that services urban communities. Series of interrelationships link energy,
transportation, water (supply and wastewater), emergency services, and
information and telecommunication sectors. Disrupting these dependencies can
have significant socio-cultural and economic consequences that may extend to
regional and national level, particularly during times of extreme flood. Research
will illustrate how changes in both the physical interfaces (flood pathways and
BGI) and institutional responsibilities (policy, planning and governance
structures) cascade across the wider urban system, and identify intervention points
to ensure rapid adoption, optimum functionality and reduced risk in other
infrastructure areas. The Three Points Approach (3PA) of Fratini et al. [23] will
be adopted and illustrates a more holistic process towards urban FRM that
simultaneously considers technical optimisation of urban drainage systems, spatial
planning to increase resiliency, and everyday performance under the green, non-
flood, condition as a foundation for social preparedness. Three system states have
been developed from the 3PA; non-flood (green condition), design standard, and
extreme event (blue condition). By understanding the interactions between

different urban infrastructure components under each of the three system states
can we hope to highlight where, when and to whom the costs and benefits of
different FRM strategies accrue.
Acceptable functioning of the flooding system is determined by meeting the
standard for flood defence despite the occurrence of possible climate changes.
Hence, we are also investigating how to optimise the functioning of the urban
water system to cope with an uncertain future, addressing recent theory that non-
traditional, Blue-Green measures may create a more resilient flooding system with
respect to long-term future change [14, 15]. Due to the non-stationarity of physical
processes, a range of scenarios will be employed to investigate the success of BGI
under different possible futures, acknowledging the full range of uncertainty that
is inherent to the outcome. This links to an ongoing uncertainty analysis which
aims to identify, and where possible, quantify uncertainty as it propagates through
the model cascade. Uncertainty is inherent in all models (empirical, conceptual,
and numerical) and effective buy-in from stakeholders regarding
recommendations for urban FRM is dependent on transparency in the research
process and acknowledgement of assumptions made. We are addressing
uncertainties that we are able to reduce, uncertainties that we can track and
propagate, and those we can only talk about. The evolving character of built
environments combined with large uncertainty in future flood inundation, for
instance, increases the complexity of modelling urban FRM strategies. Despite
such limitations, we hope to identify strategies that are robust to some of the future
uncertainties, help increase resilience, and generate a range of benefits.
3.4 Evaluation and synthesis of multiple benefits
Methodologies are being developed to assess, quantify and value the multiple
benefits of adopting BGI in urban FRM strategies at both the local/regional and
global/international scales. Such methodologies will also robustly evaluate the
multiple functionalities of BGI components and address the inherent uncertainties
of cost/benefit analysis. By evaluating the relative significance of benefits in
context specific locations we aim to establish preference ratings linked to a multi
criteria analysis for component selection. This will provide sound science and
recommendations for design guidance to assist policy makers in the choice of
FRM strategy. Despite the 2007 SuDS (Sustainable urban Drainage Systems)
Manual (C697) [24] providing extensive guidance, the lack of recent UK
legislation is a key barrier to the limited uptake of BGI and SuDS.
We adopt a novel method of performance appraisal against a set of diverse
criteria that addresses environmental, socio-cultural and economic costs and
benefits that accrue beyond the realm of effective FRM. Surface water
management objectives, such as the minimization of runoff quantity, reduction of
peak stormwater flows, and improvement to runoff quality may be achieved by
grey or Blue-Green infrastructure. Both incur costs; capital materials, energy
inputs and maintenance, yet those for BGI are typically much lower [14]. Life
cycle assessment (LCA) and similar methods of economic costing are often used
for comparison and selection of asset design [25]. Whatmore et al. [22] contend
that choice of FRM solution based solely on economic viability (benefits > costs)
restricts the range of FRM solutions to be explored. The full net-benefit of BGI
development can only be realized by a comprehensive accounting of their multiple
benefits [14]. Quantitative evaluation of benefits and the appraisal of the relative
significance of each benefit in a given location are not well understood. BGI is
acknowledged as providing additional benefits that grey infrastructure cannot,
such as counteracting urban heat island effects, reducing energy costs, creating
community amenities and improving habitats [14], and multi-functional landuse
is paramount to optimise BGI benefit accrual.
3.5 Application in the demonstration case study (Newcastle, UK)
The deliverables from Blue-Green Cities research will be exhibited in the

demonstration case study, Newcastle, UK, in the final year of the project (2015)
to demonstrate the applicability of the methods, measures and evaluations
developed by the Consortium. Newcastle encompasses hydrological, topographic,
urban density and socio-economic conditions that are representative of those found
more widely in UK cities and has experienced recent major flooding events. Much
of the city centre is impermeable and vulnerable to pluvial flooding, piped
drainage systems are often unable to cope with intense rainfall and the risk of
sewer incapacity and surcharge is relatively high. The need for increased housing
provision may also reduce available greenspace in the future. Interest in BGI for
FRM from key stakeholder groups plus active research into climate change
adaptation and mitigation and urban greenspace [15] suggests Newcastle may be
highly receptive to the Blue-Green concept.
4 Summary
The Blue-Green Cities Research Project adopts an interdisciplinary approach to
identify and rigorously evaluate the multiple benefits of natural flood risk
management strategies using Blue-Green infrastructure. This paradigm shift from
traditional grey infrastructure designed to remove water as quickly as possible
from the urban surface is in line with WSUD and urban water management that
holistically considers the environmental, social and economic consequences FRM
strategies. A Blue-Green City offers effective performance of the drainage
network to achieve high levels of flood protection and resilience to some future
climate change, while supporting multiple non-flood benefits, often maximised by
the integration of blue and green assays and creation of networks. Throughout
2014–15 the Blue-Green Cities Research Consortium will model how changes in
policy and associated agent behaviour/attitudes can impact on flooding and vice
versa. This linking of physical processes to human behavioural patterns for
different scenarios is highly innovative and will provide for an analysis of the
urban ‘system of systems’ and highlight where, when and to whom the multiple
benefits will accrue under different future scenarios. This will allow us to
rigorously, and where possible, quantitatively, evaluate the costs and benefits of
different strategies and appraise the relative significance of each benefit in a given
location. The attitudes and perceptions of people and society towards Blue-Green
and grey infrastructure is critical in demonstrating to policy makers how non-
traditional infrastructure may be utilised to achieve maximum benefit while
ensuing ‘agents’ become part of the decision-making process. The applicability of
the research methods will be tested in the demonstration case study (Newcastle,
UK) and will endeavour to incorporate the understanding and interest of key
stakeholders in urban FRM and connect this with the potential impact of adopting
the Blue-Green vision in a practical, real-life setting.
References
[1] Bennett, O., Flood defence spending in England, Standard Note: 14th
March 2013, House of Commons Library, London, 2013.
[2] Bates, B., Kundzewicz, Z.W., Wu, S., & Palutihof, J., (eds). Climate
Change and Water, Technical Paper, Intergovernmental Panel on Climate
Change (IPCC), 2008.
[3] Wilby, R.L. & Keenan, R., Adapting to flood risk under climate change.
Progress in Physical Geography, 36(3), pp. 348–378, 2012.
[4] Ashley, R., Lain, L., Ward, S., Shaffer, P., Walker, L., Morgan, C., Saul,
A., Wong, T., Moore, S., Water-sensitive urban design: opportunities for
the UK. Proceedings of the ICE-Municipal Engineer, 166(2), pp. 65–76,
2013.
[5] Wong, T., & Brown, R., The water sensitive city: principles for practice.
Water Science & Technology, 60(3), pp. 673–682, 2009.
[6] Potter, K., Ward, S., Shaw, D., Macdonald, N., White, I., Fisher, T., Butler,
D., & Kellagher, R., Engineers and planners: sustainable water management
alliances. Proceedings of the ICE-Engineering Sustainability, 164(4),
pp. 239–247, 2011.
[7] Brown, R.R., & Clarke, J. M., Transition to water sensitive urban design:
The story of Melbourne, Australia. Facility for Advancing Water
Biofiltration, Monash University Melbourne, Australia, 2007.
[8] Stahre, P., Blue-green fingerprints in the city of Malmö, Sweden: Malmö’s
way towards a sustainable urban drainage. VASYD, Malmö, Sweden, 2008.
[9] Portland “Grey to Green” initiative; The City of Portland Environmental
Services website, http://www.portlandoregon.gov/bes/47203
[10] Maksimović, S., Xi Liu, S., & M. Lalić, M., Blue Green Dream Project’s
Solutions for Urban Areas in the Future. Reporting for Sustainability,
pp. 49–54, 2013, available online at http://www.sciconfemc.rs
/PAPERS/BLUE%20GREEN%20.pdf
[11] Lowe, P. & Phillipson, J., Reflexive interdisciplinary research: the making
of a research programme on the rural economy and land use. Journal of
Agricultural Economics, 57(2), pp. 165–184, 2006.
[12] Hoyer, J., Dickhaut, W., Kronawitter, L., & Weber, B., Water sensitive
urban design: principles and inspiration for sustainable stormwater
management in the city of the future. Jovis Publications: Berlin, Germany,

2011.
[13] Novotny, V., Ahern, J., & Brown, P., Water centric sustainable
communities: planning, retrofitting and building the next urban
environment. Wiley Publishing.com, 2010.
[14] Foster, J., Lowe, A., & Winkelman, S., The value of green infrastructure
for urban climate adaptation. Center for Clean Air Policy, February, 2011,
Available online at http://dev.cakex.org/sites/default/files/Green_
Infrastructure_FINAL.pdf
[15] Planning for the Future, Core Strategy and Urban Core Plan for Gateshead
and Newcastle upon Tyne, NewcastleGateshead City Council, 2013.
Available online at https://www.newcastle.gov.uk/planning-and-
buildings/planning-policy/core-strategy-and-urban-core-plan
[16] Klein, J.T., A taxonomy of interdisciplinarity. The Oxford handbook of
interdisciplinarity, 15, 2010.
[17] Balsiger, P.W., Supradisciplinary research practices: history, objectives and
rationale. Futures, 36(4), pp. 407–421, 2004.
[18] Ashley, R., Blanksby, J., Chapman, J., & Zhou, J., Towards integrated
approaches to reduce flood risk in urban areas. Advances in Urban Flood
Management, pp. 415–432, 2007.
[19] OECD (Organisation for Economic Cooperation and Development),
Interdisciplinarity in Science and Technology, In: Directorate for Science,
Technology and Industry. OECD: Paris, 1998.
[20] Klein, J.T., Evaluation of interdisciplinary and transdisciplinary research: a
literature review. American journal of preventive medicine, 35(2), pp. 116–
123, 2008.
[21] Gibbons, M., Science’s new social contract with society. Nature, 402,
pp. 81–84, 1999.
[22] Whatmore, S.J., Odoni, N., Ward, N., & Bradley, S., Coproducing flood risk
knowledge: redistributing expertise in critical participatory modelling.
Environment and Planning A, 43, pp. 1617–1633, 2011.
[23] Fratini, C., Geldof, G. D., Kulck, J., & Mikkelsen, P. S., Three Points
Approach (3PA) for urban flood risk management: A tool to support climate
change adaptation through transdisciplinarity and multifunctionality. Urban
Water Journal, 9(5), pp. 317–331, 2012.
[24] Woods-Ballard, B., Kellagher, R., Martin, P., Jeffries, C., Bray, R., &
Shaffer, P., The SUDS manual. CIRIA, London, 2007.
[25] Wang, R., Eckelman, M. J., & Zimmerman, J. B., Consequential
Environmental and Economic Life Cycle Assessment of Green and Gray
Stormwater Infrastructures for Combined Sewer Systems. Environmental
Science & Technology, 47(19), pp. 11189–11198, 2013.
Modelling a green roof retrofit in the

Melbourne Central Business District
S. J. Wilkinson3, C. Rose1, V. Glenis2 & J. Lamond1
1
Faculty of Environment and Technology,
2
The Centre for Earth Systems Engineering Research (CESER),
School of Civil Engineering and Geosciences,
Newcastle University, UK
3
School of the Built Environment, University of Technology, Australia
Abstract
With the increasing densification in urban settlements the economic and social
disruption caused by pluvial flooding events globally is significant and growing.
Furthermore these problems are compounded where many cities are located in
areas where climate change predictions are for increased rainfall frequency
and/or intensity. One possible solution is the wide scale retrofit with green roof
technology as a means of mitigating stormwater runoff in urban settlements.
However, it is not known currently where the most effective location for and
siting of the retrofitted green roofs in a city or town would be. Moreover, the
number of and type of green roof required to reduce pluvial flooding is
unknown.
This paper describes a proof of concept framework for an assessment of the
potential to reduce pluvial flood hazard through the retrofit of green roofs
combining an evaluation of the retrofit potential of office buildings in the Central
Business District (CBD) with state-of-the-art urban rainfall inundation
modelling. Using retrofit scenarios for Melbourne CBD commercial buildings
built between 1998 and 2011 and the rainfall profile of the February 2011 event,
the modelled depths of flooding were compared. The results show that the
potential to mitigate extreme events via retrofit would be enhanced by
consideration of buildings within the wider catchment.
Keywords: flood, green roof, inundation modelling, retrofit, CBD.
doi:10.2495/FRIAR140111
1 Introduction
Globally weather patterns are changing [1], whilst it is not possible to attribute
specific extreme events to changing climates there is consensus that the
frequency of intense rainfall events is rising and will continue over most land
masses, including those where average rainfall is decreasing [1]. Intense rainfall
events can cause flash floods, particularly in dense urban areas with low
permeability.
For Australia, major flooding occurred over the densely populated East Coast
area for two consecutive years from 2010. The State Emergency Service
responded to over 100 requests for flood-related damage when a storm caused 29
mm of rain in half an hour at Perth Airport [2]. In March 2012 the Bureau of
Meteorology issued Flood Warnings and broad-scale Severe Weather Warnings
for heavy rain and flash flooding over much of northern and eastern Queensland.
The estimated costs of remediation of flood damaged buildings is A$20 billion
[3, 4].
The incidence of pluvial flooding is, in part, attributable to changing weather
patterns, and climate predictions for Australia include increased intense rainfall
in south-western and south-eastern Australia [5]. Though there has been a
general trend of declining autumn and winter rainfall in south-western and south-
eastern Australia, Australian average annual rainfall has increased slightly,
largely due to increases in spring and summer rainfall, most markedly in north
western Australia [5]. Added to this are development pressures and increasing
urban density which add to the growth in damage caused by these events [6].
Green roofs have been shown to be highly suitable stormwater controls for
retrofitting in dense urban areas [7]. As roofing areas can account for 40–50% of
the impermeable surfaces in urban locations, such modification offers the
potential to mitigate pluvial flooding without additional land-take being required
[8]. There are two main types of green roof: extensive (incorporating shallow
rooted species in a relatively thin substrate) or intensive (deep rooted species
found in roof gardens); as the load-bearing capacity of extant structures is a key
constraint, the lighter extensive type is generally more appropriate for retrofitting
applications. Mitigation of stormwater impacts occur via two processes: water is
absorbed by the growing medium, thereby delaying the onset of runoff and
attenuating peak flows; the stored water is then released by a combination of
evaporation and transpiration, mediated by the foliage [9]. These processes can,
therefore, relieve pressure on existing piped drainage systems; in extreme storm
events, however, capacity will be exceeded and the design must take this into
account [10]. Rose and Lamond’s meta-analysis [11] notes that reported
performance ranged from 42–90% of annual rainfall, whilst average retention
during storm events varied from 30–100%.
Detailed specific study is, however, needed in order to make the case for
retrofitting green roofs: firstly, the surface area and location of candidate roofs
within the urban space will have a great influence on the quantity and pattern of
runoff attenuated; secondly, a number of physical factors need to be considered
in determining retrofit suitability; thirdly, the meteorological conditions
(including the typical rainfall patterns) and hydrological factors (such as runoff
characteristics) both need to be examined. The geographical location of green
roofs has, unsurprisingly, an impact on their performance, owing to regional
climatic variation: vegetated roofs in a sub-tropical Mediterranean climate [for
example, 12] will perform differently from those in a temperate maritime climate
such as the UK [13]. Further key factors influencing performance have been
found to include roof characteristics such as overshadowing (which can inhibit
vegetation growth) and the degree of pitch of the roof [14]; other variables
include substrate type; species mix; vegetation height and local weather
characteristics (intensity of rainfall; antecedent moisture conditions) [11]. It is
also necessary to engage the support of the owners and occupiers of relevant
properties, if a retrofitting programme is to be successful; an aid to uptake can be
provided via incentive schemes such as those employed in New York City [15]
and Portland [16]. Therefore, in this paper, a framework for evaluation of runoff
attenuation through retrofit of an urban area is described, with particular
reference to a case study in Melbourne Australia.
2 Approach and data

The City of Melbourne, Australia, has adopted a policy of attaining carbon
neutrality by 2020 [17]; in this context, the aim of adapting 1200 existing
commercial buildings to incorporate sustainability was established [18]. A
proactive approach to meeting this target has been undertaken, including
research to identify suitable properties; it has been established that much of the
existing stock in the city is now at the stage where adaptation or retrofit is
typically undertaken [19]. In addition, the relevant water company has
recommended the adoption of water sensitive urban design for areas where the
drainage infrastructure cannot be upgraded [20].
Melbourne is situated on sloping terrain on the banks of the Yarra River, a
major watercourse; in common with many major cities, extensive expansion over
time has resulted in once-permeable agricultural land on the outskirts being
replaced by developments featuring largely impermeable surfaces. Furthermore,
tributary watercourses within the Melbourne Central Business District (CBD)
have been culverted: not only are culverts prone to collapse and blockages but
also, in extreme rainfall events, the volume of water can rapidly exceed the
carrying capacity, giving rise to overland flows or ‘flash floods’. This was
demonstrated in March 2010 when a severe storm resulted in 61 mm of rain
falling on the city in 48 hours, exceeding the average rainfall for the entire month
[21]. As a result of this pluvial event, a number of roads were rendered
impassable for several hours, a major railway station was flooded and tram
services in the area had to be suspended; the foregoing caused severe disruption
to the normal functioning of the CBD. The total damage for the storm across the
west of the state of Victoria exceeded a billion Australian dollars [22]: severe
weather events can, therefore, have serious adverse impacts on the economies of
affected areas, with businesses unable to function in the aftermath.
In the context of anticipated climate change, this region of Australia is

predicted to experience more intense storm events: sudden, large volumes of
runoff water are, therefore, likely to cause more frequent overland flow
responses. Measures to mitigate such events will, therefore, be needed in a major
city such as Melbourne.

Figure 1: Flash flooding in Flinders Street, Melbourne, 6 March 2010.

(Photo courtesy N. Carson.)

Figure 2: Melbourne CBD showing typical rooftops with Yarra River to
right hand side (land generally slopes left to right in this image).
2.1 Summary of method
The framework applied to the Melbourne CBD combines assessment of the

retrofit potential of the built environment with the application of an inundation
model under three retrofit scenarios.
There were three main stages within the framework:
1. Compilation and analysis of buildings database to evaluate the

commercial buildings suitable for retrofit in the Melbourne CBD.
2. Collection and compilation of a digital terrain data for Melbourne
and buildings data for the Melbourne CBD.
3. Inundation modelling of the Melbourne CBD area using the rainfall
pattern from the flood event of February 2011 under three retrofit
scenarios.

This was achieved through the development of a database of commercial
buildings within the Melbourne CBD from a variety of sources. The database
was compiled using multiple sources, including existing commercial databases
such as Cityscope in Australia, and publicly available databases such as PRISM
(Victorian Government) and the Heritage database. In addition, data from the
Property Council of Australia (PCA), Google Earth and Google Streetview [23]
was used to gather building related data. Finally, visual inspections and
photographs of CBD buildings were undertaken. Following a comprehensive
validation phase the final building database contained 526 commercial buildings
in the Melbourne CBD.
The potential for retrofit was evaluated by a qualified building surveyor
through visual inspection and using the property database based on the following
criteria: Position of the building; location of the building; orientation of the roof;
height above ground; roof pitch; weight limitations of the building.
The City of Melbourne Property Services provided contour data for the wider
Melbourne catchment was converted to a hydrologically corrected digital terrain
model using ARC GIS; shapefile data for buildings, roads and pavement for the
CBD were obtained from the same source. The building polygons layer was
then modified to create three scenarios: No green roof retrofit, 100% green roof
retrofit and green roof retrofit, for each of the commercial buildings assessed as
suitable in stage 1. Rainfall pattern data were purchased from Bureau of
Meteorology in Melbourne, and these data were combined within the CityCat
rainfall inundation model, as described below.
2.2 Description of inundation model
CityCat is an urban flood modelling system based on the shallow water

equations [24] capable of modelling pluvial and fluvial flooding in real urban
settings. CityCat uses a self-generated grid based on readily available LIDAR
Digital Terrain Models and GIS data to represent surface and building
characteristics. Areas occupied by buildings are excluded from the numerical
grid but buildings are incorporated into the numerical domain as objects and can
have different properties/characteristics such as storage on roofs. This is
radically superior to other models which represent buildings as raised ground.
Surface properties can be altered to predict influences of more permeable
surfaces (green-areas) or swales. Simulations of different flood events can be
driven by rainfall, flow/water depth time series.
The architecture of CityCat is based on the object-oriented approach which

offers development flexibility and allows easy extension of functionality due to
the fully modular structure. Also, the computational efficiency is improved by
removing the decisions (“If Then Else” statements) during run time. The solution
of the 2D shallow water equations is obtained using high resolution finite-
volume methods with shock-capturing schemes [25] which are able to capture
propagation of flood waves. CityCat can also be deployed on the Cloud for high
resolution large scale modelling of flooding that can be used for the assessment
of city-scale flood risk under climate change [26].
3 Results
3.1 Summary of retrofit potential
The map of Melbourne CBD (Figure 3) shows the historic centre area known as
the ‘Hoddle Grid’ on the north bank of the Yarra River, in which buildings were
assessed for suitability for retrofitting green roofs. Although the CBD is long
established the analysis revealed that the majority of the 526 commercial
buildings in the database (60%) were constructed after 1940 and over half were
constructed post 1960 representing a large amount of stock which is potentially
due for renewal and upgrade. However, the heights of the buildings are highly
variable, with the majority of low to medium rise buildings being mixed in with
the approximately 30% of high rise buildings and skyscrapers. Such an
arrangement of buildings could mean that existing buildings which have
adequate structural strength to accommodate retrofitting with green roofs may be
unsuitable because of overshadowing, which adversely affects planting.
However it is possible that consideration of other plant types with substrates
designed to be more absorbent than existing specifications might change this
finding. Furthermore, the orientation of many of the buildings in the CBD was
also seen to be unhelpful to successful plant growth.
Figure 3: The ‘Hoddle Grid’ area of Melbourne CBD with the modelled area
outlined in black (courtesy N. Bhattacharya-Mis).
Through visual inspection of the roof, using the Google Earth and Google
Map software, an evaluation of individual rooftops was carried out and each roof
classified as suitable, not suitable or indeterminate with respect to retrofit
potential. This evaluation was based on pitch, roof coverage by service plant and
roof construction. From this detailed study it appeared that only 15% of
Melbourne CBD buildings were considered suitable for retrofit; of the
remainder, 80% were seen not to be suitable and 5% were indeterminate. The
database covered only commercial office building stock, however, and many
other land use types exist in the CBD, including retail, residential and
educational. It is likely that some of the roofs of these buildings would also be
suited to retrofit, which would be expected to change the outcome. Finally, it
appears that the origin of the body of water which lead to the flooding in the
CBD was to the north of, and outside of, the Hoddle Grid and it is logical that
consideration of green roof retrofit potential in this stock is undertaken and
analysed.
3.2 Summary of inundation model
The model outputs are colour-coded to show water depths (as illustrated in
Figure 4). The results of the inundation model clearly showed that major flows
within the Melbourne CBD during intense storms originate from outside the
CBD grid: for example, the route of the Yarra tributary culverted below
Elizabeth Street appears here as the white line running on a diagonal from north-
west to south-east, entering the main river to the south of the railway station.
Figure 4: Water depth map – no green roofs (scenario 1).
3.3 Summary of retrofit model
A comparison of the modelled outputs reveal that the water depths could be
expected to decrease if green roofs were to be retrofitted in the area. In the
vicinity of Flinders Street Station, for example (Figure 5) the south-west corner
of the road grid has very dark shading (water over 1 metre depth) in scenario 1
(no green roofs), but this is replaced by a light grey (around 0.5 metre depth) in
scenario 3 (suitable roofs retrofitted). The granularity of the comparisons is, of
course, more apparent in the full colour images.
Figure 5: Detail of Flinders Street Station area without green roofs (left) and
with green roof retrofit (right).
3.4 Summary
The results suggest green roofs offer a potential for mitigation, in that flood
levels could be reduced in the affected areas, leading to concomitant reduction in
impacts; however, flooding of any depth presents issues for business continuity.
The key finding is that the historic flow pathways, normally hidden beneath the
urban district, will continue to re-establish their flow patterns during extreme
events, directing pluvial flows from the wider catchment towards the Yarra
River: in order to mitigate pluvial flooding in the CBD it will, therefore, be
necessary to take a wider catchment approach.
4 Discussion and conclusion

This paper has presented a framework for the evaluation of green roof retrofit in
order to mitigate the impact of extreme rainfall events in business districts. The
framework has been applied to an illustrative case study within Australia,
namely the city of Melbourne. Novel aspects of this framework include the
evaluation of retrofit potential by a qualified building surveyor using an
extensive buildings database and the incorporation of a state-of-the-art
inundation model that allows precise modelling of inundation within a complex
urban environment.
The analysis has demonstrated that such detailed analysis of specific features
of the urban environment is very important in achieving a realistic estimate of
peak attenuation and resultant flood risk reduction. Within the Melbourne CBD
about 15% of the commercial buildings were seen to be suitable for retrofitting
of green roof technology; many of these buildings, however, were concentrated
in the sub-prime areas of the business district, rather than the locations known to
have experienced the most severe flooding in 2011.
In addition, this inundation model showed that the incorporation of green
roofs on all Melbourne’s CBD buildings was not sufficient to prevent flooding in
extreme rainstorm events. However, some mitigation was observed, in that the
depth of flooding at key locations was reduced. When the expert building
knowledge was also applied, the potential to mitigate flooding by retrofit within
the CBD area was greatly reduced, reflecting the fact that high density, prime
real estate property is not particularly sympathetic to green roof technology. The
implication of this finding for the city of Melbourne is that the potential to
mitigate flooding within the CBD solely by using CBD buildings is limited,
although other benefits deriving from the use of green roofs and other green
infrastructure are well recognised.
The results of the inundation model clearly show that major flows within the
Melbourne CBD during intense storms originate from outside the CBD grid. The
historic channels that are hidden beneath the urban district re-establish their flow
patterns during extreme events, directing pluvial flows from the wider catchment
towards the Yarra River: in order to mitigate pluvial flooding in the CBD it will
be necessary to take a wider catchment approach.
The potential of the framework to aid in decision making is clearly
demonstrated through the case study and therefore it is recommended to widen
the scope if the framework to the wider Melbourne area in order to evaluate
where the greatest retrofit opportunities exist.
Acknowledgements
This research was partially funded by the RICS research trust (Project no 464)
Retrofit of Sustainable Urban Drainage (SUDS) in CBD for improved flood
mitigation
Part of the research was also performed as part of an interdisciplinary project
programme undertaken by the Blue-Green Cities Research Consortium
(www.bluegreencities.ac.uk). The Consortium is funded by the UK Engineering
and Physical Sciences Research Council under grant EP/K013661/1, with
additional contributions from the Environment Agency, Rivers Agency
(Northern Ireland) and the National Science Foundation.
Thanks are due to the following for data supplied to the project:
Contour and shapefile data: David Hassett, GIS Team Leader, Property
Services, GPO Box 1603, Melbourne, Victoria 3001 Australia
Bureau of Meteorology data: Dr Blair Trewin, National Climate Centre, GPO
Box 1289, Melbourne, Vic 3001, Australia
References
[1] Solomon, S. and D. Qin, Climate Change 2007: The Physical Science
Basis. Contribution of working Group 1 to the Fourth Assessment Report
of the Intergovernmental Panel on Climate Change. 2007, Cambridge
University Press.: Cambridge, UK and New York, USA.
[2] Bureau of Meteorology. Heavy rain and Flooding. 2012 9th July 2012;
Available from: http://www.bom.gov.au/wa/sevwx/perth/floods.shtml.
[3] Bloomberg (2012) Insurers count cost in Queensland as floods peak.
[4] Companies and Markets (2011) Australian Flood Damage Reconstruction
Likely to Cost Billions.
[5] Bureau of Meteorology. 2014; Available from: http://www.bom.gov.au

/climate/change/.
[6] Jha, A., et al., Five Feet High and Rising – Cities and Flooding in the 21st
Century, in Policy Research Working Paper 5648. 2011, The World Bank:
Washington.
[7] Voyde, E., E. Fassman, and R. Simcock, Hydrology of an extensive living
roof under sub-tropical climate conditions in Auckland, New Zealand.
Journal of Hydrology, 2010. 394(3–4): pp. 384–395.
[8] Stovin, V., The potential of green roofs to manage Urban Stormwater.
Water and Environment Journal, 2010. 24(3): pp. 192–199.
[9] Mentens, J., D. Raes, and M. Hermy, Green roofs as a tool for solving the
rainwater runoff problem in the urbanized 21st century? Landscape and
Urban Planning, 2006. 77(3): pp. 217–226.
[10] Balmforth, D., et al., Designing for exceedance in urban drainage – good
practice CIRIA C635 2006, CIRIA.
[11] Rose, C.B. and J. Lamond, Performance of sustainable drainage for urban
flood control, lessons from Europe and Asia, in International Conference
on Flood Resilience – Experiences in Asia and Europe, D. Butler and S.
Djordjevic, Editors. 2013: Exeter, United Kingdom.
[12] Fioretti, R., et al., Green roof energy and water related performance in the
Mediterranean climate. Building and Environment, 2010. 45(8): pp.
1890–1904.
[13] MacIvor, J.S. and J. Lundholm, Performance evaluation of native plants
suited to extensive green roof conditions in a maritime climate. Ecological
Engineering, 2011. 37(3): pp. 407–417.
[14] Wilkinson, S.J. and R. Reed, Green roof retrofit potential in the central
business district. Property Management, 2009. 27(5): pp. 284–301.
[15] Bloomberg, M.R. and C.H. Strickland, NYC Green Infrastructure – 2012
Annual Report. 2012, New York City – Envirionmental Protection: New
York City.
[16] Environmental Services – City of Portland. Portland’s Ecoroof Program
2011,
. cited 2013; Available from: http://www.portlandoregon.gov/bes
/article/261074.
[17] Arup Pty Ltd, Zero Net Emissions by 2020 – update 2008 (Consultation
Draft). 2008, City of Melbourne.
[18] Maunsell Australia Pty Ltd, Towards a City of Melbourne Climate Change
Adaptation Strategy: A Risk Assessment and Action Plan Discussion
Paper Responding with Resilience. 2008, City of Melbourne, Australia.
[19] Jones Lang LaSalle, Building Refurbishment – Repositioning your asset
for success. 2005, Jones Lang LaSalle: Melbourne, Australia.
[20] Melbourne Water Corporation, Flood Management and Drainage
Strategy. 2007, Melbourne Water Corporation.
[21] The Telegraph. Super storm and giant hailstones lash Melbourne. 2010 08
March 2010, cited 2014; Available from: http://www.telegraph.co.uk
/news/worldnews/australiaandthepacific/australia/7394608/Super-storm-
and-giant-hailstones-lash-Melbourne.html.
[22] Insurance Council of Australia. Historical disaster statistics, cited 2014;

Available from: http://www.insurancecouncil.com.au/industry-statistics-
data/disaster-statistics/historical-disaster-statistics.
[23] Google Maps. no date; Available from: www.google.com.au/maps.
[24] Weiyan, T., Shallow water hydrodynamics. Elsevier Oceanography Series,
1992. 55: p. 434.
[25] Toro, E., Riemann solvers and numerical methods for fluid dynamics.
Third ed. 2009, Berlin Heidelberg: Springer-Verlag. p. 724.
[26] Glenis, V., et al., Flood modelling for cities using Cloud computing.
Journal of Cloud Computing: Advances, Systems and Applications, 2013.
2(7).
Section 5
Property-level flooding
and health consequences
(Special session
organised by C. A. Booth)
Improving the uptake of flood risk adaptation

measures for domestic properties in an
insurance regime under transition
D. Cameron1 & D. Proverbs2
1
Bristol City Council, UK
2
Abstract
In June 2013 the UK Government and the ABI announced plans for a new system
of insurance called ‘Flood Re’. This announcement was the first step towards
setting up a ‘not for profit’ scheme which aims to ensure the continuation of
affordable insurance for households with the highest flood risk. This research
investigates whether the widespread provision of flood insurance is a factor in the
low uptake of property level resilience measures. In the context of transition it
further examines whether there is growing impetus for the concept of
incorporating such measures. The literature establishes that the historic insurance
regime provided few incentives for installation of flood resilience measures and
that there are a number of factors beyond the provision of insurance which
influence the low uptake of measures. The impact of potential changes in the
insurance regime is explored in more detail through a number of semi-structured
interviews with key flood risk management professionals and academics. The
research finds that links between the provision of insurance and the installation of
resilience measures are significant. Flood resilience measures will continue to be
part of the wider strategy of community engagement with an integrated approach
to flood risk management. The newly proposed ‘Flood Re’ is intended to be a
transitory measure that will allow householders to adapt and take the necessary
measures to protect themselves. However, in the long term, the anticipated move
towards risk based pricing in whatever form may provide better incentives to
households to adapt and this could be reinforced by other measures to support
households in adaptation.
Keywords: adaptation, flood risk management, insurance, resilience.
doi:10.2495/FRIAR140121
1 Introduction
Flooding is the biggest natural threat facing the UK and flood risk is predicted to
increase due climate change, development and the gradual deterioration of flood
defence assets (ABI [1]). The UK has benefited from a private insurance system
for flood risk that has existed for over half a century and this has formed the main
source of financial protection for households in flood risk areas. A series of
agreements were in place between the government and the insurance industry
which ensured that the majority of households had access to affordable insurance
for flooding. These started with what was referred to as the ‘Gentleman’s
Agreement’ and more recently the ‘Statement of Principles on the provision of
flood Insurance’ (DEFRA [2]). On the 27th of June 2013, the Government and the
insurance industry announced a new agreement which would guarantee
the availability of insurance for households in flood risk areas. The preferred
solution would be an industry-run, not-for-profit scheme called ‘Flood Re’. This
scheme will effectively cap the maximum amount paid by the 1–2 % of households
at highest risk of flooding. It would be funded by an industry backed levy set to
be £180 million per year for the first 5 years, an equivalent of £10.50 for every
UK household. It will take time for ‘Flood Re’ to become operational and therefore
the insurance industry has voluntarily agreed to abide by the Statement of
Principles until such a time that ‘Flood Re’ can be introduced (DEFRA [2]).
The considerable uncertainty surrounding the decision over the future of
household flood insurance, together with an increased frequency of flood events,
has resulted in increased awareness of the concept of incorporating flood resilience
at property level. However, uptake of these measures remains low. In 2008,
DEFRA announced less than 5000 homes have adopted flood resilient and
resistance measures (Bichard and Kazmierczak [3]). To encourage an increase in
uptake in resilience measures DEFRA launched its property level flood protection
scheme. The 2 year programme ran until March 2011 and it delivered £5.2 million
to 1,109 households, the average cost to households for these measures was £4,832
(Environment Agency [4]). Whilst the UK Government has sought to influence
householders to take up flood protection measures, the strategies employed have
not been as successful as they had hoped (Bichard and Kazmierczak [3]). One
reason for this could be that the wide availability of insurance to households has
distorted their perception of risk. This study therefore sought to investigate the
extent to which the widespread provision of flood insurance is contributing to the
low uptake of property level resilience measures.
2 A background into flood insurance and resilience

Despite the increased frequency of flood events it is apparent that take up of flood
resilience measures is still low amongst householders. As insurance is still the
main source of financial protection for domestic households, it is important to
establish how insurance influences decision of those most at risk of flooding of
whether or not to install additional protection in their property.
2.1 Flood Resilience technology (FRe)
Resilience is defined as the ability of system/community/society/defence to react

to, and recover from, the damaging effect of realised hazards. The definition of
resistance is the ability of systems to remain unchanged by external events
(SMARTeST [5]). In the context of flooding these terms are used to describe
different methods of protecting property and communities. Flood resistance or dry
proofing methods attempt to keep the flood water out of the property; these are
only suitable methods for floods up to a certain depth. Flood resilience measures
allow the water to enter the property but then enable the drying and recovery
process to be undertaken swiftly. Contemporary thought is that these methods
should not be thought of in isolation. For example, the SMARTest project
describes these and other methods under the umbrella term of Flood Resilience
technologies (FRe). Here they try to steer away from the term property level
protection because some of the products, such as demountable barriers are used at
a community level, and others do not offer full protection, they merely speed up
the recovery (White et al. [6]). FRe Technologies can be important in smaller
communities where it is not cost beneficial to consider large scale flood defence
systems (Kazmierczak and Connelly [7]). Flooding comes from multiple sources,
and FRe technologies can be considered more flexible and adaptable when dealing
with surface water and flash flooding (White et al. [6]).
The Adaptation Sub-Committee reported that the uptake of such measures is
considered to be 20–35 times lower than the rate needed to reach all of the
properties that could potentially benefit within a reasonable timeframe Adaptation
Sub-committee [8]. They also predict that by increasing investment in flood
defences and property protection measures, the number of properties at risk could
be halved by 2035, which adds economic weight to the case for property level
protection.
2.2 Flood insurance beyond 2013
The Government and the ABI have agreed upon a Memorandum of Understanding
which sets out how ‘Flood Re’ is likely to operate in order to progress with the
development of Government policy. However, there are still many issues which
need to be resolved. The Government are to introduce new legislation in the Water
Bill to enable the introduction of ‘Flood Re’. The main powers will be to compel
all insurers offering household insurance to participate in ‘Flood Re’ and provide
for ‘Flood Re’ to be funded through an industry levy (DEFRA [2]). Householders
should be aware that ‘Flood Re’ will be a transitional measure, intended to be
phased out within 20 –25 years. A ‘Sunset Clause’ will be included in the primary
legislation to set an expiry date for ‘Flood Re’, as well as powers to ensure the
orderly winding down of the scheme (DEFRA [2]). The policy objective is that
there should be a gradual transition towards risk reflective pricing (a free market),
which is intended to increase incentives for flood risk to be managed properly. The
Government intends to seek powers in the Water Bill to allow them to stand ready
to regulate if ‘Flood Re’ can’t be made to work for consumers and insurers. This
‘Flood Insurance Obligation’ will require insurance companies to insure a
proportion of properties from a register of high risk households. This should create
a level playing field and overcome the competitive pressures on insurers to
withdraw from flood risk areas (DEFRA [2]).
2.3 Property protection and insurance
In the current insurance market, there is very little to suggest that the installation
of FRe technology to protect property will result in reduced premiums and
excesses for householders. A number of surveys have concluded that for the
majority, there was no evidence to suggest the installation of FRe technologies
would result in cheaper insurance (Bell [9]; Cobbing and Miller [10]; Harries
[11]). The way in which resistance and resilience measures can help is by avoiding
the need to involve insurance companies, or reducing the size of the claim made.
This can help to maintain access to mainstream insurance but is little incentive for
the installation of such measures.
For individual properties, and for those properties in areas that flood frequently,
FRe technology can be a cost effective means of reducing damage and disruption
(Harries [11]). However, many people perceive that flood resilience measures may
adversely affect property value or make their properties harder to sell. This is
perhaps a genuine concern; research by Lamond et al. [12] found that many
property buyers and sellers are often unaware of the flood risk to their property. In
fact there is evidence to suggest that flood prone properties aren’t discounted in
price over the long term. This was illustrated by Lamond et al. [12] who point to
properties in Bewdley which showed dips in value following flood events in 2001
and 2002. However, over the long term their value recovered (cited in Lamond
[p. 332, 13]). These measures therefore could be seen as a deterrent for potential
buyers. This perceived barrier is something which needs to be overcome.
2.4 Moral hazard
There is a longstanding and growing debate that the provision of insurance may
prevent some from taking the necessary steps to protect themselves (Priest et al.
[14]; Lamond and Proverbs [15]; Harries [16]; O’Neill and O’Neill [17]). The term
‘Moral Hazard’ is defined by O’Neill and O’Neill [17] as a situation in which
individuals or organisations do not bear the costs of a particular risk and hence
lack incentives to change behaviour to reduce that risk. The question of ‘Moral
Hazard’ has been raised with regards to the behaviour of homeowners in protecting
their own properties. It is used when there is a tendency towards less responsible
behaviour by those who believe they are insulated from financial risk by insurance
(Harries [16]). Understanding ‘Moral Hazard’ on behalf of the homeowner is
critical to ensuring that a system of insurance is put in place which encourages
homeowners to protect themselves. A balance needs to be found between
providing affordable cover for those who need it and encouraging some form of
self-protection which will reduce the impact of a future flood event. The difficulty
with this is that if insurance is the default position, and it is also widely available,
then policy holders are unlikely to consider other avoidance strategies (Lamond et
al. [12]).
2.5 Standards for resilient reinstatement
To eliminate this ‘Moral Hazard’, insurance companies could encourage

homeowners to install FRe by providing incentives for adaptation measures.
Botzen and Van Den Bergh [18] explain that insurance companies could limit
damage by rewarding well designed buildings with lower premiums. An existing
property that is reinstated with resistant or resilient measures could be rewarded
with a reduced premium rate or lower excesses. The ability of insurance companies
to provide incentives for flood mitigation measures would be a critical driver for
the uptake of FRe technologies. To do this, insurers would need to know that the
measures would actually work in practice. White et al. [19] stated that “The major
insurers are key to driving FRe, but they need to be assured that their installation,
maintenance, and performance means they can price effectively”. Trust is
therefore a key theme if FRe technology is ever going to reduce householder’s
premiums. To build that trust, standardisation is needed. Boobier [20] explains that
standards are essential to ensure that the minimum acceptable level of repair is
carried out. Currently there is no definitive set of standards for resilient repair,
although some may fall under the control of building regulations. There are many
publications which outline codes of practice and propose sets of benchmarks.
However, none have yet been universally adopted. Kidd et al. [21] explain that
“although the use of guidance is generally widespread…during a major emergency
it is generally less adhered to”. For resilient and resistant repair to make a
difference to householders premiums, strict building codes would need to be in
place and regulations would need to be enforced to ensure that buildings meet the
required standard before the work is rewarded (Botzen and Van den Bergh [18]).
3 Research design method and analysis

The current uncertainty over the future of flood insurance and the dynamic of the
discussions between the ABI and the Government meant that the situation was
evolving as this research developed. A series of semi-structured interviews was
undertaken with key stakeholders just prior to an agreement being announced. The
aim of these interviews was to explore the opinions of individuals who understand
different aspects of the flood recovery process.
The questions which were devised for the interviews were influenced directly
from the issues and topics arising from the literature review. The questions were
placed into three categories: 1) General Insurance Questions. 2) Resistance and
resilience, and 3) Accountability and Training. Table 1 presents a summary of
these questions.
The interviewees were chosen because of their credentials within the FRM
community. Many have contributed research which was studied as part of the
literature review, and some belong to organisations which form an important part
of the flood recovery process. All have influence within their specialisation or
organisation and have a depth of knowledge relating to FRM that was perceived
to be beneficial to this study.
A summary of the interviewees is presented in Table 2.
Table 1: Presentation of the questions asked to the interviewees.
General insurance questions

Question 1 Do you think that the widespread provision of flood insurance
under the Statement of principles may have caused
complacency amongst householders to provide property level
protection against flood risk?
Question 2 Uptake of property level flood adaptation measures has been
low. What could be done to persuade householders to take up
flood mitigation measures?
Resistance and resilience
Question 3 What do you think is the role of resistance and resilience in
reducing flood risk?
Question 4 Whose responsibility is it to encourage and promote the use of
property level protection? How could insurers help promote
the uptake of such measures?
Question 5 Should financial incentives be provided for the installation and
purchase of resistance and resilience products for those
properties at high risk of flooding? If so, in what form could
these incentives take?
Accountability and training
Question 6 Flood events are very unpredictable. What could then be done
to develop a way of monitoring the performance of flood
resistant and resilient materials?
Question 7 Do you feel that building professionals have the necessary
training/experience to deal effectively with householders that
have experienced a flood?
Question 8 Do you think England can learn lessons on flood policy from
other parts of the world?
3.1 Method of analysis
Due to the large volumes of data it was important to adopt a method of analysis
which made sense of the information and presented the findings in a logical and
coherent way. The transcripts were analysed using methods adapted from the
hermeneutic analysis method. ‘Hermeneutics’ is characterised by Haigh [22] as
examining the inter-relationship of the response from the interviews and relating
this to the aims of the research at large. In this case, one of the main challenges
was ensuring that the analysis remained focused upon the aims and objectives of
this thesis.
Table 2: Credentials of the interviewees.
Interviewee Credentials and organisation

A Professional/surveyor
B Professional/surveyor
C Insurance expert
D Researcher/academic
E Researcher/academic
F Risk expert
G Community spokesperson
4 Analysis of interviews
The majority of Interviewees acknowledged that there was a link between the
availability of affordable insurance and households protecting their own
properties. The general feeling was that if people had always been paid out on
insurance then they would not be motivated to protect their property. However,
Interviewees C and F made the point that there is more to the issue than just
financial impact. The emotional stress that is suffered from flooding is also
significant. Interviewee C said that “they hoped that where it is beneficial for
someone to protect their property they would do this irrespective of the
affordability and the availability of insurance”. In the literature review however,
there was little evidence to support this statement. It was found that uptake of flood
protection measures is still very low. The complexity of dealing with flood risk
was cited by interviewee D as a possible reason for the low uptake of these
measures: “It’s not as clear cut as installing a safety lock to your front door…it is
less clear how these mechanisms are going to work”.
It was apparent there was no simple answer to the problem of persuading
householders to take up FRe technology. It was surprising that only one
Interviewee (B) thought reduced premiums could be used. This reflects the
perceived complexity of implementing such a scheme. Interviewee A thought that
outright refusal of cover, which would make the property un-mortgageable, would
be a possible driver. They also thought pilot grant schemes from Defra which have
now evolved into partnership funding schemes were motivational for people.
Interviewee G felt that people were more likely to take up flood resilient measures
if they were passive, such as flood doors that look like normal doors and kite
marked one way valves. “We are getting there but the PLP industry is still very
young and these products are still very expensive”.
Integration into building regulations and more robust planning were responses
that also stood out.
4.1 Resistance and resilience
In response to the question of role of resilience and resistance measures in reducing

flood risk, interviewees B and F mention passive measures such as self-closing air
bricks and front doors with integrated protection. These measures are favoured by
insurance companies because they remove the element of human error; once
installed they don’t need to be set up. Interviewee F points out that
“We protect our home with burglar alarms, and smoke alarms…it seems that
simple procedures could be put in place if the house is at risk of flooding”.
The most common response was that resistance and resilience measures shouldn’t
be taken in isolation; they should be part of a portfolio of measures. This view is
supported by the literature and the EU directive that flood risks need to be dealt
with in a more integrated way.
Interviewees A, B, E and G agreed that resistance and resilience should be used
in conjunction with other measures. In some cases flood defences may be the most
cost effective solution. Carrying out a cost/benefit analysis was highlighted by C,
D, and F all mentioned that an economic assessment of cost and benefit was vital
to prevent measures being carried out where they were not needed. Rural
communities that are scarcely populated were places which could benefit from
this. An interesting point was highlighted by interviewee C, who said that quite
often the resilience measures are designed to protect internal fixtures and fittings
that may have a design life much shorter than the expected return period of the
flood. Here, it may be more cost beneficial to assume the product will need
replacing by then anyway. Interviewee G points out that the word ‘defences’
conveys the wrong impression. It implies that they provide complete protection.
The language of flood risk management is more appropriate, and the use of
integrated techniques as part of a community flood action plan is more effective.
Interviewees A, B and G felt that for FRe technology to be promoted it needed
to be interlinked with insurance premiums. Interviewee A pointed out that if
bankers got involved and FRe measures suddenly became a condition of
mortgages then this would encourage uptake. Interviewees C, E and F thought that
flood risk was still the strategic responsibility of the Government and their
agencies. Interviewee E made the point that for insurers to get involved, it would
have to be for their own commercial advantage. Insurers are in the business to
make a profit and have no social responsibility to help householders. Interviewee
G explains that trying to get people to engage by going into communities to raise
flood risk awareness is valuable, but it is trying to push the ball uphill. Using this
analogy, he suggests that it also about trying to find policy levers so the ball can
be pulled up hill at the same time. He suggests that one such policy lever could be
to make it a condition of insurance, and this would be a strong incentive.
4.2 Accountability and training
A possible method of monitoring performance of FRe technology could be via

stricter building codes and the benchmarking of products. Interviewee E suggested
that there was scope for independent research as some manufacturers may not be
able to afford the kite mark but this does not mean their products are not useful.
From an insurer’s point of view, this kind of assurance is essential because if they
are ever going to offer discounts for flood protection measures they will need to
know the quality of the workmanship involved. Interviewee A pointed out that
there are precedents linking the benchmarking of quality with insurance. He
suggests standards for sprinkler systems and burglar alarms need to be adhered to
as a requirement for insurance against fire and theft. Respondent G says an in depth
report on what did and didn’t work in 2012 was required from the Environment
Agency and Defra in order to drive up standards.
The general feeling amongst interviewees was that building professionals
lacked the necessary experience and training to deal effectively with flood risk.
Even from those within the surveying profession. Respondent B highlighted the
complexities of the drying process was an area that needed better understanding.
There was acknowledgement that there are professionals who specialise, but that
for the moment they are in the minority. There are dedicated facilities in the UK,
such as the National Flood School, which specialise in training in flood restoration
for building professionals. There was concern with Interviewees E, D and F, that
Local Authorities, with their greater responsibility for FRM, may not yet have the
skills necessary to take on this role.
The response to whether we can learn from flood policies from other parts of
the world was that even though there are always lessons that can be learned from
other countries, there is not one specific model that will necessarily solve all the
issues. Other countries will have different climates, landscapes, populations and
social structures. There are many elements that forbid the applicability of a generic
solution to individual cases in the UK. Interviewee D rightly suggested that the
EU strategy which led to the ‘making space for water’ directive, has set out a
comprehensive new approach and triggered a rethink in the UK as to how flooding
is dealt with.
5 Conclusions
The research has established several key points that need to be achieved to promote
and incentivise the use of property level resilience. Amongst these, stronger
partnerships between insurance companies and those developing FRe technology
should be established. Passive measures which are deployed automatically could
be a key factor in encouraging insurance companies to reduce premiums as this
would remove the element of human error when it comes to deploying these
products. Insurers could be more influential in promoting and incentivising
property level resilience. For this to work there needs be assurances that the
products are going to reduce the amount the insurance companies pay out in the
event of a flood. There needs to be standardisation, both for the products and the
installation. Kite mark schemes are improving, and there are a lot more tested
products on the market. However, standards for installation need to come from
regulation, at the moment there is plenty of guidance and codes of practice, but in
an emergency these tend to be less adhered to. A relatively small reduction in the
cost of premiums will not be enough of a driver for change, because the initial
costs for the installation of FRe technology may be too high. Community schemes
need to allocate money to those households that are most vulnerable and need to
encourage innovative ideas for community resilience beyond the use of FRe
technology.
It has been established from the research that the provision of flood insurance
does impact upon householders decisions of whether or not to install property level
resilience measures. The term ‘Moral Hazard’ is used by many researchers to
describe a tendency towards less responsible behaviour by those who believe they
are insulated from financial risk by insurance. The difficulty is that if insurance is
the default position then it is unlikely that householders will consider other
avoidance strategies. It is important to emphasise that the measures proposed to
protect the availability of affordable flood insurance under ‘Flood Re’, are only
intended to be temporary. They will be phased out within 20-25 years, when the
market will move towards risk reflective pricing. There is a danger that some
householders that could benefit from FRe technology will be drawn into a false
sense of security by the availability of affordable insurance. It is therefore
important that during this transitional period of ‘Flood Re’, opportunities are taken
to develop and promote flood resilience for properties which will benefit most
from these measures. This assertion is backed up by the findings of the literature
and the interviews, which have highlighted that property level resilience measures
will need to play an increasingly important role in managing future flood risks.
References
[1] Written evidence to the Environment, Food and Rural Affairs Select
Committee inquiry into flood funding. Association of British Insurers, ABI.
Online. http://www.publications.parliament.uk/pa/cm201213/cmselect
/cmenvfru/writev/flood/m07.htm
[2] Securing the future availability and affordability of home insurance in areas
of flood risk. Department for Environment, Food and rural Affairs, DEFRA.
Online. https://consult.defra.gov.uk/flooding/floodinsurance
[3] Bichard, E. & Kazmierczak, A., Are homeowners willing to adapt to and
mitigate the effects of climate change? Climate Change (2012) pp. 112:
633–644, 2011.
[4] Guidance on surface water flood mapping for lead Local Authorities.
Environment Agency: Bristol, 2012.
[5] SMARTeST – Glossary. Online. tech.floodresilience.eu/attachments/article
/40/smartest-glossary.pdf
[6] White, I., Lawson, N., O’Hare, P., Garvin, S. & Connelly, A., Six Steps to
Property Level Flood Protection – Guidance for local authorities and
professionals: Manchester, 2012.
[7] Kazmierczak, A., and Connelly, A., Buildings and Flooding – a risk
response case study. EcoCities project, University of Manchester.
Manchester, 2011.
[8] Adaptation Sub-committee., Climate change – is the UK preparing for
flooding and water scarcity: Adaptation Sub – Committee Progress Report
2012, Committee on Climate Change: London, 2012.
[9] Bell, A., Morpeth Flood Action Group Insurance Survey – Results and
Analysis, Online. http://www.morpethfloodaction.org.uk/survey.html
[10] Cobbing, P., and Miller, S., Property level protection and insurance: Main
report – 2012, National Flood Forum: Bewdley 2012.
[11] Harries, T., Review of the Pilot Flood Protection Grant Scheme in a
Recently Flooded Area. Department for Environment Food and Rural
Affairs: London, 2009.
[12] Lamond, J. E., Proverbs, D.G., and Hammond, F.N., Accessibility of flood
risk insurance in the UK: confusion, competition and complacency, Journal
of Risk Research, 12 (6), pp. 825–84, 2009.
[13] Lamond, J. E. (2012) Financial Implications of Flooding and the Risk of
Flooding on Households, in: Lamond, J., Booth, C., Hammond, F., and
Proverbs, D. (eds.) Flood Hazards: Impacts and Responses for the Built
Environment. CRC Press: Boca Raton pp. 317–326, 2012.
[14] Priest, S.J., Clark, M.J., Treby, E.J., Flood Insurance: The challenge of the
uninsured, Area 37.3 Royal Geographical Society, pp. 295–302, 2005.
[15] Lamond, J.E., Proverbs, D, G., Flood Insurance in the UK – a survey of the
experience of flood plain residents, in: Proverbs, D., Brebbia, C.A., and
Penning-Rowsell, E. (eds.) Flood Recovery, Innovation and Response, WIT
Press: Southampton, 2008.
[16] Harries, T., Why Most “At Risk” Homeowners Do Not Protect Their Homes
From Flooding, in: Lamond, J., Booth, C., Hammond, F., and Proverbs, D.
(eds.) Flood Hazards: Impacts and Responses for the Built Environment,
Taylor and Francis Group: Boca Raton, 2012.
[17] O’Neill, J., and O’Neill, M., Social Justice and the future of flood insurance,
Joseph Rowntree Foundation: York, 2012.
[18] Botzen, W.J., Van Den Bergh, J.C.J.M., Monetary valuation of insurance
against flood risk under climate change, International Economic Review,
53, pp. 1005–1026, 2012.
[19] White, I., O’Hare, P., Garvin, S., Connelly, A. (2012) Barriers to flood
resilience: Findings from the SMARTEST project, The University of
Manchester and BRe: Manchester, 2012.
[20] Boobier, T. (2012) The Development of Standards in Flood Damage Repair:
Lessons to be learned from the United Kingdom Example in: Lamond, J.,
Booth, C., Hammond, F., and Proverbs, D. (eds.) Flood Hazards: Impacts
and Responses for the Built Environment, CRC Press: Boca Raton, pp. 129–
139, 2012.
[21] Kidd, B., Tagg, A., Escarameia, M., von Christierson, B., Lamond, J.,
Proverbs, D. Guidance and standards for drying flood damaged buildings,
Signposting current guidance – BD2760, 2010.
[22] Haigh, R. Interviews: A negotiated partnership, in: Knight, A., and Ruddick,
L. (eds.) Advance Research Methods in the Built Environment. Oxford:
Blackwell: Oxford, pp. 111–120, 2008.
Waterproofing basement apartments: technical

insights of a new flood protection solution
D. W. Beddoes1,2 & C. A. Booth1
1
Construction and Property Research Centre,
2
DrainAngel Ltd., UK
Abstract
Installing perimeter floor drains is a waterproofing option used inside buildings,
which manages water ingress in basements. Newly designed products providing
a means to access the inverts of perimeter floor drains and facilitate inspection
and maintenance have been designed and are Patent Pending GB1117089.1,
GB1102662.2, and GB1102661.4. The new system incorporates pivotally
connected fittings with water deflector plates combined with straight lengths of
perimeter floor drain. The fittings ensure that secure joints with both axial and
invert alignment are maintained throughout, which are essential for movement of
water through the level perimeter floor drains and accessibility as recommended
by British Standard BS8102: 2009. The patents demonstrate several practical
advantages over those of existing designs and systems, which is evidenced by the
commercial uptake by Safeguard Europe Ltd. and product installations in
hundreds of flood-risk basement apartments in the UK.
Keywords: property adaptation, perimeter floor drainage, flood resilience,
patent product.
1 Introduction
Provision of a building basement can increase housing density without a
reduction in habitable space and so enable more homes to be built in a
development where building height or footprint is limited [1]. Basements have
long been used in commercial buildings for plant rooms, storage space and car
parking but in recent decades the arcane construction used in basements has been
transformed by modern technology to provide fully acceptable below ground
doi:10.2495/FRIAR140131
accommodation, which is particularly beneficial in urban areas where space is

limited [2]. Moreover, energy costs can be saved by the use of a basement as
heat loss through basement floors and walls is restricted by the insulating effect
of the ground, providing possible energy savings of up to 5.6% for a semi-
detached and up to 9.5% for detached properties [3]. One of the major challenges
to be addressed with below ground accommodation is the means for preventing
the entry of water and water vapour from the surrounding ground into rooms that
are wholly or partly below ground level [4]. England has over 20 million
dwellings and some 550,000 of these have some form of basement
accommodation and the extra space provided by a basement combined with a
general shortage of building land leads to the development of sloping sites and
results in around 10,000 new basements being constructed every year in England
[5, 6]. In response to an increase in basement construction, the development of
new waterproofing materials, more deep basements in cities and the need to
mitigate inherent risks associated with below ground structures, the British
Standard has been revised. BS8102:1990 code of practice for protection of
structures against water from the ground was revised to produce BS8102:2009
code of practice for protection of below ground structures against water from the
ground [7, 8].
Three types of waterproofing protection should be initially considered [8],
where the choices are: (a) barrier protection (Type-A) relies on a separate
waterproofing barrier (applied to the structure), which must be totally free of
defects if it is to keep water out as hydrostatic pressure will cause flooding of a
basement through the smallest of holes; (b) structurally integral protection
(Type-B) is provided by the design and materials incorporated into the structure
itself and, as such, usually means a building structure using high quality
reinforced concrete. Problems come from day joints and construction joints
where ‘waterstops’ in the form of passive (rubber) or active (hydrophilic) strips
can sometimes fail at these joints; or (c) drained protection (Type-C) is installed
internally and has a major advantage in that no extra loading is placed on the
structure, it is a system of internal water management. Plastic cavity drainage
membranes combine with perimeter floor drain systems to collect water ingress
and direct it to a sump so that it can be pumped from the building. Type-C
protection is cost effective and can be retro-fitted to existing basements or
readily used as a remedial solution to basements where other types have failed.
The Type-C system is not destructive to the basement structure, can be reversed
in conservation work, provide insulation together with sound absorption, whilst
ensuring that the basement remains dry. Ease of installation, good value and
reparability are further major advantages of Type-C protection.
2 Preference and problems of basement waterproofing

As a result of developments in plastic membranes and new efficient sump pumps
the commonly preferred choice for basement waterproofing is Type-C
protection. However, problems have occurred in the past with blockages in
membranes and drainage leading to new recommendations in the revised British
Standards BS8102: 2009 that all drainage systems and installations must now be
able to be tested, incorporate accessibility and be maintainable for the life of the
structure. In the past the ‘Achilles heel’ of Type-C protection has always been
accessibility into perimeter drainage systems that were buried within the
structure, which often resulted in the removal of whole floors to investigate
defects.
New products which provide means to access the inverts of perimeter floor
drains and facilitate inspection and maintenance are now commercially available
[9]. The designs featured are Patent Pending GB1117089.1, GB1102662.2, and
GB1102661.4 [10–12]. Perimeter floor drains are used inside buildings that are
subject to water ingress, typically basements and flood situations. In a typical
basement construction the perimeter floor drain collects water from Type-C
(drained cavity) installations as described in BS8102: 2009. A waterproofing
installation to BS8102: 2009 Type-C uses waterproof structures to form a cavity
between floor, adjacent wall, and/or ceiling which is then drained into the
perimeter floor drain. However, the Type-C system is incapable of accepting any
hydrostatic pressure and the collected water must be removed from the system or
leaks will occur. The perimeter floor drain is one of the most important parts of
the water collection and management system that must function properly to
prevent water from entering the basement.
The perimeter floor drain is laid directly onto the level floor of the building at
the internal wall to floor join, therefore the invert of the perimeter floor drain has
to be laid level on the basement floor. The perimeter floor drain is installed
around the perimeter of the basement and a T-piece is incorporated at some point
to transfer all the water collected by the perimeter floor drain into a link drain
which then leads to a sump/pump unit. The sump/pump unit is typically located
within the central floor area having a discharge pipe leading to an outside
domestic drain. All water ingress through walls, floor, ceilings, the vulnerable
wall to floor join and collected from the waterproofing cavities is transported
along the perimeter floor drain to the link drain and then into the sump/pump for
removal from the building. Most importantly these perimeter floor drains and
link drains have no gradient to move the water as they are laid directly onto a
level floor. It is the successful movement of the water along the perimeter floor
drain that presents the current problem because if the water is allowed to build
up then as the Type-C installation cannot withstand hydrostatic pressure we will
get leaks into the internal basement space. Furthermore, if water is allowed to
pond within the perimeter floor drain then the standing water leads to
recrystallization of salts which have entered as a solution within the water.
Blockages can then occur due to the deposition of recrystallized salts, which
cause more standing water and more blockages etc.
Water can only move along level perimeter floor drains by hydraulic head due
to the gravitational effects on the water. This driving force is very weak and it is
therefore essential that the invert of the perimeter floor drain is completely level
and has no obstructions. Any minor obstruction or misalignment of a perimeter
floor drain invert causes an increase in depth of water upstream of the
obstruction which may then exert hydrostatic pressure on the waterproofing
structure and cause a leak into the occupied basement. As a simple illustration
we can consider a perimeter floor drain installed in a basement that is 8 x 6m on
plan. The water collected by the perimeter floor drain at the furthest point from
the link drain will have to travel at least 14m and negotiate at least two elbow
bends and a T-piece connection into the link drain when the depth of the
perimeter floor drain and all that needs to be filled up in order to cause a leak is
only 40mm depth of standing water within the drain. This example does not take
into account the fact that matters are often much worse as there may be a back-
fall due to the floor itself not being exactly level across its surface. A floor that is
8m in length may well be 25mm lower at the point furthest away from the link
drain, combine this with a mere 5mm invert obstruction at each of the two
misaligned bends and the T-piece and we have the perimeter floor drain full of
water with a potential leak into the occupied building. In these circumstances,
which are unfortunately often found on site, the perimeter floor drain cannot be
accessed in order to be flushed out and cleaned and it is not possible to maintain
the system as per the requirements of BS8102: 2009.
The perimeter floor drain itself is usually made from plastic and has a smooth
invert. Usually it is the elbow bends and T-piece, where inverts become
misaligned, that causes problems with obstructions that lead to a leak. As a result
of problems with Type-C installations and subsequent leaks all new work
involving the use of perimeter floor drains is now subject to revised BS8102:
2009 code of practice for protection of below ground structures against water
from the ground. This applies to basement and flood situations where
accessibility and repairability must be allowed for in design. For example
BS8102: 2009 page 13: shows a cross-sectional view of Type-C (drained)
protection that details a perimeter floor drain as a maintainable drainage channel,
BS8102: 2009: 10.2.1.2. requirements: where the floor cavity incorporates
perimeter floor drain channels, which discharge into sump(s), both the channels
and the sumps should be cleaned before, during and after installation of the
membrane to allow uninterrupted drainage, BS8102:2009: 10.3.1. requirements:
access points that allow routine maintenance of channels and outlets should be
incorporated into the design of the waterproofing system, BS8102: 10.3.2.
requirements: immediately after the installation of a cavity drain system the
perimeter floor drainage channels and sumps should be cleaned out and tested.
The servicing requirements for the waterproofing system should be clearly set
out in the documentation supplied by the designer to the client, including the
need for regular planned maintenance of the drainage and/or pumping systems
not less than once a year.
In order to have perimeter floor drains that function correctly and also satisfy
the requirements of British Standards for flushing out and cleaning perimeter
floor drains, it is necessary to have perimeter floor drain joints, elbow bends and
T-pieces that are securely fitted together and accurately aligned both axially and
across inverts. The current methods of perimeter floor drain installation use a
straight butt joint that is often held together with adhesive tape. These butt joints
are easily disturbed both during their own installation and also during the
subsequent floor laying and framed wall building operations that are carried out
immediately on top of the perimeter floor drains by different tradespersons. At a

corner joint in the perimeter floor drain the installers currently use a wood-saw to
roughly mitre the corners of the perimeter floor drain and then attempt to wrap
adhesive tape around the joint in less than favourable damp conditions; such
joints are easily disturbed. In a similar manner, the T-pieces are formed by
cutting out the side of a perimeter floor drain with a wood-saw and simply butt
jointing an intersecting link drain connection, attempting to tape together the
joint in the wet conditions. The end product inevitably results in a perimeter
floor drain with joints that have moved and have thus produced obstructions to
the flow of water around the perimeter floor drain. Subsequent attempts to flush
out and clean the perimeter floor drain then result in water backing up around the
perimeter floor drain leading to standing water and leaks into the occupied
basement.
Some manufacturers and contractors have introduced rigid corner pieces and
rigid T-pieces but these have brought their own problems. The rigid items are
difficult to fit and align with the runs of the perimeter floor drains as most
corners encountered in buildings are not exactly 90° and most sumps cannot be
directly accessed by a 90° T-piece. Some adjustment of the 90° T-piece is always
necessary on site because the sump has to be installed so that as far as possible
the water from the entire floor area being treated can find its way to the sump.
The sump must be placed in an area like a door opening where it always remains
accessible and the sump must be positioned so that a connection can be made to
a nearby drain or so that the installed pump can be plumbed to a drainage point.
More importantly, the use of rigid elbow bends and rigid T-pieces leads to the
need for even more straight butt joints at each side of the installed fittings where
they meet the straight lengths of perimeter floor drain that run around the floor
perimeter. A solution is needed to insure that the inverts of a perimeter
floor drain system when installed are kept level throughout and have no
obstructions to impede water flow. The system must address the problems at
joints, elbow bends and T-pieces were not only inverts must be in line but also
the axial or longitudinal axis alignment of the perimeter floor drain and link
drain must be maintained in order to ensure a secure fixing and also enable the
flow of water along a channel that has no gradient. This will then ensure that
the perimeter floor drain does not sit with standing water held continuously
within the perimeter floor drain. When flushing out and cleaning is underway the
water introduced can successfully make its way around the perimeter floor drain
and flush any sediment into the sump, as required by the revised British
Standards of BS8102: 2009. Therefore, this document discloses an accessible
system having perimeter floor drains combined with new pivotally connected
elbow bends that feature water deflectors and a new pivotally connected T-piece
with internal deflector plate. The new bend and the new T-piece can be fitted to
both one part and two part perimeter floor drains and so can be used with any
manufacturers’ perimeter floor drain to provide a secure joint with level inverts.
3 Detailed and diagrammatic description of the new system

The system is described by reference to the accompanying drawings: Figure 1 is
a line drawing of a typical perimeter floor drain that shows an end elevation of
one part perimeter floor drain and two part snap together perimeter floor drain.
Figure 1(a) shows a one part perimeter floor drain with an upstand (A), a
perimeter floor drain can be installed with or without this upstand. The upstand
is sometimes used against the inside wall of a building to hold the lower edge of
a waterproofing structure in place and hence is not always needed. The one piece
perimeter floor drain has holes (B) in the channel sidewall in order to collect
water that has passed through the external building structure. Figure 1(b) shows a
two part perimeter floor drain that consists of upper flat soffit section (C) and
lower channel section (D). The two part perimeter floor drain may also feature an
upstand (A) where needed and has the holes (B) to collect water ingress into
channel section (D). The two separate parts, upper flat soffit section (C) and
lower channel section (D) securely snap together as shown at (E).
Figure 2 is a plan view to show a typical basement installation of the new
accessible system having perimeter floor drains combined with the new pivotally
connected elbow bends and new T-piece. The external masonry structure of the
building that serves to filter the water ingress as it enters the building is shown as
(F). Inside the building the straight lengths of perimeter floor drain (as Figure 1)
are shown as (G) and the link drain (H) transfers the water to the sump/pump (J)
or gravity exit point. The pivotally connected T-piece with internal deflector (K)
transfers water from perimeter floor drain into the link drain (H). The pivotally
connected elbow bends (L) are situated at each internal corner of the building
and can be adjusted to suit each corner to ensure axial alignment of perimeter
floor drains. An access point or water jetting point (N) can be used to introduce
flushing water into the system which will then make its way around the
perimeter floor drain as shown by arrows (M). The new pivotally connected
elbow bends (L) and the new pivotally connected T-piece (K) will ensure that the
inverts are level across every joint and that axial alignment of channels across
joints is achieved. Hence without any obstructions the collected water will flow
under gravity along the level channels to the T-piece (K) where it will transfer
into the link drain (H) and into the sump (J) for removal from the building [10].
Figure 3(a) shows the new pivotally connected elbow bend with internal
deflector. The bend is pre-assembled as shown with two top flat soffit sections
(P) which are mitred and then a pivotal connection is made across the mitred join
using a connecting water deflector plate (R) and connectors (Q). The gap (W)
between the two top section mitred edges and mitred channel sections (S) allows
the two halves to rotate relative to each other so that the elbow bend can be fitted
into building corners that are not exactly 90°. This is an important feature in
maintaining both axial and invert alignment and it allows secure joints to be
made between straight lengths of perimeter floor drain and the elbow bend
fittings. The lower channel sections (S) extend out past the top sections (P) and
are a feature of the new elbow bend as they are used to form a secure joint with
the straight lengths of perimeter floor drain. If the two part perimeter floor drain
(Figure 1(b)) is being used on an installation the straight channel of the perimeter
floor drain is placed against the end of the extended elbow bend channel (S) and
then the top section of the perimeter floor drain is snapped into place, spanning
across the channel joint to create a secure staggered joint. There is no longer a
straight butt joint that passes directly through both top and channel sections of
the perimeter floor drain which is the major disadvantage associated with current
rigid bends. If a one part perimeter floor drain is being used for the straight
lengths of perimeter floor drain then the protruding channel section (S) will slide
inside any manufacturers’ one part perimeter floor drain channel currently
available to form a secure joint. The pivotally connected elbow bend is universal
and can therefore be used in installations of two part and one part perimeter floor
drains and in both cases will provide a secure joint to the straight lengths of
perimeter floor drain and the pivotal connection will ensure that soffits and
inverts are kept at the same level across the bend to prevent obstructions to water
flow [11].
Figure 1: Sectional drawings of the Figure 2: Plan drawing of the

perimeter floor drain. perimeter floor drain.
Figure 3(b) shows the pivotally connected T-piece with internal deflector
plate. The top section (U) of a length of two part perimeter floor drain is
pivotally connected (Q) to the link drain top section (V). This pivotal connection
allows adjustment of the angle of intersection at the T-piece and ensures that on
a construction site installation of the link drain is axially aligned into the T-piece
in order to prevent obstructions to water flow and also achieve a secure joint.
The link drain top section (V) is set under the top section (U) so that the invert in
the link drain is lower than that of the perimeter floor drain to encourage water to
flow from the level perimeter floor drain invert into the link drain invert. A
length of two part lower channel section has the side wall cut away and is fixed
into the top section (U), similarly a lower channel section is cut and fixed into
the link drain top section (V). In both cases the channel sections are longer than
their respective top sections and protrude out as shown (S). As previously
described above for the elbow bend (see figure 3(a)) the protruding channels
provide secure joints to both one part and two part perimeter floor drain straight
lengths to ensure axial and invert alignment and hence no obstructions to flow.
The lower channel section at the point of intersection must have the sidewall
removed to allow water to pass into the link drain and removal of the sidewall
weakens the construction of the T-piece and reduces the capacity of the flat top
soffit section to handle floor loadings. A water deflector and support is internally
fitted that spans between invert and flat soffit top section at the point of
intersection. The support sits inside the T-piece in the channel section and serves
to both support the weak flat top section and due to its shape also deflects water
into the link drain passageway [12].
(a) (b)
Figure 3: Sectional drawing to illustrate (a) the new pivotally connected

elbow-bend with internal deflector and (b) the pivotally
connected T-piece with internal deflector plate.
Figure 4 is a plan view of perimeter floor drain, T-piece, link drain and sump
chamber to show the accessibility provided by the new pivotally connected T-
piece with water deflector plate and support (b) and an existing standard T-piece
in (a). The left hand side (a) shows perimeter floor drain A with a standard T-
piece W that joins the perimeter floor drain to the link drain G. Link drain G runs
to the sump/pump chamber H. The drain inspection camera or hose R can enter
through the sump/pump chamber lid and be pushed along the link drain G. When
R reaches the T-piece it cannot negotiate the corner into the perimeter floor drain
and hits against the channel wall, there is no way to direct the camera or hose
around the corner and along the perimeter floor drain. In Figure 4 the right hand
side (b) the new T-piece has the internal water deflector plate and support D
attached. The drain inspection camera or hose R can now pass around the corner
into the length of the perimeter floor drain. By means of pushing R along
different sides of the link drain both lengths of perimeter floor drain on either
side of the T-piece can now be accessed. The internal water deflector plate and
support D provides a constant radius for the bend in the inspection camera or
hose and prevents kinking as they are fed through the T-piece W along the
perimeter floor drain A.
Figure 4: Plan views of the perimeter floor drain, T-piece, link drain and
sump chamber to show the accessibility provided by the new
pivotally connected T-piece with water deflector plate.
Figure 5(a) is a plan view to show water and/or sediment flowing into the
sump/pump chamber. The flushing water has been introduced through jetting
points set into the perimeter floor drain. In this plan the perimeter floor drain A
is connected to the link drain G with a T-piece W. The T-piece W has an internal
support D attached. This plan shows the drainage system in use. The arrows C
show the flow of water through the system. The perimeter floor drain collects
water though pre-drilled holes in the channel sides, this water runs to the T-piece
where it is passed into the link drain G and hence on to sump/pump chamber H.
The arrows C show flow of water and/or flow of sediment when the drainage
system is being cleaned by flushing out. The support D stops water and/or
sediment being washed back and forth across the end of the link drain and flows
into the link drain G. Subsequently, sediment can be removed by way of the
sump/pump chamber lid. As shown in Figure 4 one side of the new T-piece can
be used to introduce a hose into the perimeter floor drain and the flushing water
will then travel around the perimeter floor drain to carry sediment back to the
new T-piece where the deflector plate will direct it into the link drain and then
onto the sump for removal.
Figure 5(b) is a line drawing of the new T-piece to show the internal water
deflector plate that also acts as a support for the T-piece soffit. For illustration
purposes this drawing shows a rigid connection at the T-piece join, whereas in
practice the new T-piece also features an adjustable joint and an invert level
slightly lower in the link drain connection to aid water movement from perimeter
floor drain to link drain and sump [12].
(a) (b)
Figure 5: (a) Plan drawing to illustrate water and/or sediment flowing into
the sump/pump chamber and (b) a sectional drawing of the new T-
piece to show the internal water deflector plate that also acts as a
support for the T-piece soffit
5 Discussion
The products/system portrayed offers several advantages over existing
approaches to the option of installing a drained cavity construction in a basement
apartment. Pre-made elbow bends and T-piece make installation much easier and
quicker during construction. The new T-piece with internal water deflector
enables access into the perimeter floor drain system for inspection and effective
flushing via the sump chamber. The pre-made bends can be fitted to the exact
corner angle and then the secure joints hold the system together during assembly.
The T-piece link drain connection can be accurately aligned to meet the sump
location. The whole installation process needs less skill to complete, as the
operatives no longer have to try and mitre odd shaped plastic mouldings with
hand tools. With pre-made items the perimeter floor drain installation is simply
snapped together with invert levels and axial alignment guaranteed (Figure 6).
The deflector plates incorporated into corners and T-piece provide the
accessibility as recommended by the revised British Standard BS8102 2009 for
inspection and maintenance. The contractor can demonstrate on handover to the
client that the system works and is now able to build in the accessibility to
investigate any faults and also offer periodic maintenance contracts for the peace
of mind of the client.
Figure 6: Photos of the newly designed, easy to fit, perimeter floor drain
being installed by a semi-skilled professional in a basement
apartment.
6 Conclusions
Shortage of housing is encouraging the conversion of building basements into
habitable spaces. Mitigating the impact of below-ground living means there is a
need to adapt and protect accommodation against water-ingress through the
building envelope. Recommendations and guidance on the available approaches
for dealing with the entry of water from surrounding ground into a structure
below ground level include the use of a waterproofing barrier applied to the
structure, creation of a structurally integral watertight construction
or installation of a drained cavity construction. With the latter approach proving
popular, new flood-resilient products, which provide means to access the inverts
of perimeter floor drains and facilitate inspection and maintenance, are Patent
Pending GB1117089.1, GB1102662.2, and GB1102661.4. These can
demonstrate several practical advantages over those of existing designs and
systems. Recognition is demonstrated by the commercial uptake by Safeguard
Europe Ltd. and their installation in several hundred flood-risk basement
apartments in the UK
References
[1] Building Research Establishment, (2007) Good Building Guide 72:
Basement Construction and Waterproofing. Part 1: Site Investigation and
Preparation. Part 2 Construction, Safety, Insulation and Services.
Amersham: IHS BRE Press.
[2] Construction Industry Research Information Association (1995) Water-
Resisting Basements- Report 140. London: CIRIA.
[3] Tovey, A. and Keyworth, B. (1998) Basements: Land Use and Energy
Conservation – Evaluation with Market and Construction Survey.
Crowthorne: British Cement Association.
[4] Basement Information Centre (2004) Basement Information Centre
Approved Document – Basements for Dwellings. Camberley: BIC.
[5] Department of the Environment Transport and the Regions (1998) English
House Condition Survey 1996. London: The Stationary Office.
[6] Wolcox, S. and Perry, J. (2013) UK Housing Review: 2013 Briefing
Paper. Coventry: Chartered Institute of Housing.
[7] British Standards Institution (1990) BS8102:1990 Code of Practice for
Protection of Structures against Water from the Ground. London: BSI.
[8] British Standards Institution (2009) BS8102:2009 Code of Practice for
Protection of Below Ground Structures against Water from the Ground.
London: BSI.
[9] www.safeguardeurope.com/products/aquadrain.php
[10] Beddoes, D.W. (2011a) Cleanable Perimeter Drain System. Pat Pend.
117089.1.
[11] Beddoes, D.W. (2011b) Drain Deflector. Pat Pend. 1102662.2.
[12] Beddoes, D.W. (2011c) Perimeter Floor Drain T-Piece. Pat Pend.
1107397.0.
An investigation of patterns of response and

recovery among flood-affected businesses in the
UK: a case study in Sheffield and Wakefield
N. Bhattacharya-Mis1 & J. Lamond2
1
Faculty of Science and Engineering, University of Wolverhampton, UK
2
University of West of England, UK
Abstract
Despite the increasing impacts of recurrent flooding, there is dearth of research
involving businesses preparedness and recovery. This research therefore focused
on investigating the patterns of preparedness and trends in recovery among
business properties. A review of literature was performed primarily to recognize
the gaps requiring investigation followed by identification of two case studies
(Wakefield and Sheffield in the UK) for empirical data collection. The survey
enquired about the level of preparedness among a sample of the flood-affected
business community using a self-administered questionnaire. Questions addressed
the type of mitigation and preparedness activities and measures that they engaged
in and adopted for recovery along with factors like time cost of recovery and
sources of finances. Results from the survey suggest that business interruption was
highly influential in terms of differential cost and time of recovery. It was not the
direct impact of flooding rather the under-researched and lesser-perceived
business interruption through indirect factors that were more significant for cost
and time of recovery. Furthermore, evidence of businesses relying highly on self-
finance was also apparent from the survey. Knowledge gained from the survey for
preparedness measures indicated that out of flood-affected samples that flood
experience is an important indicator of preparedness and mitigation actions. The
outcome of the research has highlighted some of the least researched phenomena
in the flood-affected business property sector and can demonstrate the need for
more widespread efforts to improve disaster recovery among businesses and a
novel input for future research.
Keywords: businesses, flood risk, damage, disruption, preparedness, recovery.
doi:10.2495/FRIAR140141
1 Introduction
Ensuring continuity of businesses in times of disaster is necessary for business
sector and it is necessary to synthesize prevention and protection measures in a
pre-disaster scenario in order to respond and recover faster during and after an
event and ensure continuous business operation [1] . Reduction of direct impact
among business enterprises require emergency relief services for cleaning up,
rebuilding and restoring properties. On the other hand, mitigation of indirect
effects demand financial assistance, employees’ return to job, suppliers and
consumer adjustment to the market, and essential service management. The
Committee on Disaster Research in Social Sciences has rightly suggested that
enterprises or businesses who are engaged in preparedness and mitigation
activities will be less vulnerable to natural disasters [2]. In theory insuring property
and businesses against flood damage can be treated as one of the effective tools of
mitigation; however, literature suggest that about 90% of the small and medium
enterprises (SME) in UK are under-insured [3]. Pitt’s report after the 2007 flood
event recommended the necessity of adoption of property level resistance and
resilience measures for all types of properties in the UK [4]. Research has
previously shown that business properties lack in such sources of protection
against impacts of flooding [5, 6] but such research has been limited in scale and
scope. Therefore, the main focus of this study is to further identify and investigate
patterns of preparedness and link this to trends of recovery using a case study
approach. The paper is structured in four sections. First, existing literature is
reviewed to gather impression of the flood risk and response situation in general
among flood plain population. Based on the rationale gained from literature review
methodology for specific case study areas were discussed and finalized. This is
followed by section on observed patterns of preparedness and recovery from the
selected case studies and finally, recommendations for future studies were
proposed before concluding remarks.
2 Review of flood response and recovery

The concept of response and recovery from disastrous event such as flooding
incorporates certain basic factors: knowledge of the risk; monitoring and warning
with ample time to respond; awareness and preparedness to cope with the impacts
and recover [7, 8]. First of all it is pertinent to identify the critical assets that are
exposed to risk and have higher vulnerability to decrease operational risk [9].
Apart from the direct protection of exposed assets, literature suggests that business
preparedness and response to disasters can also be affected by indirect factors such
as level of awareness regarding available protection measures and their long term
sustainability; anticipation of actual risk and perception of being secure; as well
as timely decision making of adaptation of risk reduction practices [10–12].
Table 1 lists some factors which are frequently associated with preparedness and
recovery in literature.
Businesses at risk of flooding in general show lack of preparedness that affects
their rate of recovery [5, 16, 27]. The issue of changing strategies towards risk
Table 1: Review of main factors associated with preparedness and recovery.
Factors Insights from literature Literature

references
Low preparedness Preparedness and precautionary measures [13–17]
and longer recovery among businesses are generally low (especially
SME’s). Firm characteristics play an important
role in preparedness.
Risk perception and Low level of preparedness as a result of low [18–20]
attitude perception of risk, inadequate
Lack of recognition of preparedness and
mitigation measures by affected population
Highly vulnerable Those at greatest risk adopt hazard adjustments. [1, 14, 21, 22]
properties are more This includes businesses with previous
prone to experience of disaster had engaged in more
preparedness preparedness and mitigation activities.
More concentration Businesses are more prone to prepare against [1, 14]
on direct damages direct damages than disruption to business
as preparatory operations.
measures
Risk Lack of risk communication can affect [23, 24]
communication preparedness and recovery; early response and
warning are pre-requisites
Financial capacity Financial incapability can be a big barrier to [25, 26]
preparedness and recovery. Investment in
disaster preparedness can reduce short term
profitability.
reduction through appropriate flood response and recovery for flood plain
population involves factors such as being kept well informed through media; early
response to warnings; consideration of warning dissemination time and evacuation
time from the building [23, 24]. To respond to indirect effects of flooding, it is
essential to recover and restore vital records (insurance papers, tax return
documents, tracing orders etc.). This is greatly facilitated through appropriate
preparation and backup in advance of flooding [1, 28] and such activities may be
specified through a continuity plan. Financial constraint can make the recovery
process take longer [26] therefore adequate insurance is indicated. .Without
insurance, larger enterprises have greater financial capacity to respond to flood
effects and therefore tend to recover faster from floods while smaller enterprises
might suffer more as a result of their financial constraints [25, 26]. Factors such
as reluctance of finance companies to supply loans for repair of the affected
property and high premiums set by insurance companies for flood prone properties
can prove to be fatal for the recovery process [24]. Such actions can have
catastrophic impacts on many businesses; one report suggests that around 43% of
the properties closed down after a disaster and about 29% of those closed down
within two years [29]. Based on the insights gained from literature, the following
section will detail the methodology adopted in collection of empirical data from
two selected case study locations to analyse the situation of preparedness and
recovery in flood-affected areas for business properties of flood plain population.
3 Methodology
It was necessary to identify areas for empirical data collection which have a
historical record of flooding and have a comparatively large population of
commercial properties at risk. Case study approach (although being
geographically limited) was appropriate for the purpose because of the scattered
nature of flood-affected properties and the lack of publicly available national data
sources with evidence of commercial properties affected by flooding in the past.
Therefore to increase the probability of tracing a comparatively large sample a
larger population at risk was selected through a systematic case study selection
approach.This is general consensus in literature that better prepared businesses
will fare well in case a disaster strikes [30–32]. The questionnaire survey enquired
about the level of preparedness among flood-affected sample population by asking
questions concerning type of mitigation and preparedness activities they are
engaged in. Number of preparedness and mitigation measures was provided in the
questionnaire with a range of activities to choose from. Enquiry was also done to
observe whether the preparedness measures were adopted before or after any flood
event.
3.1 Case study areas
The national assessment of flood risk in England states that the second area after
London at highest risk of flooding with largest number of people living at risk is
Yorkshire and Humber region [33]. Yorkshire and Humber region has a long
history of flooding and flooding in 2007 caused record breaking disruptions in the
area. It was mainly caused by heavy rainfall and river overflows. An Environment
Agency data report released in November 2007 showed that number of businesses
flooded in the region was 3718 which is the highest in the entire country [34].
Therefore, this area was selected as the area of interest for the research. The four
worst-affected locations were identified in the region: Sheffield, Hull, Doncaster
and Wakefield. In both Sheffield and Hull more than 1000 commercial properties
were affected as a result of 2007 flooding. Sheffield was chosen as one of the case
study areas because of the historical evidence of higher frequency of flooding in
the area than Hull which was one of the essential factors for sampling area
selection. In Doncaster not enough businesses were flooded and most of its
vulnerable areas were residential in nature as compared to Wakefield. Therefore,
Wakefield was chosen as the second case study location suitable for this study.
3.2 Survey approach
The primary unit of analysis for the research are commercial property occupiers.
There was no readily available data set of the members of this target population
that have been affected by direct or indirect sources of flooding from which a
sample population could be selected. Therefore a sample set was constructed from
a combination of different data sources. For example, available literature and flood
risk maps were relied upon to build a picture of the areas affected. It was therefore
difficult to determine the exact sample size relative to the target population since
determination of sample frame was based mainly on indirect sources. The

available information for the selected areas was historical flooding and
approximate number of commercial properties at risk or affected by flooding in a
particular event. Valuation office dataset was geographically projected to overlay
the sampled population on maps to determine their level of risk for particular
location. The sampling strategy employed was systematic sampling stratified by
flood risk category delineated by Environment Agency maps. A remote delivery
postal self-administered survey of 3660 occupiers of commercial buildings was
performed in all risk zones within the floodplain of two selected case study areas.
The variables selected for design of the survey instrument was based on the
conceptual framework generated and operationalized based on review of literature.
The questionnaire consisted of open and closed questions for different categories
of variables (such as flood damage, preparedness, sources of recovery, property
characteristics) were required to be measured for the analysis based on the
operational framework.
4 Results and discussion

4.1 Observed patterns of preparedness
Knowledge gained from the overall scenario indicated that out of the 69 flood-
affected responses 33 (48%) undertook some sort of preparatory measures and 36
(54%) did not engage in any of the given preparatory actions. Similarly, when the
type of preparedness measures implemented by prepared part of the sample
population were analysed it was apparent that they preferred easy to procure
temporary preparatory measures and fewer long term permanent solutions for risk
reduction. Other popular measures are Environment Agency flood warning and
property and business insurance (see Figure 1).
EA warning
15% 11% 13%
Other measures Property Insurance
Business disruption 9% 10% 10%
Business Insurance
plan
6% 5%
Emergency plan 8% Resilient fittings
0% 5%
8% 2% Temporary flood
Data backup 4% 15% installations
4% 4%
Alternative fuel Permanent flood
source installations
Alternative power Alternative location
source
Figure 1: Preparedness measures adopted by flood-affected respondents.
Temporary flood installations were largely adopted which were not adequate
for higher magnitude of flooding. More than half (total 55%) of business occupiers
who adopted any sort of preparatory measures took up only one or two measures,
19% restricted themselves with 3 to 4 measures and 26% were prepared for
flooding with more than 4 different combination of measures. This shows that
although taking measures for flood risk reduction is not very prevalent among
occupiers there is certain group of business occupiers who have started preparing
for the inevitable. Flood experience can be seen as having significant impact on
the level of adoption of protection measures. An interesting pattern was observed
based on the responses from the population who were flooded once and more than
once in the study areas (Table 2).
Table 2: Flood experience vs. preparedness level.
Flood experience Percentage prepared Percentage prepared

before flood
Flooded once 39% 26%
Flooded twice 88% 71%
Flooded more than twice 100% 60%
Total 62% 43%
One hundred percent (100%) of the people flooded more than twice have taken
up some preparatory measures, 88% of those flooded twice have at least one
measure, and 39% were prepared after only one event. The average number of
measures adopted by businesses did not show much variance based on the level of
experience. The range of number of adopted measures varied between 1.6 (flooded
once), 2 (flooded twice) and 1.8 (flooded more than twice). This is slightly
different outcome from the usual trend seen in the residential sector where it takes
more than two or three times for the flood-affected population to understand the
importance of mitigation [35]. In commercial sector it seems that those who decide
to undertake mitigation activities choose to do so in the light of fewer events.
4.2 Observed patterns of recovery
Respondents were asked to rank between 1 and 5 (1-no cost incurred and 5 highly
expensive) the different factors that affect cost of recovery. Table 3 illustrates the
percentage of differential cost incurred by respondents based on their differential
ranking.
Disruption of sales was scored highest while employee compensation and legal
charges were among the lowest ranked factors. Other factors like clean up charges,
machinery and sales disruption, working hour loss and repairing ranked among the
next four most costly factors in terms of recovery. Out of 100% of total cost
incurred 62% of the total cost was incurred for indirect flood impacts. Therefore,
it is evident that the cost incurred on recovering from indirect sources of damage
was more dominating than its counterpart. Answering questions regarding
financing sources for recovery the responses were clearly dominated by two
sources of finances; self- finance and property insurance. Table 3 indicates how
Table 3: Differential ranking of importance of factors affecting cost of

recovery.
Factors affecting cost of Ranking assigned as % of total

Rank
recovery cost of recovery
Sales disruption 13% 1

Clean-up charge 12% 2
Machinery repair 10% 3
Supply disruption 10% 3
Work hour loss 10% 3
Repair inside buildings 9% 4
Structural repair 8% 5
Vacant property charges 7% 6
Data back up 6% 7
Unrecoverable rent 6% 7
Employee compensation 5% 8
Legal charges 5% 8
businesses responded to questions associated with financing the process of

recovery. However the difference in their proportions clearly emphasize that
businesses are still more reliant on self-finance rather than insuring their
properties. This might be as a result of the general perception of risk among
businesses where impact of flooding is considered as temporary.
There was another funding source indicated in the questionnaire, for instance,
commercial loan but none of the respondents indicated that they have opted for
this measure. Apart from self-finance and insurance the other factors accounted
for only 5% indicating very low adaptation. In other words, more than 50% (51%
of self-finance and other) of the business losses are hidden in the sense that they
will not appear in official claims statistics from insurers and may not be recorded
anywhere else. This suggests that estimates of disaster impacts on business
communities may be rather lower than the true cost to businesses. The time taken
by the businesses to recover from the effects of flooding was distributed among
two categories, the short term recovery and long term recovery. Table 4
summarizes the short term and long term impacts on recovery from flooding. Less
than 30% of the flooded businesses indicated insignificant effects on their
businesses in short term and 23% in the long term; 38% of respondents indicated
that they were able to get back to business within 1–3 days in short term and 10%
in long term followed by 16% and 20% who took up to 7 days, 7% and 13% had
to suffer for up to 20 days and the rest 9% and 23% took longer to recover partially
from disruption.
Although people said that floods affected their businesses significantly many
of the respondents responded that they were fully recovered within a month or so.
Some businesses indicated that they can still feel the effects of flooding and never
recovered completely. This was around 10% of the flood-affected sample.
Therefore this might be possible that those businesses which could not recover
from the impacts of flooding were not represented in this data because they might
have closed or moved to another location. This is one of the drawbacks of self-
administered questionnaires to be fully explained, especially a questionnaire with
such great detail of information. It is interesting to notice that the preparedness
actions taken by businesses before flood event were mainly concentrated on
reducing direct damages, however data indicated that the impact of indirect effect
of flooding costs them more to recover. Therefore the insight gained from the
empirical analysis suggests that attention in reducing effects of indirect disruptions
and reducing impacts which originates offsite is also necessary.
Table 4: Sources of financing used by businesses for disaster recovery.
Sources of funding for recovery Percentage of total sources of financing %

Self-finance 51%
Insurance + self-finance 10%
Insurance 9%
Business reserve 3%
Business reserve + self-finance 3%
Commercial loan 0%
No preparedness/no response 25%
Table 5: Time taken by businesses to recover.
Short term (% of total time Long term (% of total time

Time for recovery
required) required)
Within 3 days 38% 10%
Immediately 30% 23%
Within a week 16% 20%
Two weeks 7% 13%
More than a month 6% 16%
Month 3% 7%
Year or more 0% 10%
The respondents were asked to rank between 1 and 5 (1 indicating recovered in

no time and 5 indicating the highest time taken to recover) the factors affecting
time of recovery. Cleaning up of properties (ranked highest in terms of
time consumption) and bringing customers back (2nd) are the most time consuming
factors that hinders businesses from operating well after disruption. Often the loss
of work hour (3rd) could be accommodated by working more, but this results in
payment of overtime and other inconveniences. For businesses which were
affected directly, clean up and drying could take months especially if they do not
have resilient fittings installed measures to protect the property from such effects
before the occurrence of the event; and therefore, this further worsens the situation
by losing more customers and work hour loss. Other factors such as repair inside
building and supply disruption, machinery repair, structural damage and recovery
services ranked 4th, 5th and 6th respectively.
5 Conclusion and recommendations

This paper presented survey based evidence of property occupiers’ experience on
impacts of flooding, patterns of preparedness and recovery in the two selected case
study areas in Wakefield and Sheffield. A comprehensive descriptive analysis
obtained through collection of data by use of self-administered postal
questionnaire from occupiers of business property in different flood risk categories
indicated the current situation of preparedness and recovery persisting among
flooded business communities at risk. Reflections obtained from this study
illustrate that damage and disruption pattern is more skewed towards indirect
factors. It was observed that in-spite of some level of preparedness among the
flood-affected population against direct impacts, there is considerable lack of
preparatory measures for indirect effects. There is a requirement of shift in
attention towards preparedness against business interruption. It is important to
focus on appropriate measures and efforts to adopt them in risk reduction process.
Relatively little attention has been paid to conduct assessment of effects of
flooding on properties with repeated flood experience. Based on the glimpse of
interesting result obtained for repeat flooded property in terms of preparedness it
is recommended that more research should be diverted towards deeper
understanding of business properties with previous experience of flooding.
References
[1] Y. Zhang, M. K. Lindell, and C. S. Prater, “Vulnerability of community
businesses to environmental disasters,” Disasters, vol. 33, no. 1, pp. 38–
57, 2009.
[2] Committee on Disaster Research and Social sciences, “Facing Hazards
and Disasters: Understanding Human Dimensions.” National Academy
Press., Washington D.C., 2006.
[3] K. Clemo, “Preparing for Climate Change: Insurance and Small
Business.,” Geneva Pap. Risk Insur. – Issues Pract., vol. 33, no. 1,
pp. 110–116, Jan. 2008.
[4] M. Pitt, “The Pitt Review – Learning Lessons from the 2007 floods,”
Cabinet office, London, 2008.
[5] G. Wedawatta, B. Ingirige and D. Proverbs, “Adaptation to flood risk: the
case of businesses in the UK,” in International conference on building
resilience, 2011.
[6] E. P. Evans, J. D. Simm, C. R. Thorne, N. W. Arnell, R. M. Ashley, T. M.
Hess, S. N. Lane, J. Morris, R. J. Nicholls, E. C. Penning-Rowsell, N. S.
Reynard, A. J. Saul, S. M. Tapsell, A. R. Watkinson, H. S. Wheater, “An
update of the foresight future flooding 2004 qualitative risk analysis,”
Cabinet Office, London, 2008.
[7] G. J. P. Strunz, K. Zosseder, S. Wegscheider, M. Mück, T. Riedlinger,

H. Mehl, S. Dech, J. Birkmann, and H. Harjono N. Gebert, H. Z. Anwar,
Sumaryono, R. M. Khomarudin and A. Muhari, “Tsunami risk assessment
in Indonesia.,” Nat. Hazards Earth Syst. Sci., vol. 11, pp. 67–82, 2011.
[8] R. Basher, “Global early warning systems for natural hazards: systematic
and people-centred,” Philos. T. Roy. Soc. A, vol. 181, no. 9, pp. 2167–
2182, 2006.
[9] Committee on Disaster Research and Social Sciences, The Impacts of
Natural Disasters. Washinton D.C.: National Academy Press, 1999.
[10] H. S. Wheater, “Flood hazard and management: a UK perspective,”
Philos. Trans. R. Soc. A Math. Phys. Eng. Sci., vol. 364, no. 1845,
pp. 2135–2145, 2006.
[11] G. Peters-Guarin, M. K. McCall and C. van Westen, “Coping strategies
and risk manageability: using participatory geographical information
systems to represent local knowledge,” Disasters, vol. 36, no. 1, pp. 1–27,
2012.
[12] M. Bleda and S. Shackley, “The dynamics of belief in climate change and
its risks in business organisations,” Ecol. Econ., vol. 66, no. 2–3, pp. 517–
532, 2008.
[13] H. Kreibich, M. Müller, A. H. Thieken and B. Merz, “Flood precaution of
companies and their ability to cope with the flood in August 2002,
Germany,” Water Resourses. Res., vol. 43, no. W03408, 2007.
[14] J. M. Dahlhamer and M. J. D’Souza, “Determinants of Business Disaster
preparedness in two U.S. Metropoliton areas,” Int. J. Mass Emerg.
Disasters, vol. 15, no. 2, pp. 265–281, 1997.
[15] B. Ingirige, D. Proverbs and G. Wedawatta, “Impact of flooding on SMEs
and their relevance to Chartered Surveyors,” RICS, London, 2012.
[16] D. Crichton, “Climate Change and its Effects on Small Businesses in the
UK,” AXA Insurance, London, 2006.
[17] A. Gissing, “Flood action plans – making loss reduction more effective in
the commercial sector,” Aust. J. Emerg. Manag., vol. 18, no. 3, pp. 1–9,
2003.
[18] F. Berkhout, J. Hertin and N. W. Arnell, “Business and Climate Change:
Measuring and Enhancing Adaptive Capacity,” Tyndall Centre, Technical
Report 11, 2004.
[19] F. Messner and V. Meyer, “Flood damage, vulnerability and risk
perception – challenges for flood damage research,” Discussion paper.
UFZ, Leipiz, 2005.
[20] B. Merz, H. Kreibich, R. Schwarze and A. Thieken, “Assessment of
economic flood damage,” Nat. Hazards Earth Syst. Sci., vol. 10, pp. 1697–
1724, 2010.
[21] T. E. Drabek, “Understanding disaster warning responses,” Soc. Sci. J.,
vol. 36, no. 3, pp. 515–523, 1999.
[22] D. Mileti, Designing Future Disasters: An Assessment and Bolder Course
for the Nation. Washington: Joseph Henry Press., 1999.
[23] P. D. Howe, “Hurricane preparedness as anticipatory adaptation: A case

study of community businesses,” Glob. Environ. Chang., vol. 21, no. 2,
pp. 711–720, May 2011.
[24] A. Jha, J. Lamond, R. Bloch, N. Bhattacharya, N. Papachristodoulou,
A. Bird, D. Proverbs, J. Davies and R. Baker, Cities and Flooding: A guide
to integrated urban flood risk management for 21st Century. Washinton
DC: GFDRR, World Bank, 2012.
[25] K. J. Tierney, “Business Impacts of the Northridge Earthquake,”
J. Contingencies Cris. Manag., vol. 5, no. 2, pp. 87–97, 1997.
[26] M. Aba-Bulgu and S.M.N. Islam, Corporate crisis and risk management:
modelling strategies and SME application. Oxford: Elsevier Ltd., 2007.
[27] I. Seifert, H. Kreibich, B. Merz and A. H. Thieken, “Development of
FLEMOcs-a new model for the estimation of flood losses in the
commercial sector,” Hydrol. Sci., vol. 55, no. 8, pp. 1302–1314, 2010.
[28] N. Bhattacharya and J. Lamond, “Flood Vulnerability and Hazard
Adjustment of UK commercial sector : A synthesis of research,” 2011.
[29] D. Wenk, “Is ‘Good Enough’ storage Good Enough for compliance?,”
Disaster Recover. J., vol. 17, no. 1, pp. 1–3, 2004.
[30] K. J. Tierney and R. G. Webb, “Business Vulnerability to earthquakes and
other disasters,” University of Delaware, DRC Preliminary papers, 2001.
[31] R. Raaijmakers, J. Krywkow and A. van der Veen, “Flood risk perceptions
and spatial multi-criteria analysis: an exploratory research for hazard
mitigation,” Nat. Hazards, vol. 46, pp. 307–322, 2008.
[32] G. R. Webb, K. J. Tierney and J. M. Dahlhamer, “Predicting long-term
business recovery from disasters: A comparison of the Loma Prieta
earthquake and Hurricane Andrew,” Environ. Hazards, vol. 4, no. 1,
pp. 45–58, 2002.
[33] Environment Agency, “Flooding in England : A national assessment of
flood risk.” Bristol, 2009.
[34] Common Recognised Information Programme 45, “2007 Summer
floods,” 2007.
[35] C. Rose, D. G. Proverbs, C. Booth, C. and K. I. Manktelow, “Three times
is enemy action – flood experience and flood perception,” in Flood
Recovery, Innovation and Response III, 2012.
Resilient reinstatement: what can we learn

from the 2007 flooding in England?
R. Joseph1,2, D. Proverbs1 & J. Lamond1
1
2
Cunningham Lindsey, UK
Abstract
In the face of increased flooding in the UK, it is becoming increasingly
important to understand the ways in which flood experience can affect
homeowners’ attitude towards taking precautionary measures to protect their
homes. This could include investing in flood adaptation measures to reduce
likely flood damage and hence exposure to flood risk. This research sought to
investigate, the level of awareness, implementation and the costs of resilience
measures, from those homeowners who had experienced flood damage to their
properties in 2007 summer flooding in England. A questionnaire survey was thus
employed to elicit the extent to which flood experience influenced the decision
to adopt flood resilient measures during reinstatement works. The findings
revealed that some 82% of houses inundated were returned to their pre-incident
condition i.e. with no improved resilience to future flooding. It was found that
the level of awareness of resilience measures among the respondents was high;
however, the level of implementation was quite low. Only 10% of those who
indicated that they invested resilience measure actually implemented a full
package of the measures. This shows that more needs to be done by flood risk
management stakeholders to encourage full uptake of resilience measures. Loss
adjusters and surveyors are better placed to advice homeowners of the potential
risk reduction measures, which can be implemented during reinstatement period.
Further, there is a need for policy development in the form of revising the current
Building Regulations for refurbishing or reinstating flood damaged buildings in
order to encourage the up-take of resilient reinstatement.
Keywords: flood adaptation, flood damage, flood experience, resilience
measures.
doi:10.2495/FRIAR140151
1 Introduction
The cost of flood damage in the UK has risen significantly since 1998 [1].
Currently, in the UK, over 5.2 million properties and 2.4 million people are at
risk of flooding, and annual average damages are estimated to be more than
£1 billion [2]. However, climate change and the increasing urbanisation of our
societies are increasing flood risk [2, 3]. In particular, there now appears to be
clear evidence that climate change will lead to an increase in the frequency and
severity of extreme precipitation and other weather events [4]; for the UK, this
may well result in wetter and stormier winters [5]. As such, The “Foresight
Future Flooding” report raises the prospect of a 4–10-fold increase in coastal
flood risk by the 2080s as a result of sea level rise alone [2]. The UK
Government policy on flood management can be summed up by the strategy of
“Making space for Water” which combines the provision and maintenance of
engineered flood defences with the recognition that flooding can never be
prevented entirely [6].
The direct financial damages related to the flooding of residential properties
can be significant. Depending on flood depth, duration of flooding and property
types, it is estimated that the cost of flooding can range from £15,000 to over
£80,000 for a single residential property and its contents [7]. The impact of
flooding at an individual household level can also result in less direct,
insurance-related impacts [8, 9], with premiums and flood-related excesses
potentially increasing following a flood event and as a result of making
insurance claims [10, 11].
Whilst large scale flood defences can be effective in reducing widespread
flood risk, such developments are costly, both in terms of time and financial
resources. Consequently, cost benefit analysis does not always yield a favourable
result for large scale defence schemes, and the extensive flooding that has
recently occurred within the UK has strengthened calls for greater use of
adaptation measures [12, 13]. In the UK, such measures are generally classified
as resistance (measures to keep water out of properties) or resilience measures
(installed to reduce the damaged impact of flooding on the fabric of building)
[14]. Keeping water out is a natural desire of property owners but, it is not
always possible or cost effective to prevent flooding of property, especially when
the anticipated flood depth is up to 1000 mm. Resilience measures are often
preferred, allowing water into the property in the knowledge that preparations
have been taken to minimise the damage caused. These adaptation measures are
designed to achieve two important objectives: to limit the financial impact on the
flood victim or their insurer by reducing damage to contents and building fabric
and to reduce the time used to reinstate properties, thereby, allowing
communities to return to normality quickly in the aftermath of the flood event.
Research has shown that implementing adaptation measures during flood
recovery period can effectively reduce the cost of the measures [15].
Review of extant literature revealed that the uptake of resilience measures in
residential properties remains persistently low [15], with one study finding that
only 16% of households and 32% of small-medium enterprises (SMEs) in areas
of significant flood risk have taken practical steps to reduce their exposure to the
potential flood risk [16]. Common reasons for the low uptake of the measures
include underestimation of flood risk, a lack of understanding about flood
protection responsibilities and concerns over the costs and aesthetics of such
measures [17, 18].
This study explores, the level of awareness, implementation and the costs of
resilient reinstatement after 2007 summer flooding in England. Those
homeowners who had experienced flood damage to their properties were the
focus of the study. The 2007 summer flood event provides an interesting case
study as it was reported to be widespread, and the cost of reinstatement work
during the recovery process was the highest insurers had ever paid (prior
to 2007) in England on a single flood event. Adapting existing properties to
potential future flood risk can be achieved by investing either in resistance or
resilience measures, or the combination of the two measures. The focus of this
study is on resilience measures. The concept of resilient reinstatement and
specifications, which can be incorporated during the flood reinstatement process
and the costs of resilient reinstatement are discussed in this paper. The
concluding part of the paper outlines some of the lessons learned from the level
of awareness, implementation and the actual cost spent by those homeowners
who implemented one form of resilience measures as a result of reinstatement
work to their properties following the 2007 summer flood event.
2 Concept of resilient reinstatement

Flood water can enter buildings swiftly, causing pervasive damage to floors,
walls, finishes and services, and in more severe floods the flood water can cause
structural damage [16, 19]. The vulnerability of buildings depends on the
construction methods and building materials used in its construction. The
processes and pathways by which water enters a building during a flood depends
on the characteristics of the flood, specifically flood depth and duration, and
water velocity [20]. Nevertheless, for floods deeper than 1000 mm, it is
recommended that no attempt should be made to keep the water out of the house,
because the build up of water pressure could cause external walls to become
unstable, leading to serious structural damage [20]. It has been suggested that if a
property is vulnerable to repeated flooding, it is important to limit damage to
speed up drying/re-occupation by making the inside of the property more
resilient to floodwater.
2.1 Flood resilient measures/specifications
Due to the additional cost involved in implementing resilience measures, they

are generally recommended for buildings with exceptionally high risk of
flooding. Materials such as water-resistant paints and coatings, for example, can
prevent floodwater soaking into the external face of the walls. Other materials
such as lime-based plaster, as opposed to gypsum plaster have good
water-resilient properties and dry out quickly. Solid concrete floors can also
prevent water seeping into the fabric of a building. Other measures include re-
fitting electrical sockets and electricity meter boxes above the anticipated flood
levels. Despite the extra cost of these measures, it has been suggested that the
implementation of resilient measures will reduce the repair costs in the long-term
assuming repeat flooding [16]. Table 1, shows the most widely used and
recognised resilience measures/specifications.
Table 1: List of resilience specifications.
Resilience measures
Replace timber floors with concrete and cover with tiles.
Replace carpet with ceramic tiles.
Replace chipboard/MDF kitchen and bathroom units with plastic equivalents or
stainless steel.
Replace gypsum plaster with more water-resistant material, such as lime plaster or
cement sand render.
Apply water resistant paint to walls.
Move service meters, boiler, and electrical points well above likely flood level.
Replace softwood timber skirting with plastic or hardwood and apply water resilience
paint.
Replace softwood door and window frames with water resilient alternative.
Replace mineral insulation with cell insulation.
Source: Joseph [7].
The effectiveness of such resilient measures is dependent on the expected

volume and duration of the flood water and it has been established that in some
cases these measures are not always cost effective [16, 19], therefore proper
flood risk assessment should be carried out before investing in resilient
reinstatement. Conversely, there are resilience measures for buildings that are
inexpensive, especially if implemented during other building works [19] or may
be cost neutral, for example setting electrical sockets further up the wall where
the electricity supply is dropped down from the ceiling [14].
2.2 Cost of resilient reinstatement
Previous research carried out on behalf of the ABI [15], revealed that, on
average, resilient reinstatement costs over 40% (£12,000) more than traditional
reinstatement. It was stressed that there are significant variations around this
40% average, both between house types (i.e. bungalow, block of flats, terraced,
semi-detached and detached houses) and within house types. Although, the
authors further reiterate that resilient reinstatement could costs as little as 15% or
as much as 70% more than traditional reinstatement [15]. The reasons for the
wide variation were; property owners’ individual preferences and different
approaches to reinstatement methods adopted by different surveyors, despite the
available guidance such as Proverbs and Soetanto [21]; Garvin et al. [22] and
PAS 64 [23]. Some resilient measures can be introduced on a cost neutral basis,
and therefore not all aspects of resilient reinstatement measures increase the cost
of reinstatement. According to the economic modelling study which was

conducted on behalf of Department for Environment, Food and Rural Affairs
(DEFRA) and Environment Agency (EA) resilience measures are most cost
effective when conducted as part of a programme of resilient repair following a
flood [16].
Table 2 shows the additional cost of resilience measures, these costs are the
extra over cost incurred during reinstatement of flood damaged building to make
those properties flood resilience against future flooding. The cost ranges from
as low as £12,000 for a terraced house flooded to a depth of 150 mm and as high
as £28,300 for a bungalow flooded to a depth of 1000 mm. The additional cost of
resilience measures presented in Table 2 was based on resilience specifications
presented in Table 1. Understandably, as the depth of the floodwater increases,
so does the cost of resilience measures, therefore, accuracy of expected future
flood depth is important when estimating the cost of resilience measures. Getting
this wrong may invalidate the resilience measures which were taken, thereby
leading to waste of money already spent on implementing resilience measures.
Table 2: Costs of resistance and resilience measures for different building

types, flood depths and deployment methods.
Cost of resilience measures (CMrt) in flood depth (mm)

Building Types categories
0–150 151–300 301–500 501–1000 > 1000
Bungalow £15,200 £16,200 £20,395 £28,300
Detached £13,300 £14,600 £16,700 £23,700 £24,800
Semi-detached £12,500 £13,600 £15,800 £15,000 £22,600
Terraced £12,000 £15,300 £16,800 £15,400 £20,200
Source: Joseph [7].
3 Research methodology
An extensive survey was undertaken among those homeowners, who
experienced flood damage to their properties in the summer 2007 flood event, in
order to gain a better understanding of their flood experiences; to investigate
their understanding of resilience reinstatement; and to examine their
responsiveness to resilient reinstatement while their properties were being
repaired. This contributed to the evidence base needed to inform the effective
promotion of resilient reinstatement during flood recovery period. The
investigation took the form of postal questionnaire surveys. The mix of
the targeted population, which comprises of young and elderly people, dictates
the postal approach instead of online method of questionnaire distribution. The
summer 2007 flood event in England was selected as the focus of the study. This
flood event was widespread and it affected much of the UK during June and July
2007 which followed the wettest-ever May since national records began in 1766
[12]. The survey was carried out in 2013 some 6 years after the event and was
designed to gather information in two key areas;
1. Flood experience (previous and subsequent flood experiences).

2. Level of awareness and implementation of resilience measures.
Prior to distributing the questionnaire to the main respondents, a pilot survey
was conducted among homeowners who were not part of the main survey to
determine the suitability of the questionnaire format and the contents, before
being distributed to the targeted population. The feedback received from the pilot
survey showed that the questions were easy to understand, therefore, it was
decided that the main questionnaire survey could proceed. Figure 1, shows the
survey location, which comprises of cities in the North and South of England.
The survey locations were selected from amongst the locations flooded during
the 2007 flood event. The selection criteria was based on the need to represent
the widest possible variation both geographical and flood typology while
retaining minimum numbers of properties within each selected site. To that end
only sites with greater than 50 affected properties were included in the survey. In
total, 2309 questionnaires were distributed via post to homeowners. The survey
yielded 280 responses, representing a response rate of 12.1%, which is
considered a reasonable return for an unsolicited postal survey.
Figure 1: Survey site locations.
4 Research results
Detailed analysis of the dataset was carried out and is presented in this section.
Respondents were asked if they had experienced flood damage to their properties
before and after 2007. Most respondents (77%) had no previous flood experience
prior to the 2007 flood event; 16% reported that they had experienced one
previous damaging flood to their properties prior to the 2007 event; and
approximately 4% had been flooded twice and 3% had been flooded more than
twice. This information is important because it is anticipated that those
respondents who had been flooded more than once, are more likely to invest in
resilience measures. Research has shown that experience of flooding can be a
source of motivation to individuals to undertake precautionary measures against
future flooding [24]. Further, respondents were asked if they had experienced
further flood damage to their properties following the 2007 flood event. Some
91% of respondents did not experience a flood event after the 2007 summer
flood event. This means that only 9% of respondents had experienced further
flooding after 2007.
4.1 Level of awareness of resilience measures
Figure 2 illustrates the analysis of the level of awareness of different types of

resilience measures, which homeowners can implement during flood recovery
period. The result shows that, the level of awareness ranges from 11% to 61%.
Some 61% of respondents are aware of replacing floor carpet with tiles as one
form of resilience measure. In total, 60% of respondents are aware of replacing
suspended timber floor with concrete floor as one form of resilience measure.
Over half of the respondents (51%) are aware of raising electrical socket above
the anticipated flood level, as one form of resilience measures. These results
differ from earlier UK studies, which suggest a lower level of awareness of
resilience measures [14]. Majority of respondents were unaware of replacing
mineral insulation with cell insulation (89%) and using plastic (85%) or stainless
steel (81%) kitchen units instead of MDF boards as form of resilience measures.
The low level of awareness of these measures can be linked to the fact that,
these measures are not readily available. The use of stainless kitchen units is
synonymous to commercial kitchens, and the plastic kitchen units are not
currently readily available in the building construction market. It can be inferred
from these results, that, the majority of the respondents are aware of the most
commonly used resilient measures.
4.2 Level of implementation of resilience measures
Despite the relatively high number of respondents being aware of (at least) one
form of resilience measure to protect their property, the results presented in
Figure 3, show that fewer people actually used the opportunity of the 2007 flood
Cell insulation 11% 89%

Stainless steel kitchen units 18% 81%
Plastic kitchen units 15% 85%
Upvc doors 46% 54%
Aware
Plastic skirting 25% 75% (%)
Water resistant paint 25% 75%
Tanking 19% 81% Not
Aware
Gas meter above flood line 44% 56% (%)
Electrical socket above flood line 51% 49%
Water resistant plaster 31% 69%
Floor tile 61% 39%
Concrete floor 60% 40%
0% 20% 40% 60% 80% 100%

Percentage (%) of responses
Figure 2: Percentages of respondents who are ‘aware and not aware’

of resilience measures.
event to invest in resilience measures. Among those who were aware

of resilience measure of using plastic or hardwood skirting board instead of
softwood timber skirting, 23% actually invested in plastic or hardwood skirting.
The relatively high percentage of people who invested in this resilience measures
can be linked to the fact that, the cost increase from softwood skirting to plastic
skirting is very low [15]. In some cases, the cost may be incorporated in the total
cost of reinstatement, which means the insurer may have paid for it
unknowingly. Some 14% and 18% of respondents decided to raise gas and
electric meters and electrical sockets above the anticipated flood levels
respectively. These resilience measures are normally cost neutral, if implemented
during the reinstatement process.
The fact that not all respondents who indicated that they were aware of these
forms of resilience measures actually implemented the measures shows that,
apart from awareness, there are other barriers, such as aesthetic considerations
and emotional attachment to the existing layout of fittings and features this
accords with previous study [18], that barrier to uptake of resilience
reinstatement is not only hinges on financial constraints.
Cell insulation 6%
Stainless steel kitchen units 5%
Plastic kitchen units 4%
Upvc doors 11% Series1
Plastic skirting 23%
Water resistant paint 25%
Tanking 2%
Gas meter above flood line 14%
Electrical socket above flood line 18%
Water resistant plaster 12%
Floor tile 20%
Concrete floor 25%
0% 5% 10% 15% 20% 25% 30%
Percentage (%) implemented
Figure 3: Distribution of respondents who had implemented one form of

resilience measures.
4.3 Analysis of costs invested by respondents in resilience measures
Research has shown that implementing resilience measures during flood

reinstatement process is less expensive, because the contractor’s site set up cost
would have been paid by the insurer as part of the normal insurance
reinstatement work. Thus, the homeowner would only be required to pay the
extra cost of resilience measures, for instance, when a suspended timber floor is
to be replaced with concrete floor, the insurer would paid for the cost of
replacing the timber floor including the cost of preliminaries, however, the
homeowner would be required to pay the difference between the cost of timber
and concrete floors, excluding any preliminaries costs.
Analysis of the cost invested by those respondents who implemented at least
one form of resilience measure was carried out. The total amount homeowners
invested in resilience measures during the flood reinstatement process ranged
from £1000 to £45,000. With 49% of respondents investing up to £1000
(equivalent to extra over cost of replacing softwood skirting board with plastic or
hardwood). 16% of those who implemented resilience measures, invested up to
£3000, whilst only 2% invested up to £45,000.
The overall median, which homeowners invested in resilience measures, was
£1,500. This figure is lower when compared to the earlier research such as
[13, 15, 16], which suggest a higher value of £12,000 minimum for full package
of resilience measure. This indicates that among those respondents who had
implemented resilience measures, full package of resilience measures were not

implemented. 10% of those who invested in the measures actually implemented
full package of the resilience measure, such as replacing timber floor with
concrete, raising electrical socket and gas meters above the anticipated flood
level.
5 Discussion and conclusions

The aim of governments to place more of the responsibility for flood
management onto the floodplain population requires the floodplain population to
take action to reduce the impact of flooding on their properties. Resilience
measures have a place in the hierarchy of flood risk management solutions for
existing properties. However, they are generally regarded as the last resort for
locations and situations where no other measure, such as large scale flood
defences, can be provided. The findings reported herein are part of a research
into the development of a comprehensive costs and benefits of property level
flood risk adaptation measures in England. In addition to broadly confirming the
findings of earlier studies into the level of awareness and take-up of resilience
measures, the findings from this research have shed some light onto some of the
key issues surrounding the uptake of resilience measure, especially among those
that have been flooded before. Five (5) key lessons revolving round the level of
awareness and implementation of resilience measures were learned from the
output of this research, these are summarised at the end of this section, under
the heading ‘key lessons learned’.
The emergence of effective public awareness and engagement campaigns by
organisations such as Environment Agency; Department for Environment, Food
and Rural Affairs (DEFRA); National Flood Forum (NFF); and Association of
British Insurers (ABI) seem to have led to an increase in awareness of resilience
measures, amongst the respondents. However, the fact that majority of the
respondents are aware of one form of resilience measures, did not result in
increase uptake of the measures. This research shows that some 82% of
properties did not adopt any form of resilient reinstatement. Of those who did
implement resilience measures in the reinstatement process, a vast majority
(90%) failed to implement a full range of resilient measures. The reasons why
partial resilience measures were taken by those people were generally unknown.
Further research to investigate this is therefore recommended.
For effective flood risk management strategies, understanding the reasons
why some homeowners did not take up resilience reinstatement during flood
recovery period in 2007 is important, as this is a key step in developing strategies
to increase the uptake of resilience measures. In order to embrace the principle of
resilience reinstatement, loss adjusters and surveyors, often the link between the
insurer and homeowners are better placed to advise their clients (homeowners
and insurers) of the potential risk reduction measures, which can be implemented
during reinstatement period, most especially, those measures that are cost neutral
if implemented during reinstatement period.
The last resort is that, if the level of uptake of resilience reinstatement after
flood event continues to be as low as what was revealed in this study, it is
recommended that, Government needs to revise the current Building Regulations
for refurbishing or reinstating flood damaged buildings in order to force the
up-take of resilient reinstatement after flood event.
Key Lessons Learned:
 The level of awareness of resilience measures among the respondents is

relatively higher than reported in previous research.
 Despite the high level of awareness, a majority of respondents did not
invest in resilience measures while their properties were being repaired in
2007.
 Among those respondents who invested in the measures, only 1.8% of the
whole sample can be said to have implemented a full package of
resilience measures.
 By not implementing a full package of resilience measures, the full
benefits of resilience measures cannot be achieved.
 There is still a need to encourage the implementation of resilience
measures during flood recovery period, perhaps, by updating the current
Building Regulations to force the implementation of resilience measures.
References
[1] Association of British Insurers. Summer floods 2007: Learning the
lessons. London. Association of British Insurers. London, 2007.
[2] Evans, E. P., Ashley, R., Hall, J., Penning-Rowsell, E., Sayers, P., Thorne,
C. R. & Watkinson, A. Foresight. Future Flooding. 1 and 2 London, 2004.
[3] Office of Science and Technology Climate Change Science. Postnote No.
295. Parliamentary Office of Science and Technology, London, 2007.
[4] IPCC Managing the risks of extreme events and disasters to advance
climate change adaptation. A special report of working groups I and II of
the intergovernmental panel on climate change. Cambridge University
Press, Cambridge, 2012.
[5] UKCIP. UK Climate Projections. UK Climate Impacts Programme,
DEFRA. http://www.defra.gov.uk/publications/files/pb13274-uk-climate-
projections-090617.pdf, 2009.
[6] Rooke, D. The summer of storm. Water and Environment Magazine, (10),
pp. 8-9, 2007.
[7] Joseph, R. Development of a comprehensive quantification of the costs
and benefits of property level flood risk adaptation measures in England.
PhD Thesis. University of the West of England, Bristol, 2014.
[8] Ball T., Geddes A., Werritty A., Black A. & Easton A. Flood insurance
provision and affordability beyond the statement of principles: implication
for Scotland. CREW, University of Dundee, 2012.
[9] Association of British Insurers. Revised Statement of Principles on the

Provision of Flood Insurance. ABI, London, 2008.
[10] Association of British Insurers. Under-pricing of the flood element of
home insurance for domestic customers at significant flood risk. ABI,
London, 2011a.
[11] O’Neill M. & O’Neill J. Social Justice and the Future of Flood Insurance.
Joseph Rowntree Foundation, 2012.
[12] Pitt, M. The Pitt review – Learning lessons from the 2007 floods. H.M.
Government Cabinet Office, 2008.
[13] Joseph, R., Proverbs, D., Lamond, J. & Wassell, P. An analysis of the
costs of resilient reinstatement of flood affected properties: A case study
of the 2009 flood event in Cockermouth. Structural Survey, 9(4),
pp. 279-293, 2011.
[14] DEFRA. Consultation on policy options for promoting property-level
flood protection and resilience. DEFRA Report, London, 2008.
[15] Wassell, P., Ayton-Robinson, R., Robinson, D., Joseph, R., Hack, K.,
Butler, D., Salkeld, I. & Twomey, J. Resilient Reinstatement: The costs of
flood resilient reinstatement of domestic properties. Association of British
Insurers, 2009.
[16] Thurston, N., Finlinson, B., Breakspear, R., Williams, N., Shaw, J.
& Chatterton, J. Developing the Evidence Base for Flood Resistance and
Resilience. Joint DEFRA/EA Flood and Coastal Erosion Risk
Management R&D. DEFRA. London, 2008.
[17] Werritty A., Houston D., Ball T., Tavendale A. & Black A. Exploring the
social impacts of flood risk and flooding in Scotland. Report to the
Scottish Executive, 2007.
[18] Proverbs, D. & Lamond, J. The barriers to resilient reinstatement of flood
damaged homes. Proceedings of 4th International i-Rec Conference-
Building Resilience: Achieving Effective Post-disaster Reconstruction.
Christchurch, New Zealand, 2008.
[19] Lamond, J.E. & Proverbs, D.G. Resilience to Flooding: Lessons from
International Comparison. Urban Design and Planning, 162(2), pp. 63-70,
2009.
[20] Samwinga, V., Proverbs, D. & Homan, J. Exploring the Experience of UK
Homeowners in Flood Disasters. International Construction Research
Conference. Royal Institute of Chartered Surveyors, 2004.
[21] Proverbs, D. & Soetanto, R. Flood Damaged Property: A guide to repair.
Blackwell, Oxford, 2004.
[22] Garvin, S., Reid, J. & Scott, M. Standards for the Repair of Buildings
following Flooding. CIRIA: London, 2005.
[23] PAS 64. Mitigation and Recovery of Water Damaged Buildings – Code of
Practice. London: British Standard Institution, 2013.
[24] Kreibich H., Seifert I., Thieken, A. H., Lindquist E., Wagner K. & Merz
B. Recent changes in flood preparedness of private households and
businesses in Germany. Regional Environmental Change 11, 59-71, 2011.
The role of flood memory in the impact of

repeat flooding on mental health
J. Lamond
Abstract
A highly important but under researched impact of flood events is the long term
psychological effect of the distress and trauma caused by damage and losses
associated with repeated flooding of communities. As a part of the recovery
process responders need to consider flooded households and offer support to
mitigate against the stress of flooding. This research aims to consider how the
risk of repeat flooding and flood memory can affect the needs of communities
with respect to post disaster support. Previous research has identified a variety of
influencing factors that affect the prevalence of mental health disorders in the
aftermath of flooding. Using a structured literature review and novel conceptual
model this research examines the role of flood experience and memory in the
impact of flooding on mental health and the needs of flooded communities. It is
found that the memory of previous flooding can influence future outcomes in a
variety of ways, with some positive incentives towards actions that may result in
lower damages in future events. These actions, that affect future trauma, have the
potential to mitigate the impact of repeated flooding. Therefore appropriate post
disaster needs assessment should not only identify vulnerable individuals but
also take account of the risk of future flooding.
Keywords: flood memory, PTSD, Anxiety, frequent flooding, flood impact,
mental health, flood recovery.
1 Introduction
The impact of flooding on the physical and mental health and wellbeing of
communities can endure long after the loss and damage due to direct contact
with floodwater is repaired. Quite apart from the possible loss of life and
irreversible injury; studies have demonstrated that a variety of other physical
doi:10.2495/FRIAR140161
ailments and mental health issues can arise in the aftermath of a flood [1]. It is
clear that, for the UK, Europe and the majority of the developed world, the
mental health impacts of flooding are at least as important as the risks from
physical illness [1, 2]. However, the detailed level of understanding needed by
responders and agencies in order to provide appropriate support throughout the
disaster cycle is lacking [3–5]. Furthermore it is not entirely clear how the
resources available to offset the longer term effects of flooding on the
psychological resilience of individuals and communities should be directed.
Research has explored various factors that can influence the severity of
mental health impacts of flooding including flood characteristics [6, 7];
individual characteristics [8]; socio economic factors [6]; preparedness [9]; and
duration of reinstatement activities [10]. It also seems intuitive to suggest that
flood memory or experience of past flooding will have bearing on the severity of
mental health impacts. However research in this area is lacking and it has been
identified by the UK Health Protection Agency as an area in need of further
research [11].
Therefore this paper seeks to explore various factors, including flood memory
on the mental health and wellbeing of flood affected communities and
individuals. The eventual aim is to improve the understanding of mental health
consequences from repeated flood experience leading to improved provision of
support services and targeting of resources to those potentially most vulnerable
to future mental health problems as a result of flooding and flood risk [5].
2 Methodology
The research adopted an enquiry based qualitative approach through a structured
review of available literature on the basis of research questions designed to
answer the main research aim. Literature from the wider field of disaster
management was combined with flood specific research in order to address the
following research questions:
1. What are the main mental health problems caused by flooding?
2. What factors affect the prevalence and severity of mental health issues
in flooded communities?
3. How long does the impact of flooding on mental health endure and does
the memory of flooding affect mental health issues following flooding?
A keyword search of academic literature databases provided the majority of the
literature, recent publications were prioritised and the presence of several
overarching reviews was capitalised upon in order to optimise the coverage of
older and diverse literature. Over 80 studies were accessed directly but the pool
of background studies was far larger because of many wide ranging reviews.
A novel conceptual framework was then developed based on the available
evidence on the research questions. This illustrates the influencing factors, role
of memory and mitigating interventions in the context of communities at risk of
frequent and repeated flood events. The construct validity of the proposed
framework derives from the thorough nature of the qualitative enquiry.
3 Research results
Studies investigating the impact of flooding on mental health span the disciplines
of Flood risk management, disaster management, public health, epidemiology,
environmental management, climate change and more. Findings from this
diverse knowledge base are structured below as they relate to the three research
questions. However the emphasis is on identifying lessons for appropriate needs
assessment rather than examining appropriate clinical diagnosis or treatment.
3.1 Mental health impacts caused by flooding
Flooding can be regarded as a stressful and sometimes traumatic experience and

some psychological reaction is therefore expected and natural [4] and much of
this may quickly dissipate. Where this is not the case, Post traumatic stress
disorder (PTSD) is the most commonly reported side effect of natural disasters
including flooding and symptoms of depression and anxiety are also frequently
seen [4]. For example Norris et al. [12] reviewed multiple disaster studies,
concluding that experiences ranged from inconvenience to severe trauma and
that the reaction to those experiences ranged from severe mental health
deterioration to some positive developmental aspects. It is apparent that some
individuals suffer from more than one mental health issue. For example Norris et
al. [13] observed both PTSD and Mild depressive disorder (MDD) in
populations affected by floods in Mexico and found that co-morbidity was
substantial.
Estimated prevalence of mental health disorders varies widely. Alderman et
al. [14] collated literature estimating the prevalence of mental health disorders
ranging from 8.6% to 53% in the first two years following flooding. Some of this
variability may be due to sampling and methodological differences in estimation
methods. Measurement of the impact of flooding in prompting mental health
issues is complicated by the underlying level of psychiatric disorders already
present in the population. Notwithstanding, studies imply that the uplift in need
after a flood is substantial and variable. For example, in Lewes after the 2000
event, Reacher et al. [15] found a four-fold increase in psychological distress in
flooded households when compared with non-flooded. Given the wide disparity
in observed occurrence of PTSD and other psychological issues, the research
clearly demonstrates that some unpicking of influencing factors would be critical
in identifying vulnerable communities and individuals and in directing support.
3.2 Factors affecting the development of mental health disorder
Models of health impacts in the literature include Few [16] and Tapsell et al. [5].
They suggest a list of factors that make a difference to the prevalence of mental
health issues post disaster. These can be grouped into pre-existing conditions,
impact of the stressor event and post event conditions and stresses [9]. However
the scale and direction of the influencing factors are not consistent across studies.
There are some confirmatory and strong results such as the conclusion that
low socio economic status relates to higher level of distress [17]. However, other
factors such as age display much more complex and conflicted relationships with
mental health problems. Furthermore the risk of development of severe mental
health issues, or PTSD, was found to be related to individuals with extreme
pre-existing conditions and the presence of extreme aspects in the stressor event.
So for examples survivors who may have a higher than typical risk for PTSD
include those with a history of trauma exposure; chronic illness; chronic social
problems; or other major life stressors such as single parenting [5]. Table 1
summarises the influencing factors identified in the literature review.
Table 1: Factors influencing the likelihood of experiencing mental health

disorder after flooding.
Pre existing conditions Features of the Post event stress and

stressor event coping strategies
History of psychiatric Severity of exposure Presence of other
disorder stressors
History of other health Perception of human Lack of resources for
related problems control or responsibility recovery
for the event
Gender General scale of loss of Distress of others,
life or massive injury particularly spouse
Disaster experience or Level of personal Living with the threat
training property of constant or growing
flood threat
Age Social support
Ethnicity Religion
Socio economic status Family structure
Dependent children Coping strategies
Urban/rural setting Need for relocation
Personality factors
Vulnerability is related to gender. Tapsell et al. [5] observed that women and
girls exhibited stronger effects than men and boys in 42 out of 45 studies. The
effects were most marked in the study of PTSD and within traditional cultures.
The presence of strong spousal support in mitigating stress was also less helpful
for women than men with women apparently burdened by close social ties [5].
The effect of age as an influencing factor is more complex. While it is clear
that age contributes to physical challenges that result in increased physical health
impacts, injury and mortality [18], conflicting results are reported in mental
health studies in the disaster field [19]. For example, in Korea younger people
(under 45) were found to have most symptoms after a flood event [20] and other
authors have found similar patterns [21] possibly due to the older generation’s
coping strategies [22]. However the protective influence of older age was not
observed in Vietnam [23]. Some studies have demonstrated that middle aged
adults are the most prone to mental health problems after disasters [12]. Tapsell
et al. [5] suggest an explanation for this is that flooding adds to the greater
responsibilities they already face. Children as a distinct group have also been
studied, but no conclusive evidence demonstrates whether children are more or
less likely to display mental health impacts [5, 24]. Age as a factor is highly
confounded with other life stage, physical abilities and stressor variables to the
extent that any true impact of chronological age can rarely be established.
The available research on culture and ethnicity shows ethnic minorities are
more vulnerable to disasters in general possibly due to social deprivation and
marginalisation [12, 25]. The impact of cultural expectations may also have an
influence on the tendency of an individual to seek help [26, 27].
Evidence that implicates the flood severity in mental health disorders strongly
suggests this is related to direct, indirect, tangible and intangible losses [13].
These findings also seem to hold true for disasters generally [5, 12, 28].
Indicators include the number of casualties, deaths, losses and disease. For
example in Thailand severe flooding quadrupled the incidence of PTSD
symptoms whereas in Korea risk of PTSD and depression were influenced by
injury, death of a relative and damage [14]. However, the categorisation of
severity differs across studies and a full diagnostic would be difficult to establish
without considerable further research [9, 29].
On an individual level, pre-existing health conditions and personality factors
are good predictors of post-disaster mental health problems for a given disaster
severity [12]. This has also been demonstrated specifically for flood events [5].
After a flood the stress associated with lack of basic services, evacuation and
poor living conditions, can also damage mental health [4, 13, 30]. Events seen to
be accidents cause less distress than those seen as preventable [5].
Disaster studies that focus on aspects of the stressor have found that
psychological impacts are most likely when at least two severe event factors are
present: Extreme and widespread property damage; serious financial hardship;
human causes for the disaster; high levels of injury and deaths [12]. However
research has also demonstrated differences between different types of mental
health issues with PTSD related more to event stressors and depression
associated with both event stressors and life stressors [9].
It is clear from the above discussion that it is necessary to consider multiple
influencing factors relating to the flood and the population. However research
has tended to focus on factors in isolation, rather than investigate interactions
between multiple stressors and characteristics rendering the evidence indicative
rather than predictive. Recognising these factors may nevertheless help to
identify those most at risk of developing PTSD, anxiety and depression, however
a deeper understanding of the interactions may allow disaster managers and
health professionals to offer specifically targeted support.
3.3 Duration of impact and the influence of flood memory
Duration of impact and flood memory are linked because remembering a flood
will have an influence on the length of psychological symptoms. While treatment
pathways are outside the scope of the research the existing evidence does allow
for the development of some conceptual framing that may be helpful in

designing the timing and focus of intervention strategies.
3.3.1 Duration of impacts on mental health

The duration of symptoms is highly relevant to the influence of repeated
flooding. Many trauma symptoms may disappear quite quickly perhaps due to
particular coping strategies and support. Other symptoms may arrive after some
delay such as PTSD, and different symptoms can emerge in the short and long
term [14]. Apart from the burden on mental health services, psychological effects
from flooding can influence long term physical health and mortality [18, 31].
Longitudinal studies of mental health impacts after flooding are rare but they
suggest that for some individuals the effects are long lasting. For example Norris
et al. [13] found that symptoms persisted more than 2 years after a flood for
some individuals in Mexico and suggested that a minority of those suffering with
PTSD would never recover. Tapsell and Tunstall [21] in England, found impacts
persisted for the full four years of their longitudinal panel. Furthermore
Assanangkomchai et al. [32] studying flood affected populations in Thailand
found resurgence of symptoms on the anniversary of a flood suggesting a strong
link between mental health symptoms and flood memory.
Other cross sectional studies may be carried out long after the traumatic event
yet still find significant uplift in mental health symptoms [12]. The possible
phases in the health effects of floods have been outlined by Parker et al. [33].
These range from threat anxiety in anticipation of an event for those who have
knowledge and experience of flood risk; stress and shock effects during the
event; worry and depression during early recovery; stress and stress related
illness during long-term recovery; and post-event anxiety over future threat and
impaired mental health which brings the phases full circle if a subsequent event
occurs but may eventually dissipate if the threat is removed or forgotten.
3.3.2 Impact of previous flood experience and flood memory

As noted above, previous research has suggested that experience of disaster has
some influence on the mental health outcomes following a second or subsequent
event. Flood memory as a concept has been studied but not generally within the
context of impact on mental health and it can be thought of as an individual,
collective or institutional property of populations at risk [34, 35]. The analysis in
3.3.1 demonstrates a possible logical causal route for the influence of both
experience and memory on the future severity of symptoms. However according
to Mason et al. [4] the direction of this influence is unclear from observed health
outcomes in post flood evaluations. In their recent study in the UK prior
experience was found to have a negative effect on mental health outcomes [4].
Galea et al. [36] and Heo et al. [20] also observe that past trauma increases the
likelihood of developing PTSD, whereas other authors propose an innoculation
theory that prior experience with a stressor increases capacity to deal with it [37,
38]. The positive aspect of this controversy is that the differences in observed
outcomes may be as a result of coping strategies and interventions applied
between one disaster and another. For example people could be encouraged to
engage in detached coping in the short-term at least as this allows for more
effective assessment of the options to prevent recurrence of flooding [4].

Therefore it may be helpful to consider in detail the mechanisms through which
flood experience could influence mental health outcomes in order to identify
potential interventions that could result in more positive outcomes.
4 Conceptual model of the impact of flood memory on

mental health
In conceptualising how flood memory could affect the prevalence of symptoms
in the population each of the identified influencing factors has been considered
in turn. Those factors that remain unchanged due to flood experience (such as
gender) or marginally impacted (such as socio-economic status) were removed.
The remaining factors were considered as to whether they would potentially
increase or decrease the prevalence of mental health issues (table 2).
For many of the influencing factors it was seen that their influence on future
mental health could be in either direction and therefore the mitigating or
contributing actions also need to be considered. This analysis leads to a
conceptual framework of the impact of flood memory on psychological distress
that recognises the role of health and disaster management professionals in
partnership to reduce the future mental health footprint (as seen in Figure 1).
This framework, derived from the existing evidence as categorised under the
research themes above is directional in nature as it is not calibrated with
evidence of the strength or duration of the individual influences on overall
mental health. However, the structured approach would be suitable for
operationalisation if suitable data were available. Furthermore in explicitly
expressing the temporal aspects of the many influencing factors it can be used to
encourage appropriate long term investment in necessary support services.
4.1 Research implications
Mental health consequences of flood events have not been fully addressed in the
past either in disaster or health fields. But the purpose of this present review is
not to predict the required mitigating actions in advance of flooding. Few [16]
and Tapsell et al. [5] have pointed out the difficulty and futility in making this
attempt. It is important to avoid pathologising a natural reaction to trauma and
offer appropriate post disaster support that can detect those individuals that may
be more vulnerable and likely to develop more severe and longer term symptoms
[17]. In this respect the memory and frequency of flooding could be instrumental
in setting up conditions that could trigger higher levels of emotional distress.
Therefore it is relevant to consider what actions can be taken in recently flooded
locations to mitigate against the impact of a second flood.
We can deduce that mitigation of mental health impacts for a given individual
with fixed personality and socio economic conditions for the next flood might be
achieved through one of the following approaches:
The first approach is to assess what steps could be useful in preventing the
development of psychological problems after an event. The model suggests
Table 2: Mental health influencing factors affected by flood memory.
Pre existing Features of the stressor Post event stress and

conditions event coping strategies
History of Level of personal property Need for relocation
psychiatric disorder damage and loss (could be (could be improved
(could be made improved by preparedness through better resilient
worse because of but otherwise could be reinstatement)
lasting impact of cumulative)
previous event)
History of other Perception of human Lack of resources for
health related control or responsibility recovery (May be more
problems (could be (Frequent flooding could prepared but previously
made worse because feel like victimisation but held resources could be
of lasting impact of improved preparedness exhausted)
previous event) could give feeling of
control)
Disaster experience Distress of others,
or training (Could be particularly spouse
improved because of (could be worse through
previous experience) memory of flooding)
Social support (may be
improved due to past
experience or subject to
compassion fatigue)
Coping strategies (May
be enhanced through
past experience but may
have to be abandoned)
Living with the threat of
constant or growing
flood threat (likely to be
made worse by flood
memory)
helpful interventions might be: support for faster reinstatement particularly for
those without insurance; sympathetic insurance and reinstatement professionals;
good advice on coping strategies that help individuals forget or assimilate the
trauma of their experiences; strengthening of community networks and other
social support; and provision of counselling and direct mental health support.
The second approach reduces future vulnerability to flooding through taking
steps to increase the capacity and resilience of people and the built environment
[39]. Helpful interventions might include: resilient reinstatement of buildings
[40], disaster preparedness training; and warning systems. Indeed a whole range
of flood risk management measures can be used to lower the risk of physical
Figure 1: Impact of flood memory on mental health impacts in frequently

flooded communities.
damage and loss [41]. The resilience of the population at risk can also be
enhanced through many means including: the provision of peace of mind via
insurance; good advice on coping styles that enable rational actions while
reducing trauma; preparedness training; improving general wellbeing; and
boosting community cohesion [39].
5 Discussion and conclusions

Impact on the mental health of affected households, usually PTSD, depression or
anxiety, is often the largest health effect observed in the context of flood events.
The effects can last for many years within a population although the majority of
individuals will recover quite quickly. There are several indicators that may
allow responders to identify those most at need of support in the aftermath of
disasters but large scale predictive mapping of likely need is problematic.
Flood memory is one of the indicators found to have an influence on the
prevalence of mental health disturbance but there are some contradictory
findings and theories surrounding this issue. Underlying causes of the differential
observed outcomes could be related to the balance between worsening pre-
existing health conditions and improvements in preparedness that may lessen the
stress of the subsequent flood and recovery period.
Interventions, physical and psychological, in the immediate aftermath of a
flood event, designed to restore or improve the pre-flood conditions will
therefore be expected to mitigate against the worsening pre-existing conditions
for successive floods. Furthermore, action to limit the severity of flooding in the
successive flood will contribute to lower levels of trauma and therefore may lead
to improved mental health outcomes.
Coping strategies need to be explored. Those strategies appropriate to short
term recovery following a single flood may be different from those appropriate
for those at risk from frequent and repeated flooding. However an initially
detached coping style may enable a more rational consideration of the options
available to act.
Finally the results show that future research in this field may benefit from a
multi-dimensional approach to measurement of impacts and further
consideration of the complex relationship between concepts of flood memory
and mental health. The formulation of a conceptual model of the impacts of
flood memory on mental distress following repeated flooding will be helpful in
deriving appropriate multi-dimensional research designs that include the effect of
previous flooding.
Acknowledgement
This research was funded by the UK Engineering and Physical Sciences
Research Council under grant EP/K013513/1 Flood MEMORY: Multi-Event
Modelling Of Risk & recoverY.
References
[1] B. Carroll, R. Balogh, H. Morbey, and G. Araoz, “Health and social
impacts of a flood disaster: responding to needs and implications for
practice,” Disasters, vol. 34, pp. 1045–1063, 2010.
[2] L. Fewtrell and D. Kay, “An attempt to quantify the health impacts of
flooding in the UK using an urban case study,” Public Health, vol. 122,
pp. 446–451, 2008.
[3] S. Hajat, K. L. Ebi, S. Kovats, B. Menne, S. Edwards, and A. Haines, “The
human health consequences of flooding in Europe and the implications for
public health: a review of the evidence,” Applied Environmental Science
and Public Health, vol. 1, pp. 13–21, 2003.
[4] V. Mason, H. Andrews, and D. Upton, “The psychological impact of
exposure to floods,” Psychology, Health & Medicine, vol. 15, pp. 61–73,
2010/01/01 2010.
[5] S. M. Tapsell, S. M. Tunstall, and S. Priest, “Developing a conceptual
model of flood impacts upon human health,” FloodSite Report T10-09-02,
2009.
[6] T. W. Collins, A. M. Jimenez, and S. E. Grineski, “Hispanic Health
Disparities After a Flood Disaster: Results of a Population-Based Survey
of Individuals Experiencing Home Site Damage in El Paso (Texas, USA),”
Journal of Immigrant and Minority Health, vol. 15, pp. 415–426,
2013/04/01 2013.
[7] S. Paranjothy, J. Gallacher, R. Amlôt, G. J. Rubin, L. Page, T. Baxter, et

al., “Psychosocial impact of the summer 2007 floods in England,” BMC
Public Health, vol. 11, p. 145, 2011.
[8] C. S. Rosen, M. M. Matthieu, and F. H. Norris, “Factors Predicting Crisis
Counselor Referrals to Other Crisis Counseling, Disaster Relief, and
Psychological Services: A Cross-Site Analysis of Post-Katrina Programs,”
Adm Policy Ment Health vol. 36, pp. 186–194, 2009.
[9] J. M. Shultz, A. McLean, H. B. Herberman Mash, A. Rosen, F. Kelly, H.
M. Solo-Gabriele, et al., “Mitigating flood exposure: Reducing disaster
risk and trauma signature,” Disaster Health, vol. 1, pp. 30–44, 2013.
[10] K. Azuma, K. Ikeda, N. Kagi, U. Yanagi, K. Hasegawa, and H. Osawa,
“Effects of water-damaged homes after flooding: health status of the
residents and the environmental risk factors,” International Journal of
Environmental Health Research, vol. 24, pp. 158–175, 2013.
[11] C. Stanke, V. Murray, R. Amlôt, J. Nurse, and R. Williams, “The effects
of flooding on mental health: Outcomes and recommendations from a
review of the literature,” PLOS Current Disasters, vol. 1, 2012.
[12] F. H. Norris, M. J. Friedman, and P. J. Watson, “60,000 disaster victims
speak: Part II, summary and implications of the disaster mental health
research,” Psychiatry, vol. 65, pp. 240–260, 2002.
[13] F. H. Norris, A. D. Murphy, C. K. Baker, and J. L. Perilla, “Postdisaster
PTSD Over Four Waves of a Panel Study of Mexico’s 1999 Flood,”
Journal of Traumatic Stress, vol. 17, pp. 283–292, 2004.
[14] K. Alderman, L. R. Turner, and S. Tong, “Floods and human health: A
systematic review,” Environment International, vol. 47, pp. 37–47, 2012.
[15] M. Reacher, C. McKenzie, C. Lane, T. Nichols, I. Kedge, A. Iverson, et
al., “Health impacts of flooding in Lewes: a comparison of gastrointestinal
and other illness and mental health in flooded and non-flooded
households.,” Communicable Diseases and Public Health, vol. 7, pp. 39–
46, 2004.
[16] R. Few, “Health and Climatic hazards: framing social research on
vulnerability, response and adaptation,” GLOBAL Environmental Change,
vol. 17, pp. 281–295, 2007.
[17] K. Tierney, “Controversy and Consensus in Disaster Mental Health
Research,” Prehospital and Disaster Medicine, vol. 15, pp. 55–61, 2000.
[18] S. N. Jonkman, “Loss of life estimation in flood risk assessment, theory
and applications,” Technische Universiteit Delft, 2007.
[19] V. Murray, H. Caldin, R. Amlôt, C. Stanke, S. Lock, H. Rowlatt, et al.,
“The Effects of Flooding on Mental Health,” Health Protection Agency
2011.
[20] J. H. Heo, M. H. Kim, S. B. Koh, S. Noh, J. H. Park, and J. S. Ahn, “A
Prospective Study on Changes in Health Status Following Flood Disaster,”
Psychiatry Investigation, vol. 5, pp. 186–192, 2008.
[21] S. Tapsell and S. Tunstall, “An examination of the health effects of
flooding in the United Kingdom,” Journal of Meteorology, vol. 28,
pp. 341–349, 2003.
[22] F. Huerta and R. Horton, “Coping Behavior of Elderly Flood Victims,”

The Gerontologist, vol. 18, p. 541, 1978.
[23] A. B. Amstadter, R. Acierno, L. K. Richardson, D. G. Kilpatrick, D. F.
Gros, M. T. Gaboury, et al., “Post-Typhoon Prevalence of Post-traumatic
Stress Disorder, Major Depressive Disorder, Panic Disorder and
Generalized Anxiety Disorder in a Vietnamese Sample,” Journal of
Traumatic Stress vol. 22, pp. 180–188, 2009.
[24] Z. Chen, Y. Zhang, Z. Liu, Y. Liu, and A. Dyregrov, “Structure of the
Children’s Revised Impact of Event Scale (CRIES) with Children and
Adolescents Exposed to Debris Flood,” PLoS ONE, vol. 7, p. 41741, 2012.
[25] S. M. Tapsell and S. M. Tunstall, “I wish I’d never heard of Banbury”:
the relationship between ‘place’ and the health impacts from flooding,”
Health and Place, vol. 14, pp. 133–154, 2008.
[26] A. Maercker and J. Muller, “Social Acknowledgment as a Victim or
Survivor: A Scale to Measure a Recovery Factor of PTSD,” Journal of
Traumatic Stress, vol. 17, pp. 345–351, 2004.
[27] Galen E. Switzer, “Posttraumatic Stress Disorder and Service Utilization
Among Urban Mental Health Center Clients,” Journal of Traumatic
Stress, vol. 12, pp. 25–39, 1999.
[28] S. H. Bland, E. S. O’Leary, E. Farinaro, F. Jossa, and M. Trevisan, “Long-
Term Psychological Effects of Natural Disasters,” Psychosomatic
Medicine vol. 58, pp. 18–24, 1996.
[29] P. Verger, M. Rotily, C. Hunault, J. Brenot, E. Baruffold, and D. Bard,
“Assessment of exposure to a flood disaster in a mental-health study,”
Journal of Exposure Analusis and Environmental Epidemiology, vol. 13,
pp. 436–442, 2003.
[30] R. Whittle and W. Medd, “Living with Flood: Understanding residents’
experience of recovery,” in Flood hazards, impacts and responses for the
built environment, J. E. Lamond, D. G. Proverbs, C. A. Booth, and F. N.
Hammond, Eds., ed New York: Taylor CRC press, 2011.
[31] G. Bennet, “Bristol floods 1968: controlled survey of effects on health of
local community disaster,” British Medical Journal, vol. 3, pp. 454–458,
1970.
[32] S. Assanangkornchai, S. Tangboonngam, N. Samangsri, and J. G.
Edwards, “A Thai community’s anniversary reaction to a major
catastrophe,” Stress and Health, vol. 23, pp. 43–50, 2007.
[33] D. J. Parker, C. H. Green, and P. M. Thompson, Urban Flood Protection
Benefits: A Project Appraisal Guide. Aldershot: Gower, 1987.
[34] J. Boardman and K. Vandaele, “Soil erosion, muddy floods and the need
for institutional memory,” Area, vol. 42, pp. 502–513, 2010.
[35] J. Lamond, “Elephants or ostriches, does the collective memory of flood
events fade with time?,” in First World Conference for Accelerating
Excellence in the Built Environment, Birmingham, 2006.
[36] S. Galea, A. Nandi, and D. Vlahov, “The Epidemiology of Post-Traumatic
Stress Disorder after Disasters,” Epidemiologic Reviews, vol. 27, pp. 78–
91, 2005.
[37] F. H. Norris and S. A. Murrell, “Prior experience as a moderator of

disaster impact on anxiety symptoms in older adults,” American Journal of
Community Psychology, vol. 16, pp. 665–683, 1988.
[38] Morrissey S and J. Reser, “Natural disasters, climate change and mental
health considerations for rural Australia. ,” Aust J Rural Health vol. 15,
pp. 120–125, 2007.
[39] I. de Terte, J. Becker, and C. Stephens, “An Integrated Model for
Understanding and Developing Resilience in the Face of Adverse Events,”
Journal of Pacific Rim Psychology, vol. 3, pp. 20–26, 2009.
[40] R. Joseph, J. Lamond, and D. Proverbs, “Property-level flood adaptation:
A tool for reducing the intangible impact of floods on households in the
UK,” presented at the International conference in building resilience:
Interdisciplinary approaches to disaster risk reduction and the development
of sustainable communities, Kandalama, Sri Lanka 2011.
[41] A. Jha, J. Lamond, R. Bloch, N. Bhattacharya, A. Lopez, N.
Papachristodoulou, et al., “Five Feet High and Rising – Cities and
Flooding in the 21st Century,” The World Bank, Washington DC 2011.
The long-term health impacts of repeated

flood events
J. Stephenson, M. Vaganay, R. Cameron & P. Joseph
School of the Built Environment and the
Built Environment Research Institute, University of Ulster, UK
Abstract
During the past 30 years, floods have resulted in over 200,000 fatalities and
affected more than 2.8 billion others worldwide. Flood victims are vulnerable to
long-term physical and psychological health effects, which persist for an
undefined time period in the aftermath of a flood event. Following a flood event,
secondary stressors, which are indirectly related to the event, can potentially
prolong and intensify the health impacts on affected individuals and communities.
These secondary stressors consist of economic stressors, including loss of income,
but also social stressors such as isolation due to prolonged flooding. A significant
gap in the research to date is in relation to repeated flooding and its impact on the
extent to which individuals are affected by these secondary stressors. This review
examined studies focusing on repeated flooding, concentrating on the secondary
stressors resulting from repeated flood events. It also considered the awareness,
preparedness and resilience of the study populations in order to determine the
potential for these communities to be impacted by secondary stressors. This review
indicated that both rural and urban communities in developed and developing
countries are significantly affected by economic, social and psychological
secondary stressors. The majority of communities do have a basic awareness of
flood risk; however, many residents do not take flood risk seriously and thus take
little preventative action. Community resilience was higher in urban and rural
areas in developing countries, but also in rural areas in developed countries. Future
work should take into consideration the secondary stressors that affect different
communities and how to minimise their impact in order to increase resilience.
Keywords: flooding, repeated flooding, health, secondary stressors, resilience,
awareness, preparedness, urban, rural.
doi:10.2495/FRIAR140171
1 Introduction
Flooding has become the most frequent type of major disaster globally within both
developing and developed countries [1]. The World Health Organisation
concluded that during the past 30 years, flooding resulted in over 200 000 fatalities
and affected more than 2.8 billion others worldwide [2]. The statistics illustrate
that flooding is a worldwide phenomenon and an unquestionable cross border
issue.
Flooding poses multiple risks to health and growing evidence worldwide
indicates that the health impacts of flooding penetrate a lot deeper than the
immediate physical impacts such as injuries and drowning [3]. An increasing
recognition is that following extreme events such as floods, secondary stressors,
which are indirectly related to the event, can potentially prolong and intensify the
health impacts on affected individuals and communities [4]. These secondary
stressors take in economic stressors including the impact on property values, but
also social stressors such as forced isolation due to a prolonged flood event.
Although numerous secondary stressors of extreme events have been identified,
there remains a need to establish whether repeated flooding has an impact on the
extent to which communities are affected by these stressors and also to investigate
if these stressors have similar impacts on different types of communities, such as
urban and rural areas.
Therefore the aims of this paper are:
1. To identify studies which have examined the long-term impact of
repeated flooding.
2. To summarise and critically review the published literature to date on the
secondary stressors impacting urban and rural flooded populations.
3. To establish the awareness, preparedness and level of resilience of
communities which have suffered repetitive flooding in order to
determine their vulnerability to secondary stressors.
4. To determine the knowledge gaps in the research relating to the long-
term health impact of repeated flooding on communities.
2 Methods
A literature search using Proquest, Science Direct, Medline and Web of Science
was conducted. The search was limited to peer-reviewed articles published in
English. Table 1 outlines the search strategy that was used to identify studies to be
included in the paper. It included a combination of key words relating to exposure,
health outcomes, susceptibility to flooding and the location of the flood event.
Studies were eliminated that did not focus on the health impacts of flood events or
secondary stressors of flood events. It was also decided to disregard articles which
focused solely on immediate impacts of flooding such as mortality, diarrhoeal
diseases etc. In addition, papers were excluded that addressed only the health
impacts of single flood events. After the search strategy was executed it was
decided to discount studies where the sample population did not permanently
reside in the area that had been repeatedly flooded i.e. studies on caravan sites.
The full texts of the remaining articles which met the inclusion criteria were then
critically reviewed by the first author, the key findings of which are summarised
in this paper.
Table 1: Search strategy used to identify studies for inclusion.

Key words relating to repeated flood events (exposure)
flood* OR “repeated flood*” OR “frequent flood*” OR “successive flood*” OR
“continuous flood*” OR “reoccurring flood*” OR “regular flood*” OR “habitual
flood*” OR “intermittent flood*” OR “recurring flood*” OR “constant flood*” OR
“repetitive flood*” OR “continual flood*” OR “perpetual flood*” OR “routine flood*”
AND
Key words relating to health outcomes (outcome)
health OR “secondary stressors” OR “health impacts” OR “mental health” OR stress
OR anxiety OR depression
AND/OR
Key words relating to susceptibility to flooding
resilience OR awareness OR “flood risk” OR vulnerab*
AND/OR
Key words relating to location of flood event
urban OR rural
3 Results and discussion

The initial search resulted in 4970 potentially relevant references which matched
the key words used in the search strategy. Following the application of the
inclusion criteria, only 118 articles were reserved for full article review, excluding
articles which did not focus on the human health impact of floods. Upon further
examination 93 articles were eliminated as not meeting the full criteria, namely
repeated flood events. The full texts of the remaining 25 studies were critically
analysed by the first author. The studies were divided, where possible, into urban
and rural case studies.
Examination of the 25 key articles identified for inclusion in this study revealed
that 16 of the studies were in rural locations, 8 were in urban locations and 1
study had both an urban and a rural case study. 12 of the rural studies were in
developing countries and 4 in developed countries. 4 of the urban studies were
in developing countries and 4 in developed countries. The urban and rural study
took place in a developed country.
3.1 Secondary stressors impacting flood affected populations
Flood victims are exposed to a traumatic and frustrating sequence of events which
can potentially contribute to health implications long after the subsidence of the
flood. This is particularly the case for individuals affected by repeated flood
events, who are constantly re-exposed to these health implications, often before
they have had a chance to recover from a previous flood experience. A significant
issue for flood victims is the secondary stressors which come to the forefront in
the aftermath of flood events, which can be just as detrimental to health as flood
related injuries and diseases.
3.1.1 Economic stressors

3.1.1.1 Loss of income, disruption to livelihoods and debt One of the most
significant secondary stressors identified in this study was that flooding affected
livelihoods and resulted in substantial loss of income [5–21]. Reasons for loss of
income can include: damage to road infrastructure preventing respondents
reaching their workplace or dealing with the damage caused to property [5–10, 12,
15, 17, 19, 21, 22].
Loss of income was a major concern in both urban and rural areas and
developed and developing countries. Braun and Aßheuer found that 70% of
households in an urban area in Bangladesh faced a significant decrease in their
income following flooding [7]. A rural study in Vietnam established that
individuals experienced flooding for two to six months annually, thus resulting in
a substantial loss of income [13].
Flooding also posed a risk to traditional livelihoods in rural areas [9, 11, 12,
14, 15, 20]. Dun found that the disruption caused to rice farmers in Vietnam by
repeated flooding resulted in the decision to migrate as there was no alternative
livelihood [14]. In Vietnam seasonal migration often occurs as a means for flood
affected families to boost their income [13, 14].
Debt was an additional economic stressor identified in this review [13, 14, 16,
18]. Nguyen and James identified that the disruption to income caused by flooding
often led poor households to acquire high interest loans [13]. Low socio-economic
groups often found it very difficult to repay these loans, leading to further stress
and anxiety.
It is important to recognise that business owners are often under as great a strain
as homeowners following flood events due to the impact on their livelihood [6,
23]. Hoggart et al. found that business owners faced significant worries regarding
flooding as it caused physical damage to merchandise, disrupted transportation of
deliveries, prevented customer and employee access to the business and had the
potential to cause electricity blackouts [6].
Loss of income was a universal problem for the majority of flood victims in the
studies examined by this review. Location and socio-economic status did not have
any bearing on its impact, although it was more difficult for individuals from
developing countries to recover as they often had little or no savings to assist them
post-flood.
3.1.1.2 Damage to property and possessions Damage to property is particularly

traumatising for flood victims as they often lose not only valuable assets, but also
irreplaceable items of sentimental value [5–10, 12, 15, 17, 19, 21, 22]. However,
an additional consequence of damage to property noted in this study was that
damage to assets such as vehicles could have far reaching implications as it could
prevent travel to work, resulting in loss of income, as well as limiting social
interaction [5, 7]. Furthermore the potential for damage to possessions to occur
can affect lifestyle choices, as repeated flood risk can restrict the appliances or
furniture that flood victims can realistically purchase [5]. Returning properties to
their pre-flood status is very difficult as state compensation is often inadequate
and many do not have enough savings to restore dwellings immediately [15, 19].
Damage to property and possessions can be extremely traumatising as flood

victims lose both essential items and items that are irreplaceable to them. Due to
repeated flooding, individuals may have to go through this stress multiple times.
The review shows that damage to property and possessions was also a widespread
stressor, suffered in rural and urban areas and developed and developing countries.
However, it can be concluded that it has a much more significant impact in
developing countries, as individuals do not have the financial resources to replace
everything that was lost.
3.1.1.3 Insurance and house prices Following a flood event, the key priority for
homeowners is to restore their property to its pre-flood state. However, with the
increasing prevalence of flooding, a rising problem is obtaining flood insurance
[17, 23, 24]. A UK study by Lamond et al. noted that 13% of respondents were
refused a quote for insurance due to flood risk and 3% were denied a renewal due
to flood risk [24].
In developed countries the increased prevalence of flooding has led to the need
for homeowners to invest in flood insurance. Flood insurance is often non-existent
in developing countries and flood victims have to use their savings to cover the
financial costs. Obtaining insurance for flood risk areas is now very difficult in
developed countries and although the majority of homeowners do eventually
obtain insurance, the strain on individuals already traumatised by other flooding
related stressors should not be underestimated.
3.1.2 Social stressors

3.1.2.1 Fear of reoccurrence For individuals with previous flood experience, the
stress and anxiety associated with flood memories leaves them fearful of repeated
flooding [5, 16, 19, 21]. A study in Guyana, South America found that 58% of
respondents acknowledged that they worried every time it rained heavily [21].
Fear of reoccurrence can be a particular problem in developing countries, where
lack of formal flood warning systems means households personally monitor signs
of flooding, paying particular attention during prolonged rainfall. This on-going
stressor can have a significant impact on daily life, straining family relationships,
leaving homeowners wary of going on holiday and often prompting individuals to
come home early from work when it starts to rain [5]. In developed countries it
can often result in obsession with weather forecasts [16].
Fear of reoccurring flooding is evident in this study in both urban and rural
areas, but particularly in urban areas in developing countries. Other secondary
stressors can contribute to it such as loss of income and damage to property. It is
an extremely difficult stressor to alleviate as only a reduction in flood occurrence
or migration can overcome it.
3.1.2.2 Migration and displacement Temporary evacuation is a common

occurrence during flood events, in relation to preventing mortality and injuries,
but also in order to avoid the adverse health impacts associated with living in a
flooded or damp home [5, 7–10, 13–15, 17]. Studies in Bangladesh found that 50–
95% of all respondents had to leave their homes during recurrent flood events [7,
9]. This is particularly difficult for young children and the elderly who may not
understand the situation and feel anxious away from normal surroundings and
belongings.
Repeated flooding can eventually leave residents with only one option, that of
relocation. A study in New Orleans found that 23% of respondents were
considering, trying or actually selling their home [25]. Recurrent damage to
property or successive flooding resulting in damage to crops can encourage
migration [14]. Migration, although a means of removing exposure to repeated
flooding, can be a secondary stressor in itself as displacement from familiar
surroundings and severing of ties from family and neighbours can be extremely
distressful [5, 17, 24]. In many circumstances, flood victims wish to remain in
their homes despite the high flood risk, due to fear of the unknown. A study in the
USA found that while 190 households accepted the relocation offer, 47 rejected it
[22]. However, often the constant fear of reoccurring flooding can become too
stressful, leaving migration as the only alternative.
It is important to recognise that when the decision to migrate is eventually
made, the stress does not necessarily come to an end, due to financial issues such
as problems with selling homes or higher house prices in non-flood risk areas [5,
15]. A study in Alaska identified that the main obstacle to migration was
government funding, as the cost of relocating the village was too high [22]. In
addition, home owners are not always offered what they consider the true value of
their property, leading to added stress when deciding whether to agree to
government resettlement [15]. Property owners who perceived the condition of
their property as higher or were attached to their neighbourhood found it more
difficult to relocate, however, fear of future flooding made acceptance easier [16].
Dun identified that when government resettlement occurs, households are
provided with a five year loan to buy a housing plot; thus resettlement can actually
become an economic stressor, causing individuals to go into debt [14].
Temporary evacuation is common in both rural and urban areas. It is
unavoidable in both developed and developing countries in order to protect the
health of flood victims. However, this study has found that migration is generally
a rural issue, as individuals decide to move permanently to urban areas where they
feel that flood risk is a lesser concern. It is essential to recognise that while
migration and resettlement reduces physical exposure to the health impacts of
repeated flooding, it can increase economic and social vulnerabilities; therefore
migration can act as both a solution and a stressor.
3.1.3 Physical and psychological stressors

3.1.3.1 Long-term malnutrition During flood events, food supplies are often
scare, both due to destruction by floods and inability to reach shops or bring in
supplies via damaged roads. In order to survive, flood victims often have to reduce
the number of meals they consume [7, 9, 12, 26]. Rodriguez-Llanes et al.
concluded that children from flooded homes were more likely to be underweight
and underdeveloped compared to children from non-flooded homes. This is a
particular concern as child malnutrition is associated with underdevelopment, poor
school performance and even premature mortality [26].
Long-term malnutrition is a key issue in developing countries, particularly in

rural areas where it is more difficult for government and voluntary agencies to
reach flood victims to provide aid. It is vital for government agencies to do more
to assist flood victims in rural communities in developing countries to prevent
long-term health problems.
3.1.3.2 Stress, anxiety and depression Repeated flood events place individuals
and communities under severe stress and anxiety due to both social and economic
reasons [18, 19, 24, 27]. Stress and anxiety can sometimes have a significant
impact on human behaviour. Biswas et al. found that 70% of mothers and 40% of
fathers abused their children during flood events [18]. The possible long-term
health effects of abuse at an early age are widely recognised, including a tendency
towards eating disorders and depression [18]. It is important to recognise that the
extreme stress and anxiety placed on recurrent flood victims has the potential to
lead to mental health problems. In a 2013 study on repeated flood events, Wind et
al. found that mental health symptoms, such as anxiety and depression were
notably higher than other studies on natural disasters, suggesting that repeated
events have an even greater influence on mental health [27].
Stress, anxiety and depression due to flooding were identified in both urban
and rural studies in this review. Individuals from developing countries are
particularly susceptible as they have limited resources to recover from flooding.
Stress, anxiety and depression generally occur as a result of other secondary
stressors such as damage to property. It is therefore crucial to acquire a greater
knowledge of the impact of flood related secondary stressors to minimise the
mental health impacts of recurrent flooding.
3.1.3.3 Isolation Recurrent flood events can prevent individuals commuting to

work and school and can often result in families or individuals being isolated for
significant periods until the flood waters recede [8, 19]. Isolation can contribute to
mental health conditions as it can cause stress and anxiety to escalate.
The review indicated that this is a problem in both urban and rural areas; in
rural areas due to the remote location, while in urban areas flood victims may be
unable to leave their homes until the waters recede.
3.2 Awareness and preparedness for flood events
It is essential to understand that awareness of and preparedness for flood events
are completely separate concepts. In many cases individuals are fully aware of the
flood risk, but do not take action, despite previous flood history. In this review, 16
out of the 25 studies considered how aware populations were regarding the risk of
repeated flooding. In all of those 16 studies homeowners and business owners
were aware at least to some extent of flood risk. However, lack of awareness can
be a significant problem in some developing countries due to issues such as low
education, lack of effective warning systems etc. [10, 11]. Although awareness of
flood risk may exist, an additional issue can be lack of awareness in relation to the
health implications of flooding, flood insurance cover or to weather forecasting
that could increase likelihood of being prepared [10, 11, 17, 24].
This review established that lack of preparedness for flooding was a significant
problem [6, 7, 10, 17, 20, 28, 29]. This problem was evident in both urban and
rural and developing and developed countries. An additional issue was that the
majority of preparation made was reactive rather than proactive [6, 10, 11]. The
key reasons were lack of awareness regarding mitigation measures, lack of
financial resources to invest in preventative measures, lack of ‘know-how’ and a
limited time frame between successive flood events [5–7, 11, 13, 15, 20, 29].
Several of the studies also recognised that often residents thought they were
prepared for flooding, but in reality they had not taken sufficient mitigation action
or their insurance did not cover flood events [17, 23, 24]. Furthermore, some
residents, despite being aware of future flood risk, chose to ignore it, believing that
it would not reoccur [16]. In developed countries, homeowners and business
owners are often dismissive of flood preparedness, suggesting it is entirely the
government’s responsibility [17, 23].
A key finding of this review was that the majority of flood victims who were
flood prepared resided in developing countries [5, 9, 11–13, 15, 19, 21]. However,
two of the studies in developed countries also illustrated flood preparedness [24,
25]. A potential explanation is that in developing countries flood victims see
mitigation and preventative action as their own responsibility rather than that of
government agencies, while flood victims in developed countries have a higher
expectation in relation to government assistance. Additionally Mishra et al. found
that place attachment significantly influenced flood preparedness, indicating that
residents who had lived in the same location for generations would be more likely
to take mitigation action in order to protect their heritage [28].
The review indicates that although a basic level of flood awareness did exist in
all the studies, there was a lack of awareness relating to mitigation measures and
the health implications of flooding. Lack of preparedness was a major issue of
concern in both developed and developing countries, however, flood victims in
developing countries, in both urban and rural areas were more likely to be flood
prepared due to their lack of dependence on agencies. The findings of the review
suggest that communities are very vulnerable to being affected by secondary
stressors due to their lack of flood preparedness. It is important to raise awareness
concerning the ramifications in communities if they do not prepare for flooding,
emphasising that lack of awareness and preparedness for flood events can escalate
the long-term impact of secondary stressors on health [10].
3.3 Community resilience: adaptation and mitigation
Communities are increasingly being encouraged to work together collectively to

recover from and become more resilient to flooding, with little or no assistance
from government agencies. Community resilience is particularly important in
developing countries, where neighbours and relatives are an essential source of
both emotional and physical support during flood events.
This review found significant community resilience in flood affected areas,
with strong community spirit helping during the flood and making the recovery
period easier [5, 7, 8, 10, 12, 17]. It was even suggested that flooding increased
community resilience, encouraging closer relationships between neighbours [8].
In developing countries, despite dire economic conditions almost all

residents were willing to loan money and offer shelter to neighbours during flood
events [7, 8].
It was established that both individual and community resilience was especially
strong in developing countries were following flood experience as residents
developed their own coping strategies to protect against flood events. The methods
employed included placing barriers around houses, construction of floating
houses, using concrete blocks instead of timber to reduce likelihood of floodwater
entering homes and taking shelter on higher ground during flood events [5, 12].
Farmers were found to be particularly resilient, adapting to flooding in new
innovative ways, such as turning to an alternative livelihood during the flooding
season or selecting crops around the timeframe of seasonal flooding [12, 13]. This
is crucially important in developing countries as flooding can destroy livelihoods,
however, exploring innovative ways to benefit from flooding may be something
that could also be considered in developed countries.
It is important to understand that one of the key reasons as to why individual
resilience often remains low is lack of financial resources. Upgrading properties
as a mitigation technique to reduce flood risk allows individuals to become more
resilient, however, in both developing and developed countries it is not always
financially viable [11, 13, 20]. One of the key factors identified to increase the
likelihood of responding to flood hazard and adapting properties was housing
tenure. It was identified that owner occupier households were more likely to
mitigate against flood risk through actions such as yard raising, essentially because
owner occupier households tend to be a higher income group [21]. Another
problem identified was that individuals took temporary short-term measures to
deal with floods each year, without taking any action that will assist them during
the next flood [11]. This is often due to a lack of awareness and education [12]. It
is essential to improve flood education and mitigation programmes, especially
within rural areas in developing countries in order to help homeowners to
understand that proactive mitigation measures will reduce stress and save money
in the long-term. A final problem identified was that many homeowners and
businesses appeared to act alone to mitigate flood risk rather than collectively [6,
19, 20].
This review found that community resilience was higher in developing
countries, both in urban and rural areas, but also in rural areas in developed
countries. The development of community resilience is sometimes hampered by
lack of financial resources or the use of short-term mitigation measures. It is
essential to improve community resilience, particularly in urban areas in
developed countries as it plays an essential role in limiting the detrimental long-
term impact that secondary stressors can have on health.
4 Conclusion
This review concluded that both rural and urban communities in developed and
developing countries are significantly affected by secondary stressors. We
identified that loss of income and damage to property had an almost universal
impact, affecting the majority of communities, despite geographical and socio-

economic differences. However, we found that other secondary stressors only
impacted specific communities, for example, insurance and house prices were a
significant issue in developed countries, but not in developing countries as
insurance was often not available.
In relation to flood awareness and preparedness we concluded that although the
majority of communities do have a basic awareness of their flood risk, many
residents do not take this risk seriously and thus take little preventative action.
Flood victims in developing countries, despite having fewer financial resources,
were recognised as being more likely to take preventative action than residents in
developed countries, due to a lesser dependence on government agencies.
Developed countries must learn this lesson from the developed world, recognising
that responsibility to flood proof homes does not solely lie with government
institutions.
Finally, we concluded that community resilience was high in urban and rural
areas in developing countries and also in rural areas in developed countries.
However, we established that community resilience in urban areas in developed
countries was low. A significant finding was that farmers in developing countries
used innovative methods to adapt to repeated flooding. This led us to conclude that
flood victims in developed countries must find their own means of adapting to
repeated flood events in order to improve their coping strategies, making them
more resilient and prepared for future flooding.
Future studies, which we hope to accomplish, should focus on the specific
secondary stressors faced by rural and urban communities and how to minimise
their impact in order to increase resilience. We concluded that different priorities
exist for flood victims and it is probable that government agencies need to deal
with them in different ways, even if they are located near to each other
geographically. Future work could include proposing a model in relation to dealing
with the long term health impacts of floods among different communities.
Communities should be educated on cost effective proactive preventative
measures that they can implement, allowing them to become more aware of the
detrimental health impacts if only reactive action is taken. There should be a
particular focus on the impact of repeated flooding in urban areas as we identified
that most studies to date have concentrated on rural communities. The increasing
prevalence of flooding in the developed world emphasises the need for more work
in this field, thus further work will contribute to an increased knowledge and
understanding of the secondary stressors that affect communities, allowing
policymakers to establish if a generic approach during the recovery stage of flood
events is appropriate for different communities.
References
[1] Ahern, M., Kovats, R.S., Wilkinson, P., Few, R. and Matthies, F., Global
health impacts of floods: epidemiologic evidence. Epidemiologic reviews, 27
(1), pp. 36–46, 2005.
[2] World Health Organisation. Floods in the WHO European Region: Health
effects and their prevention. WHO: Copenhagen, 2013.
[3] Few, R. and Matthies, F., Flood hazards and health: responding to present
and future risks, Earthscan: London, 2006.
[4] Lock, S., Rubin, G.J., Murray, V., Rogers, M.B., Amlôt, R. and Williams,
R., Secondary stressors and extreme events and disasters: a systematic
review of primary research from 2010–2011. PLoS currents, 4, 2012.
[5] Zahari, R.K. and Ariffin, R.N.R., Community-Based Disaster Management
in Kuala Lumpur. Procedia-Social and Behavioral Sciences, 85, pp. 493–
501, 2013.
[6] Hoggart, S., Hanley, M., Parker, D., Simmonds, D., Bilton, D., Filipova-
Marinova, M., Franklin, E., Kotsev, I., Penning-Rowsell, E. and Rundle, S.,
The consequences of doing nothing: The effects of seawater flooding on
coastal zones. Coastal Engineering, 87, pp. 169–182, 2014.
[7] Braun, B. and Aßheuer, T., Floods in megacity environments: vulnerability
and coping strategies of slum dwellers in Dhaka/Bangladesh. Natural
Hazards, 58 (2), pp. 771–787. 2011.
[8] Boon, H.J., Disaster resilience in a flood-impacted rural Australian
town. Natural Hazards, 71 (1), pp. 683–701, 2014.
[9] Paul, S.K. and Routray, J.K., Flood proneness and coping strategies: the
experiences of two villages in Bangladesh. Disasters, 34 (2), pp. 489–508,
2010.
[10] Tschakert, P., Sagoe, R., Ofori-Darko, G. and Codjoe, S.N., Floods in the
Sahel: an analysis of anomalies, memory, and anticipatory learning. Climatic
Change, 103 (3-4), pp. 471–502, 2010.
[11] Shimi, A.C., Parvin, G.A., Biswas, C. and Shaw, R., Impact and adaptation
to flood: A focus on water supply, sanitation and health problems of rural
community in Bangladesh. Disaster Prevention and Management, 19 (3), pp.
298–313, 2010.
[12] Mavhura, E., Manyena, S.B., Collins, A.E. and Manatsa, D., Indigenous
knowledge, coping strategies and resilience to floods in Muzarabani,
Zimbabwe. International Journal of Disaster Risk Reduction, 5 pp. 38–48,
2013.
[13] Nguyen, K.V. and James, H., Measuring Household Resilience to Floods: a
Case Study in the Vietnamese Mekong River Delta. Ecology & Society, 18
(3), 2013.
[14] Dun, O., Migration and displacement triggered by floods in the Mekong
Delta. International Migration, 49 (1), pp. 200–223, 2011.
[15] Chhotray, V. and Few, R., Post-disaster recovery and ongoing vulnerability:
ten years after the super-cyclone of 1999 in Orissa, India. Global
Environmental Change, 22 (3), pp. 695–702, 2012.
[16] Kick, E.L., Fraser, J.C., Fulkerson, G.M., McKinney, L.A. and De Vries,
D.H., Repetitive flood victims and acceptance of FEMA mitigation offers:
an analysis with community–system policy implications. Disasters, 35 (3),
pp. 510–539, 2011.
[17] Keogh, D.U., Apan, A., Mushtaq, S., King, D. and Thomas, M., Resilience,
vulnerability and adaptive capacity of an inland rural town prone to flooding:
a climate change adaptation case study of Charleville, Queensland,

Australia. Natural Hazards, 59 (2), pp. 699–723, 2011.
[18] Biswas, A., Rahman, A., Mashreky, S., Rahman, F. and Dalal, K.,
Unintentional injuries and parental violence against children during flood: a
study in rural Bangladesh. Rural & Remote Health, 10 (1), 2010.
[19] Linnekamp, F., Koedam, A. and Baud, I., Household vulnerability to climate
change: Examining perceptions of households of flood risks in Georgetown
and Paramaribo. Habitat International, 35 (3), pp. 447–456, 2011.
[20] Munji, C.A., Bele, M.Y., Nkwatoh, A.F., Idinoba, M.E., Somorin, O.A. and
Sonwa, D.J., Vulnerability to coastal flooding and response strategies: The
case of settlements in Cameroon mangrove forests. Environmental
Development, 5 pp. 54–72, 2013.
[21] Pelling, M., What determines vulnerability to floods; a case study in
Georgetown, Guyana. Environment and Urbanization, 9 (1), pp. 203–226,
1997.
[22] Marino, E., The long history of environmental migration: Assessing
vulnerability construction and obstacles to successful relocation in
Shishmaref, Alaska. Global Environmental Change, 22 (2), pp. 374–381,
2012.
[23] Manock, I., Islam, R., Hicks, J., Sappey, R.B. and Ingham, V., Perceptions
of institutional and social response to frequent flooding in an Australian rural
town. Australian Journal of Emergency Management, 28 (1) pp. 42–48,
2013.
[24] Lamond, J.E., Proverbs, D. and Hammond, F., Accessibility of flood risk
insurance in the UK: confusion, competition and complacency. Journal of
Risk Research, 12 (6), pp. 825–841, 2009.
[25] Laska, S.B., Homeowner Adaptation to Flooding an Application of the
General Hazards Coping Theory. Environment and Behavior, 22 (3),
pp. 320–357, 1990.
[26] Rodriguez-Llanes, J.M., Ranjan-Dash, S., Degomme, O., Mukhopadhyay, A.
and Guha-Sapir, D., Child malnutrition and recurrent flooding in rural
eastern India: a community-based survey. BMJ open, 1 (2), 2011.
[27] Wind, T.R., Joshi, P.C., Kleber, R.J. and Komproe, I.H., The impact of
recurrent disasters on mental health: a study on seasonal floods in northern
India. Prehospital and Disaster Medicine, 28 (03), pp. 279–285, 2013.
[28] Mishra, S., Mazumdar, S. and Suar, D., Place attachment and flood
preparedness. Journal of Environmental Psychology, 30 (2), pp. 187–197,
2010.
[29] Mattei, N., Stack, S., Faris, M., Adeinat, I. and Laska, S., Mitigation of
repetitively flooded homes in New Orleans, Louisiana. Ecosystems and
Sustainable Development VII, 122 pp. 365–378, 2009.
Section 6
State-of-the-art on flooding
damage survey and assessment
(Special session
organised by D. Molinari)
Implementing tools to meet the

Floods Directive requirements:
a “procedure” to collect, store and manage
damage data in the aftermath of flood events
D. Molinari1, M. Mazuran2, C. Arias1, G. Minucci3, F. Atun3
& D. Ardagna2
1
Department of Civil and Environmental Engineering,
Politecnico di Milano, Italy
2
Department of Electronics, Information and Bioengineering,
3
Department of Architecture and Urban Studies,
Abstract
The aim of this paper is to present a “procedure” to collect and store damage data
in the aftermath of flood events. The activity is performed within the
Poli_RISPOSTA project (stRumentI per la protezione civile a Supporto delle
POpolazioni nel poST Alluvione), an internal project of Politecnico di Milano
whose aim is to supply tools supporting Civil Protection Authorities in dealing
with flood emergency. Specifically, the aim of this paper is to discuss the present
implementation of the project, highlighting challenges for data collection,
storage, analysis and visualisation. Data can have different formats (e.g. paper
based vs. digital form, different digital files extensions), refer to different aspects
of the phenomenon (i.e. hazard, exposure, vulnerability and damage), refer to
different spatial and temporal scales (e.g. micro vs. meso scale, different phases
of the flood event) and come from different sources (e.g. local authorities, field
surveys, crowdsourcing). Therefore a multidisciplinary approach which includes
expertise from ICT, geomatics, engineering, urban planning, economy, etc. is
required. This paper first describes a conceptual map of the issue at stake, then it
discusses the state of the art of the implementation, taken as reference the
Umbria flood in November 2012. Impacts of the project are discussed with
doi:10.2495/FRIAR140181
particular attention to their utility to meet some of the “Floods Directive”

(Directive 2007/60/EU) requirements: (i) to create a reliable and consistent
database; the latter should be the basis on which damage models can be
defined/validated and thus risk can be mapped; (ii) to supply a comprehensive
scenario of flood impacts according to which priorities can be identified during
the emergency and recovery phase.
Keywords: flood risk management, flood damages, disaster databases, flood
damage maps.
1 Introduction
In recent years, awareness of a need for more effective disaster data collection,
storage, and sharing of analyses has developed in many parts of the world, also
in the wake of several policies that, at different levels of government, implicitly
or explicitly required to face the problem at stake (e.g. the Hyogo framework for
Action [1], the EU disaster prevention framework [2], the European Union
Solidarity Fund [3], the Green Paper on Insurance of Natural and Man-made
Disasters [4]).
Among natural disasters, this paper focuses on floods. Having more reliable
data on flood impacts is of paramount importance for improving pre and post
event risk reduction strategies. For instance De Groeve et al. [5] suggest three
application areas for (flood) loss data: loss accounting, disaster forensics and risk
modelling. In the aftermath of flood events, the principal motivation for
recording the impacts of floods is loss accounting. This information is crucial at
different levels of governance/risk management. At the local level, civil
protection and policy makers (i.e. mayors) need loss accounting in order to
identify priorities for the emergency and the recovery-reconstruction phases
while insurers use this information to compensate victims. At the sub-
national/national level, loss accounting is required by policy makers for fund
allocation, for addressing damage compensation and recovery. At the
international level the interest is on financial and humanitarian aid.
In peace time, flood loss data are required to improve knowledge of the
mechanisms leading to flood impacts; to analyse the causes of disasters through
measuring relative contribution of hazard, exposure, vulnerability and coping
capacity (i.e. the response to the flood). This is what is called disaster forensic.
The objective of disaster forensic is twofold: (i) to enhance disaster management
from lessons learnt, and (ii) to improve risk mitigation strategies by increasing
the capacity of modelling and forecasting flood damage.
Within this context, this paper presents the Poli-RISPOSTA project
(stRumentI per la protezione civile a Supporto delle POpolazioni nel poST
Alluvione), an internal project of Politecnico di Milano supporting
interdisciplinary research with a direct impact on the society. The main intention
of Poli-RISPOSTA is to build with and for the Civil Protection (CP) a model,
tools and advanced technical solutions for collecting, mapping and evaluating
post-flood damage data. In fact, as the consequence of the policies discussed
above, the need for enhanced methods and procedures for post-event damage
assessment has been increasingly demanded also by Italian local authorities.

Moreover, requirements imposed by the European “Floods” Directive
2007/60/EC play a major role for flood risk.
In order to protect people and assets from the impact and consequences of
floods, the EU Floods Directive requires that flood risk management plans
(FRMPs) will be based not only on various flood hazard scenarios but also on
risk assessments which must present the potential adverse consequences of
floods for “human health, the environment, cultural heritage, and economic
activity.” There is then an increasing need to obtain more reliable, high quality
flood impact data that can serve numerous purposes:
(i) to create a reliable and consistent database; the latter should be the basis
on which damage models can be defined/validated and thus risk can be
mapped. Flood risk maps are, in their turn, the main tools on which FRMPs
are defined.
(ii) to supply a comprehensive scenario of the consequences of floods
according to which priorities can be identified during the emergency and the
recovery phases.
Poli-RISPOSTA wants to address these needs.
In order to provide appropriate solutions, one of the key principles of the
project is working with stakeholders, not for them. Stakeholders involvement has
been increasingly demanded by both European policies on risk mitigation (i.e.
the quoted EU Floods Directive is emblematic for the problem at stake) and
research project calls at both national and international levels (see e.g. the FP7
program) but there is a significant difference between interviewing stakeholders
to obtain feedback on work that has already been carried out in research centres
and developing tools and methods jointly. The last modus operandi is, in
authors’ opinion, the only way to get efficient and feasible solutions to the
problem at stake. Accordingly, stakeholders will be actively involved during
the entire project by means of meeting, participatory activities, exercises, etc.
Last but not least, it is worth noticing that Poli-RISPOSTA is an
interdisciplinary project where experts from several fields are involved as ICT,
geomatics, engineering, urban planning, economy, etc.
The conceptualisation of the problem addressed, the approach followed in the
project as well as challenges of Poli-RISPOSTA (also in terms of required
expertise) are discussed more in detail in section 2. Section 3 describes the
current state of the project: challenges regarding data acquisition and
requirements that are needed to gather data before, during and after a flood are
identified. Section 4 concerns the steps required for the remaining part of project.
The paper ends with conclusion including deductions coming from the paper.
2 Problem conceptualisation
The general objective of Poli-RISPOSTA can be identified in the development of
a “complete” flood scenario describing both the physical features of the forcing
event (i.e. the flood) as well as its impacts and the capacity of societies to face
them. In order to accomplish with risk mitigation objectives (e.g. those imposed
by the EU Floods Directive) such a scenario must be developed both ex-ante and
ex-post. Before an event occurs “complete” scenarios provide a picture of the
most risky areas and allow identifying strategies to mitigate risk and to cope with
hazardous events. After the event occurrence, the objective is instead to figure
out real impacts and to identify priorities for the emergency and the recovery
phase (fig. 1). A comparison of the ex-ante and ex-post scenarios allows finally
to infer lessons towards an improvement of both the capacity of predicting the
event (and its consequences) and to cope with it.
To achieve these goals, tools and advanced technical solutions to collect,
store, analyse and represent a multitude of data must be developed within Poli-
RISPOSTA (fig. 1). After an event occurs, such data regard both physical effects
of the forcing event (as flooded areas, water depth and velocity inside it, the
occurrence and localisation of landslides, etc.) and observed damages on the
different sectors of the society (i.e. people, economic, and human activities),
the natural and built environment (i.e. residential and industrial buildings,
infrastructures, public and cultural heritage, ecosystems). Damages can be due to
the physical contact of the flooding water (i.e. direct damages) or induced by the
first (i.e. indirect damages); both (ii) tangible (i.e. monetary) and intangible data
must be taken into account. Moreover, data on mitigation actions implemented
by emergency services and lay people before and during the flood is of interest
as these actions influence both physical effects and damages.
Before the event, data regard instead results from hazard, exposure and
vulnerability modelling. Also in this case, information must be managed with
respect to the different variables characterising the physical scenario as well as
required to estimate risk on the different items potentially affected by the floods.
PEACE TIME
Exposure Vulnerability Hazard modeling
modeling modeling
Flood scenario (ex-ante) Physical scenario

EVENT
Flood scenario (ex-post) Physical effects
(flodeed areas, water
depth, landslides, etc.)
Observed damage (direct/indirect,

tangible/intangible) to exposed
Implemented
sectors
(mitigation) actions
EMERGENCY/RECOVERY
Figure 1: Objective of Poli-RISPOSTA and data of interest.
Poli-RISPOSTA objectives can be achieved by developing of an Information

System (IS) to coordinate and support all the activities earlier described. With
respect to this, a cyclic process is followed in the project (see fig. 2). First,
features of data to be managed and required elaborations to develop complete
scenarios are analysed in order to identify critical aspects to be managed by the
IS; such criticalities are translated into requirements for the IS. The next step is
the design and the development of a prototype which keeps into account
identified requirements. The application of the prototype to a real case allows to
update data analysis and corresponding IS requirements; according to this the
prototype is revised for new applications and tests. In this way, the final product
will be created based on step-by-step refinements, also according to new data
and analyses which could be available/required after the first development of the
IS.
Flood
analysts
Flood
Flood analysts/
analysts ICT experts
ICT experts Flood

analysts/
ICT experts
Figure 2: The cyclic process adopted in the Poli-RISPOSTA.
The complexity of the problem at stake implies several challenges for Poli-
RISPOSTA, with respect to the current state of the art. In the following the most
relevant are discussed.
First, tools for systematic loss accounting are not very well developed. The
way in which flood damage data are presently collected and stored implies
several problems for an efficient, multipurpose use of data as wished in [5] and
[6]. The main problem of existing disaster databases concerns data comparison
and management. This is due to a lack of agreed standard to collect and store
damage data. Specifically, several differences can be found in existing databases
regarding:
- recorded losses. This depends on: (i) the intent of the reporting activity (i.e.
insurance companies, governmental agencies and NGOs collect data for
different purposes; for this reason, flood loss records are often not
representatives of the real impact of floods as they focus only on certain items
at risk and/or types of damage), (ii) the time of reporting, and (iii) present
capacity of estimating all types of damages (e.g. indirect or secondary damages
are not so evident in the aftermath of an event and are difficult to evaluate in
monetary terms).
- The scales of reporting. Flood loss data can be recorded at different spatial and
temporal scales, according to the intent of the report and to who is leading the
reporting activity (i.e. their role and responsibility). However,
aggregating/disaggregating damage over space and time is not straightforward.
- The economic rationale. There are different methods to evaluate monetary loss,
e.g. taking into account inflation (i.e. depreciated value), purchasing parity (i.e.
replacement value), insured losses, etc.
One of the main challenges Poli-RISPOSTA has to face is to develop tools
for the survey and collection of flood loss data that overcome the above limits,
guaranteeing high quality, consistent and reliable data, in the philosophy that
“the quality of disaster databases can only be as good as the reporting system”
[7]. Contrary to common practice, Poli-RISPOSTA wants to work at the local
level in order to meet two basic requirements of flood loss data: (i) going into
details of phenomena/aspects leading to damage and (ii) reporting all the events,
including small ones (like multi spot flash floods in mountain regions) which are
presently discounted by national/international databases [8][9]. Data at upper
levels, for strategic and policy making purpose, can be obtained in a second step
by proper aggregation rules. On the other hand, Poli-RISPOSTA wants to
provide a “complete picture” of a disaster, identifying damage to various sectors
of the economy and society. From this perspective, the PDNA - Post Disaster
Needs Assessment methodology resulting from the collaboration of a number of
institutions, including the EU Commission, United Nations, the World Bank and
others is a very important example (for an application see [10]).
Linked to the previous point is the development of technological solutions for
data acquisition. Indeed, while damage data at the meso or macro scale can be
inferred from indirect sources (e.g. public accounting, researches, newspapers,
and regulations), local data are often collected by means of field surveys. Tools
should then be developed in order to support data survey in digital format. Such
tools should provide real time data storage (in a database) and their visualisation
in terms of maps, supporting this way the field survey/emergency phase (e.g.
supporting the coordination of survey team). With respect to this, the DARMsys
developed by the Queensland Reconstruction Authority in Australia can be taken
as reference [11].
The need of managing collected data also in terms of visualisation and spatial
analysis represents another challenge for Poli-RISPOSTA. Since there is not a
standardized way to collect spatial data in the case of floods, data collected in the
aftermath of flood events are commonly in different formats that make data not
immediately usable for spatial analysis. This is not the case, e.g., in ex-ante risk
assessments where data of interest are directly produced to be handled by GIS
tools [12]. Creating flood databases is common practice, but not always in
GIS standard compatible formats. This point must be addressed by Poli-
RISPOSTA, providing tools for data storage which also satisfy requirements of
data visualisation and spatial analysis.
Last but not least, as any technological problem, an interdisciplinary approach
is required by the project that put together computer scientists and domain
experts.
With respect to this, the further challenge of Poli-RISPOSTA is that not only
domain experts come from different disciplines but also computer scientists are
heterogeneous. As regard domain experts, the need of analysing both the
physical feature of the event as well as its consequences (both in monetary terms
and with respect to intangible damages) implies expertise from engineering,
urban planning, sociology, economy, etc. With respect to computer sciences ICT
experts are required for the collection, storage and management of data of
interest. Moreover, the need of representing data in terms of maps as well as of
carrying out spatial analyses (see section 2) requires involving experts from
geomatics.
3 Present implementation of Poli-RISPOSTA

The project is currently evolving towards the third step of the cycle in fig. 2. A
significant effort has already been put into the first two steps that represent a
crucial part of the cycle since their outputs are the starting point for the
development of the IS. Therefore, a good analysis both of the domain and of the
requirements is fundamental. Particularly at this stage of the project, the
interdisciplinary approach discussed above is needed to combine the two areas of
expertise; flood analysts and ICT experts have to collaborate side by side to
define the most complete picture of the application scenario. The better the result
is, the better the basis for the implementation of the IS will be. Indeed, during the
lifetime of the IS, new needs and requirements might come up that need to be
incorporated into the system. Therefore, we need an agile approach where the
feedbacks gathered from the running system are used to improve and refine its
specifications (i.e. the cyclic approach shown in fig. 2).
In the next two subsections the results of the first two steps are discussed in
details. Results highlight the complexity of the problem at stake.
3.1 Data analysis
According to the cyclic process described in fig. 2, the first step in designing the
IS consists of an analysis concerning both data characteristics and types of
required elaboration to be performed. This analysis was carried out on the basis
of the flood event occurred in the Umbria Region – Central Italy in November
2012 [13]. On that occasion the regional CP asked Politecnico di Milano to
develop a report (under construction) describing the event and its consequences
at the regional level. Researchers activity focused on two aspects: (i) the
development of an ex-post scenario to help CP to figure out event impacts [14],
to identify priorities for recovery and reconstruction and to verify effectiveness
of emergency plans; (ii) the development of an ex-ante scenario to be compared
with the first to verify whether or not existing risk assessment and mitigation
strategies are suitable to deal with flood risk in Umbria. Indeed, results from this
experience are the starting point of Poli-RISPOSTA, allowing both to recognise
needs and requirements in terms of data analysis and elaboration for the scenario
development and to analyse the features of (available or required) data on which
such activity should be performed (see section 3.2).
With respect of data features, besides the fact that they refer to the different
domains discussed in section 2 (i.e. risk modelling, observed physical effects and
damage on the different sectors, mitigation actions), other important features
were recognised which imply requirements for the IS (see section 3.2);
availability in time is one of them. Data are available at different times. This is
due, on the one hand, to the nature of the data itself; for example, modelling data
are available before an event occurs, indirect damages (e.g. disruption of
economic activities, of basic services to the population, the loss of rental income)
are not evident in the aftermath of an event but some months later, etc. On the
other hand, norms regulating damage compensations count. The latter identify
which damages are refunded by law and which are the deadlines to ask for
compensation; for this reason, both public and private subjects give priority to
determine reimbursable damages while other types of losses are assessed in a
second step (e.g. in Italy damage to infrastructures must be declared by regional
authorities 20 days after the event while damage to residential buildings 90 days
after). However, generally speaking, data of interest are available before the
event or after the event, in a time window ranging from few days to 1 year.
Another important feature is the spatial scale; data can refer to individual
objects (e.g. damage to a building, a bridge, a fabric), the local scale (e.g.
number of evacuees in a municipality), the large scale (e.g. traffic disruption at
the province, flood zones in the river basin) or to the regional
/national/international scale (e.g. indirect damage to ecosystems). The Euclidian
dimension is also linked to this feature and to the need of representing data
(analyses) in terms of maps (see below). Data can be represented as points (e.g.
the damage to a building), lines (e.g. length of damaged roads) or areas
(e.g. flooded areas).
The source of data is another important aspect. Some data are acquired by
means of field surveys (e.g. direct damage to buildings, water depth inside the
flooded area) for which suitable tools should be designed (see sections 3 and 4).
Other data are directly produced by the CP (e.g. flood forecasts); finally several
data are recorded by other subjects (e.g. local authorities, service suppliers,
research centres) and must be “simply” collected by the CP.
The present data format is an additional characteristic to be taken into
account. Coming from different sources, data can have different formats: papery
based or digital. In the second case, recognised formats are heterogeneous:
features, texts, spreadsheets, images or multimedia.
Last but not least, not only quantitative (e.g. observed damage) but also
qualitative data (e.g. vulnerability features, emergency actions) are of interest.
In order to reproduce the complete event scenarios, the experience in the
Umbria region highlighted that data of interest not only are heterogeneous but
should also be managed in several ways. Most common analyses regard

aggregation/disaggregation in space and time (e.g. determine damage over a
municipality or one months after the event), filtering (e.g. evaluate damage to a
specific sector), visualisation (e.g. in terms of graphs, diagrams, maps) and
spatial analysis (e.g. determine the water depth at certain location, compare
damage occurring in different municipalities).
In the next section requirements for the IS linked to both data features and
required elaborations are discussed.
3.2 System requirements
Data heterogeneity implies several challenges for the IS development. To

manage data effectively (in order to produce those elaborations which are
required to develop a complete event scenario), criticalities highlighted in section
3.1 were translated into system requirements, according to the cyclic process in
fig. 2:
1. Temporal tracking and storage of data: some data are available before the
event, some are gathered during, other are collected after it; thus the system
should allow to store and manage data collected at different times. As some
information might deteriorate quickly after a flood event (e.g. the level of water,
the memories of people affected by the event, etc.), the system must recall users
on data to be recorded at each time. Moreover, for some data the interest is in
keeping their history over time (that is, the way they change over time), for
others only their current (or most updated) status needs to be known. Criteria
must be defined in this regard to be embedded in the IS.
2. Data aggregation/disaggregation: Data are gathered at different scales; an
approach is required to identify rules according to which data must be
aggregated/disaggregated to guarantee information coherence.
3. Data redundancy prevention: most of data come from different sources; this
reflects in several issues. In fact, data are gathered in many different formats
(spreadsheets, documents, audio, video, etc.) that provide information of
invaluable importance that could also be repeated. The system should define
criteria according to which data are stored or not (e.g. quality of data, time of
acquisition, source reliability, etc.).
4. Data pre-processing: Data can come in several formats that are not necessarily
compatible with their storage in a database or for spatial analysis. The IS can
support only pre-defined format(s). Accordingly, procedures for data pre-
processing (for users) must be defined. Likewise, it is important to support the
process of structuring (when possible) and organizing that information that is
semi-structured or unstructured (e.g. pictures, drawings, audio files and so on).
5. Data acquisition: Some data are collected by means of field surveys. A tool
should be developed to support data survey in digital format (e.g. tablet). Other
data must be simply collected from other sources; accordingly the IS must allow
data acquisition from different sources/users (see the next point).
6. Multi-owners environment: Different users will use the IS in different way
(e.g. to insert data, to analyse data, to visualise data elaboration). Possible users
must be identified as well as allowed actions for each users. This implies to
create different user permissions in the IS (also in remote).
7. Data management: The IS must support several data analysis:
aggregation/disaggregation, visualisation, filtering, querying, etc. both during the
collection/survey phase and at the end of this activity. Pre-defined tools for data
analysis must be developed within the IS to facilitate/make quicker the scenarios
development.
4 Next steps
Coherently with the cyclic process in fig. 2, next steps of Poli-RISPOSTA are
towards the implementation of the IS; in particular, the first milestone concerns
the development of a first IS prototype. The design and implementation of the
prototype will be performed in close collaboration with the CP by means of
participatory processes, exercises, etc. according to the project philosophy
which considers the involvement of stakeholders as key to get efficient and
feasible solutions.
Next efforts of Poli-RISPOSTA can be grouped specifically into three main
activities which are all required to develop the IS:
1. Database design and development. The IS is supported by a database where

all the collected data are stored. Therefore, it is important to define its structure
in order to be able to manage all the data introduced so far in a flexible way, so
to support future changes or updates. The database must keep into account all the
features of the data and the kind of manipulations to be performed on them, thus,
it is designed according to the data and requirements analysis introduced in
Section 3. Particular effort will be put into the depiction and management of
geographical data that play a key role in the project.
The first version of the database will be designed and developed according to the
data available from the 2012 Umbria flood, however, future events will be used
to enrich data and requirement analysis and as a consequence to improve the
design and development of the database itself.
2. Providing the software for data management. According to data requirements
identified in section 3.2, software tools are needed with a twofold aim. On the
one hand, tools for data acquisition will be developed. Such tools will support
both the direct survey of flood damage data on the field and the collection of data
from other sources by the CP. In both cases the software should embed/match
with tools and procedures presently adopted by CP to perform data collection.
On the other hand, tools for the reconstruction of the flood scenario must be
provided. This means to supply the software required to analyse, interpret and
represent collected data. It is worth noticing that the two “objectives” are not
disconnected. In fact, a first reconstruction of the scenario is required in the first
hours after the event to make available the identification, even with limited
precision, of the flooded areas and affected items. This information will be used
during the collection/field survey of data to identify and track investigated items.
This activity will be performed taken as reference data elaboration

produced/required after the November, 2012 flood in Umbria. At the same time
the structure of the DB developed during activity 1 will be carefully taken into
consideration. Actually, the development of the DB and of the software must be
considered as complementary activities, requiring continuous interaction and
feedbacks. Accordingly, the two activities should be done in parallel rather than
one after the other.
3. Definition of a procedure to carry out data collection and elaboration. The
last activity consists of the definition of a procedure to help practitioners to use
the IS, in terms of both data collection and analyses. With respect to the first
point, guidelines will be produced specifying which data should be collected,
when, at which scale, by whom, in which format, etc. Regarding the second
aspect, a framework will be drawn up detailing steps, data analyses and
elaborations required to produce a complete event scenario. The objective is to
create a procedure to be adopted by CP as a standard in case of flood, to make
easier and quicker the analysis of the event. Also this last activity is strictly
interconnected with the others and must be carried out in parallel with them.
The prototype that will be produced at the end of the three activities will be
tested during a CP exercise in the Umbria region in autumn 2014. The test will
allow to possibly modify IS requirements (and corresponding features) as
described in the iterative process in fig. 2. The evolution of the IS according to
test results will be the objective of next research efforts.
5 Conclusion
The objective of this paper is to present the Poli-RISPOSTA project, an
interdisciplinary project of Politecnico di Milano providing novel and enhanced
methods and procedures for post-flood damage assessment. The latter are a key
prerequisite for improving pre and post event risk reduction strategies as required
(among the others) also by the EU Floods Directive. Having more reliable flood
loss data is of paramount importance for loss accounting, disaster forensic and
risk modelling.
Efficient solutions imply the use of advanced technological tools; for this
reason an interdisciplinary approach is required that put together expert’s domain
(i.e. flood analysts) and computer scientists.
By describing the current level of implementation of the project, the paper wants
to highlight two peculiarities of the problem at stake. On the one hand, its
complexity both in terms of methodological gaps, data to be handled,
elaborations to be performed and the variety of expertise which is required. On
the other hand, the need to work with stakeholders (i.e. the users of developed
tools) to get feasible and effective solutions.
Acknowledgements
The authors acknowledge all the people involved in the Poli-RISPOSTA project
for their useful feedback on the paper. Authors also acknowledge with gratitude
the Umbria Region Civil Protection authority (and its staff), which strongly
encourages/actively takes part in this research.
References
[1] ISDR, Hyogo framework for Action 2005-2015: Building the resilience of
nations and communities to disasters, http:\\www.unisdr.org/wcdr, 2009.
[2] Council of the European Union, Council Conclusions on a Community
framework on disaster prevention within EU. 2979th Justice and Home
Affairs Council meeting. Brussels, 30 November 2009.
[3] Council Regulation (EC) No 2012/2002 of 11 November 2002
establishing the European Union Solidarity Fund.
[4] EC, 2013. Green Paper on the Insurance of Natural and Man-made
Disasters. COM/2013/0213 final.
[5] De Groeve, T., Poljansek, K. & Ehrlich, D., Recording Disaster Losses.
Recommendations for a European Approach, JRC Scientific and Policy
Report. Report EUR 26111 En, 2013.
[6] Wirtz, A., Kron, W., Low, P. & Steuer M., The need for data: natural
disasters and challenges of database management, Nat Hazards, 70,
pp. 135-157, 2014.
[7] Guha-Sapir D. & Below R., The quality and accuracy of disaster data. A
comparative analyses of three global data sets. ProVention Consortium
(World Bank), 2002.
[8] Llsat et al., Towards a database on societal impact of Mediterranean floods
within the framework of the HYMEX project, Nat. Hazards Earth Syst.
Sci, 13, pp. 1337-1350, 2013.
[9] Mysiak, J., Testella, F., Bonaiuto, M., Carrus, G., De Dominicis, S.,
Ganucci Cancellieri, U., Firus, K. & Grifoni, P., Flood risk management in
Italy: challenges and opportunities for the implementation of the EU
Floods Directive (2007/60/EC), Nat. Hazards Earth Syst. Sci., 13,
pp. 2883-2890, 2013.
[10] Wergerdt, J. & Mark, S.S., Post-Nargis Needs assessment and monitoring.
ASEAN’s Pioneering Response, Final report, Asean Secretariat, 2010.
[11] Queensland Reconstruction Authority, Australia,
http://qldreconstruction.org.au/about/darmsys
[12] Jonkman, S.N., Bočkarjova, M., Kok, M. & Bernardini P., Integrated
hydrodynamic and economic modelling of flood damage in the
Netherlands, Ecological Economics, 66, pp. 77-90, 2008.
[13] Servizio Protezione Civile – Regione Umbria. Evento alluvionale 11-14
Novembre 2012: Rapporto di evento. Available on line at:
www.cfumbria.it
[14] Molinari, D., Menoni, S. , Aronica, G.T., Ballio, F., Berni, N., Pandolfo,
C., Stelluti, M., Minucci, G., Ex-post damage assessment: an Italian
experience, Nat. Hazards Earth Syst. Sci (accepted for publication).
Flood damage survey after a major flood in

Norway 2013: cooperation between the
insurance business and a government agency
H. Berg1, M. Ebeltoft2 & J. Nielsen2,3
1
Norwegian Water Resources and Energy Directorate, Norway
2
Finance Norway, Norway
3
Norwegian Natural Perils Pool, Norway
Abstract
Results from cooperation between the insurance business and the Norwegian
Water Resources and Energy Directorate on a flood damage survey after a major
flood in Norway 2013 is presented, as well as results from similar cooperation
after a flood in 1995. Benefits for flood risk management of including flood
parameters in future damage surveys are presented.
Keywords: damage survey, flood, natural hazard, insurance, risk management.
1 Introduction
Data from flood events are collected by different stakeholders for different
purposes. This paper presents results from cooperation on flood damage survey
after a major flood in Norway in May 2013, between the insurance business and
the Norwegian Water Resources and Energy Directorate (NVE). The idea is that
the survey made by the insurance business for their purpose potentially could
provide valuable information for other purposes within flood risk management. In
the first instance the idea was to improve the basis for damage functions relevant
for Norway. The paper expands on this and presents ideas for future collection of
data.
2 Flood risk management in Norway

A White Paper with the title “How to live with the hazards” was issued by the
Government in 2012, White Paper no. 15 [1]. This outlines the national policy in
doi:10.2495/FRIAR140191
dealing with floods and landslides. The Government states that it will continue its
efforts in preventing damage from floods and landslides according to a holistic
approach including mapping, land use planning, protection measures, monitoring,
early warning, contingency and crisis management. The Norwegian Water
Resources Directorate (NVE) is the agency at directorate level responsible for
coordinating the implementation of the national policy. NVE’s work in preventing
damage from floods and landslides is structured according to the holistic approach,
in the following tasks:
- Hazard and risk mapping
- Assistance and control of land use planning in the municipalities
- Planning and construction of structural protection measures
- Monitoring and early warning: Floods, Debris flows, Snow avalanche
- Assistance to the police and municipalities in emergency situations
- Research & Development, Communication
The White Paper [1] outlines how responsibilities for dealing with floods and
landslides are distributed among the main actors. The importance of cooperation
between the relevant actors is highlighted, and a national strategy for cooperation
and coordination will therefore be developed.
Every municipality is obligated according to the Civil Protection Act to
perform an overall Risk and Vulnerability (RAV) analysis for its territory as a
basis for preparedness to deal with harmful events and for land use planning. The
municipality is responsible for making sure that natural hazards are being
evaluated and taken properly into account in every new development scheme,
according to the Planning and Building Act.
3 Natural hazards insurance

In order to limit the losses to private stakeholders different compensation systems
are established. One of these is the insurance against natural hazards. In Norway
all buildings insured against fire are automatically insured against natural hazards
such as floods, landslides and storms, according to the Natural Hazards Insurance
Act. The system is based on a solidarity principle as the premium is based on the
value of property and not differentiated according to risk. Damage to the building,
its content as well as the garden and the courtyard adjacent to the building is
covered. Insurance companies offering fire insurance are mandatory members of
the Norwegian Natural Perils Pool (NNPP). The insurance companies have all
contact with their policy holders, whereas the pool equalizes losses between the
companies. The administration of the pool is run from a separate Pool office within
Finance Norway.
Finance Norway (FNO) is the federation for banks, insurance companies and
other financial institutions in Norway. Finance Norway fulfills both the business
policies and employer-related cooperation in the financial sector. It is part of
FNO’s climate strategy to work with public authorities in the prevention of
damage caused by increased frequency and intensity of natural events. This
includes understanding the risk implicated by the Intergovernmental Panel on
Climate Change (IPCC) climate scenarios.
4 Damage survey after major flood 1995 – development of

stage-damage functions
Internationally substantial work has been done on establishing stage-damage
functions for floods, i.e. relations between water level during a flood and damage.
There has not been much work on damage functions based on data from Norway.
The residential houses in Norway outside city centers are to large extent wooden
constructions. Typical residential houses before, during and after flooding is
shown in figure 1. It is important to develop new or test existing damage functions
on data from Norway.
(a) (b)
(c)
Figure 1: (a) A typical residential house; (b) flooded houses during a flood in
1995; (c) the interior of a house during repair after flood damage. (All
photos: NVE.)
After a major flood in South Eastern Norway in 1995, cooperation between the
Norwegian Natural Perils Pool (NNPP) and NVE was established in order to
collect data on water level in buildings and the corresponding damage. The
surveyors for the insurance companies were asked to register maximum water
level in buildings relative to ground floor level. Wathne et al. [2] developed stage-
damage functions based on a limited part of the data set (as shown in figure 2).
Some years later more of the data from the survey in 1995 was systemized and
used by Gottschalk and Krasovskaia [3] in the Interreg III B project FLOWS. They
established damage functions for different categories of objects, as shown in
figure 3. Some statistical parameters related to the same data set is presented
in table 1.
1200
1000
800
D am age (kN O K )
Constr. yr 1850-1942
600 Constr.yr 1946-94
Trend line 1850-1942
400 Trend line 1946-94
200
0
0 100 200 300 400 500 600
Water level (cm) (ground level = 200)
Figure 2: Individual damage cost for residential houses (detached) in terms of

insurance payouts as a function of water level above basement floor
(from Wathne et al. [2]).
10000
single family res.
secondary buildings
farms
ind./com. buildings
Other
Linear (single family res.)
Linear (secondary buildings)
Linear (ind./com. buildings)
1000
damage cost [kNOK]
100
10
-1 0 1 2 3 4 5
water level from basement floor [m]
Figure 3: Scatter plot of damage for different categories of damaged objects

against local water level in building (from Gottschalk and
Krasovskaia [3]).
Table 1: Statistical parameters of flood damage in kNOK for different

categories of buildings (from Gottschalk and Krasovskaia [3]).
Category All data Single Secondary Public Farms Industrial- Other

family houses buildings Commercial buildings
residential buildings
houses
total number
of cases 2296 607 211 97 649 359 373
number of
cases with
complete data 1420 367 161 69 432 218 173
mean 570 311 160 1116 805 914 268
median 142 93 93 521 281 184 81
std. dev. 1170 678 184 1490 1333 1734 619
minimum 0.1 1.4 3.5 11 0.1 0.1 795
maximum 15655 5512 1029 10535 6803 15655 5512298
coeff. var. 2.052 2.180 1.149 1.335 1.656 1.897 2.308
skewness 4.758 4.737 2.217 3.995 2.821 4.524 5.628
5 Damage survey after the flood of 2013

5.1 Registration of flood parameters
A major flood occurred in Norway late May 2013. The most severe flood and
subsequent damage occurred in River Gudbrandsdalslågen and its tributaries.
Figure 4 show photos of damaged buildings in the village Kvam. After the event
the insurance companies immediately starts the process of assessing damage as
basis for the compensation to the policy holders. For this the insurance companies
hire surveyors with relevant education and experience.
Shortly after the flood, contact was established between NVE, Finance Norway
and the NNPP. An agreement was made to include data on water levels in the
survey, similar to what was done in 1995. Ad hoc a form was developed for the
Figure 4: Damaged buildings in Kvam, June 2013. (Photos: NVE.)
purpose and circulated to the surveyors by the NNPP, with accompanying

instructions. The cost related to filling in the forms was included in the survey cost
and covered by the insurance companies.
The form included a table with parameters related to the flood (see table 2). To
be able to link this information to the rest of the survey, the following information
was requested: surveyor name, survey no., insurance company, location and
municipality no.
Table 2: Flood parameters included in registration form used by insurance

surveyors.
Building no. ID or type of building

Water level cm +/- relative to ground
floor level
Basement? yes/no
Erosion, under-mining of building? yes/no
Mass deposition outside of the building? yes/no - thickness
Damage due to floating objects etc. hitting the yes/no
building?
Supplementary information
5.2 Preliminary results
As of February 2014 a total of 243 cases have been reported by the surveyors. The
processing of these data is not yet completed as the compensation process is still
ongoing in a substantial part of cases. Compensation paid to the policy holders so
far span from more than 1 Mill NOK to 6000 NOK.
Some preliminary figures concerning the data from the forms could still be
presented. Concerning water level the following data appear:
1- 270 cm below ground floor level: 89 cases

1- 170 cm above ground floor level: 85 cases
0: 43 cases
Blank (no value): 27 cases
Further investigation needs to be made into the cases with value “0” or no value.
A preliminary review indicates that supplementary information in the form in
some cases includes information on water level. In other cases it appears that
damage was only to the garden and the courtyard.
Concerning the other parameters in the form, the following results appear as
shown in table 3.
We see from Gottschalk and Krasovskaia [3] that there is a great variability in
the data set. The question is if more factors could be identified to create relations
with less variability. This was the idea behind including more parameters than
water level, such as erosion and mass deposition, in the form used in 2013.
Another obvious parameter to test, is the total value of the building. This is a topic
for further research.
Table 3: Results from damage survey related to the 2013 flood.
Basement? Erosion, Mass Damage due

(yes/no) under-mining deposition to floating
of building? outside of the objects etc.
(yes/no) building? hitting the
(yes/no – building?
thickness) (yes/no)
Yes 117 17 70 7
No 110 214 161 224
Blank 16 12 12 12
6 Damage data in flood risk management

A key element in flood risk management is risk assessments, cost benefit analyses
and other types of analyses. Assessing consequences of events is part of the
analyses. Thus information from historic events provides important input to risk
assessments at all levels.
Damage data at an aggregated level is important for decision making at higher,
strategic levels e.g. to illustrate the size of the challenges in a national perspective.
In this the existing statistics from the NNPP on damage has been important, for
instance as basis for the White Paper no. 15 [1] in 2012. To be able to draw a
complete picture it is important to include all types of costs. There is a lack of
access to similar statistics on damage to public property and infrastructure such as
roads, railroads, power grid, water supply, sewage etc.
This chapter focuses on the benefit of data at a more detailed level and in
particular how data collected by the insurance business could be of value to other
stakeholders in flood risk management. More and better data will improve decision
making and ultimately reduce the damage caused by floods.
6.1 Stage-damage functions
The stage-damage functions developed based on the data from the 1995 flood, is
among others useful for cost-benefit analyses related to flood protection schemes.
To NVE cost-benefit analyses is key input to the decision on governmental
financial support or not.
NVE’s cost-benefit analyses are based on a common concept of risk among
engineers: risk is a product of probability and consequences. The probability part
is usually well covered, for instance through flood mapping. NVE has since 1998
produced flood inundation maps presenting areas prone to flooding with high
precision based on analyses of flood frequency, hydraulic modeling and GIS-
analysis with a detailed digital elevation model. Limited access to data on damage
opens for more subjective judgments of consequences and subsequently greater

variability in quality of the consequence analysis. Hence, improved stage-damage
functions will benefit decision making regarding flood risk by improving the
quality of risk analyses. We assume this reasoning is relevant to other stakeholders
investing in protection measures, such as developers and infrastructure owners.
6.2 Detailed positioning of damage points
Based on the cooperation from the 2013 flood, NVE and Finance Norway has
started a discussion on the possibilities for including registration of flood
parameters as standard in damage surveys by the insurance business. Included in
this is an investigation of the advantages of more detailed positioning of damage
points in map coordinates, including relating water levels to the standard map
elevation basis.
NVE has highlighted that a better positioning will open for a much wider use
of the data, and hence increase the value substantially. All parts of flood risk
management benefit from information on events and improved mapping.
A good positioning of damage will pinpoint areas at risk and thus provide
important input to the municipalities in their overall RAV-analyses. Put together
such point observations could be the basis for flood event maps showing areas
exposed to one particular flood. If more sophisticated mapping has not been
performed, event maps are valuable for land use planning, flood protection and
emergency preparedness.
Observation of water level from actual flood events is important for validation
and calibration of flood models, such as the hydraulic models used in flood
inundation mapping. The access and quality of calibration data in the form of
observed flood levels significantly affects the quality of the maps.
Information from events in itself or via flood maps is the key for taking flood risk
properly into account in land use planning. Similarly the quality of flood maps is
important for the planning of protection measures. Better models could potentially
lead to reduced development cost as safety/uncertainty margins could be reduced.
The models developed in the mapping process are also used during flood
situations as a tool for the crisis management. Better models will accordingly
improve the basis for decisions on measures to be taken during crises.
7 Conclusions
FNO has started a pilot project aimed at clarifying if damage data from storm
water, backwater in sewer systems and natural hazards could be useful for the
municipalities in their work on identifying vulnerable areas and the performance
of RAV analyses. The project is due to deliver its results by the summer of 2014.
The cooperation referred in this paper is not part of the pilot project, but stand
as an example of how data from the insurance business could be used by NVE and
other authorities in the prevention of flood damage. Given that the data are being
used as suggested above, it could contribute to improved decisions regarding flood
risk. Ultimately the result would be reduced damage related to flood events and/
or reduced cost for measures taken.
Registrations of more parameters related to floods, should preferably not be
based on ad hoc initiatives such as in 1995 and 2013, but rather be part of standard
procedures of the survey after a flood event.
Before deciding on this in any direction, more investigation is necessary on a
number of issues:
- How to perform the registration in practice.
- What are the most cost-effective solutions providing sufficient quality
of data
- Format and organization of data.
- Sensitivity of data; what could/could not be published.
Potentially such investigations could be part of a follow-up project. In any case,
the processing and evaluation of the 2013-data will continue and hopefully
contribute to the further investigation of these issues.
References
[1] White paper no. 15. Meld St 15 (2011-2012) Melding til Stortinget. Hvordan
leve med farene – om flom og skred. In Norwegian. Ministry of Petroleum and
Energy, Oslo, 2012.
[2] Wathne, M., Skoglund, M. & Eggestad, H.O. Samfunnskostnader på grunn
av flom i vassdrag. HYDRA report no. R02. Norwegian with English
summary. Norwegian Water Resources and Energy Directorate, Oslo, 1999.
[3] Gottschalk, L. & Krasovskaia, I. Expected damage (risk) of flooding. Interreg
IIB FLOWS report, sub project 1b, Oslo, 2006.
Section 7
Emergency preparedness
and response
An overview of the applications for early

warning and mapping of the flood events in
New Brunswick
D. Mioc1, E. McGillivray2, F. Anton1, M. Mezouaghi2,
L. Mofford2 & P. Tang3
1
National Space Institute, Technical University of Denmark, Denmark
2
New Brunswick Emergency Measures Organization, Canada
3
New Brunswick Department of Environment, Canada
Abstract
This paper gives an overview of the on-line flood warning implementation in the
province of New Brunswick, Canada. The on-line flood warning applications are
available via the “River Watch” website provided by the New Brunswick
Department of Environment. Advanced GIS technology combined with
hydrological modelling, provide a mapping and visualization tool that can be used
by emergency managers and the general public to predict possible flood zones.
The applications developed for “River Watch” support the processing of large
amounts of digital terrain and hydrological data, which are then, quantified and
displayed on digital maps allowing decision makers and the general population to
comprehend and visualize the possible area and impact of the flooding. The
WebGIS applications that are available from the “River Watch” web site provide
snow reports and maps, flood warnings and interactive maps. The searchable
historical database containing reports about the impact of past floods and
estimated damages provides a valuable insight into the past of the province of New
Brunswick and the motivation for development of the system for flood prediction
and management.
Keywords: flood maps, flood prediction, flood management.
1 Introduction
In the province of New Brunswick (Canada), river valleys and flood plains can
pose a risk because of ice jams, harsh weather and floods of annual spring thaw.
doi:10.2495/FRIAR140201
Another danger comes from hurricanes, tropical storms, erosion, or other harsh
seasonal weather events, which may cause tidal and ice surges in coastal areas (see
Figure 1). All of these can cause a threat of flooding with material damage to
people and even the loss of human lives [1].
In Canada, the province of New Brunswick was the first province to join the
Flood Damage Reduction Program signing General, Mapping and Studies
Agreements in March 1976 [2]. The first outcome of this agreement was mapping
of the flood plain, where one in a 100-year flood was used to delineate and
designate flood plains in 13 areas [2].
Figure 1: Historic ice jams in the Stain John river basin (from:
http://www2.gnb.ca/content/dam/gnb/Departments/env/pdf/Water-
Eau/SaintJohnRiverBasin-BassinFleuveSaintJean.pdf).
Within this program, a sub-agreement on structural controls centered on
building sea dykes in the Petitcodiac area was made. This separate agreement for
flood control (where each party assumed one third of the costs in the Marsh Creek
area) was negotiated with the federal and provincial governments and the city of
Saint John. The flood management related works provided by this program
included channel improvements, improvements to outlet control structures, the
construction of a reservoir and the reconstruction of a bridge [2].
Additional studies agreement funded ice research on the Restigouche River and
the international section of the Saint John River [2].
Figure 2: River watch web site (from: http://www2.gnb.ca/content

/gnb/en/news/public_alerts/river_watch.html).
This sub-agreement on flood forecasting followed by the Geoconnections
funded project on flood prediction and mapping helped the province of New
Brunswick to establish a flood forecasting centre for the Saint John River,
including the required technology development and transfer. The River Forecast
Centre (RFC), located in Fredericton, forecasts river levels and produces
interactive near real time flood maps along the Saint John River and its main
tributaries below Fredericton where the major flood damages are experienced in
the province (see Figures 1 and 2). The RFC provides this service on everyday
basis during the spring freshet as well as during flood events following heavy
rainfall [2].
The basic flood facts [3] are presented here:
 A heavy rainfall can result in flooding, particularly when the ground is
still frozen or already saturated from previous storms, for example the
flooding that usually happen in spring time [3], sometimes even with the
ice jams (see Figure 3).
 Flash flooding – in which warning time is extremely limited – can be
caused by hurricanes, violent storms or dams breaking, that became more
frequent in recent years [3] what is attributed to the climate changes.
 Many Canadian rivers experience frequent flooding. The potential for
flood damage is very high if residential or commercial development is
allowed on low-lying, flood-prone lands [3]. The regulation for building
permits should exclude construction on the flood plain.
Figure 3: Ice Jam Flooding: an ice jam in the St. John River caused major
flooding, impacting homes, businesses and public infrastructure in
the Perth-Andover area (from: http://www2.gnb.ca /content
/gnb/en/multimedia/mrenderer .2012.03.2012-03-25_1.jpg.html).
2 The on-line available applications for flood forecasting and

management in “River Watch”
There are several interactive applications available to the users accessing “River
Watch” web site. The general web site (shown on Figure 2) provides the links to
the individual applications about snow (see Figure 4), ice or flood status and
warnings. There is a link to the information about the conditions on the roads
provided by New Brunswick Department of Transport shown on Figure 5. The
additional information about the road accessibility or closure is given in tabular
Figure 4: The Web map application showing the snow depth for the whole
province of New Brunswick (from: http://www2.gnb.ca/content
/gnb/en/news/ public_alerts/river_watch/survey_depth_cm.html).
form directly at the web site [4]. The interactive maps that show flooded areas in
near real time are updated on a daily basis (see Figure 6). In Figure 7, a more
detailed view of the Saint John River watershed is shown, and Figure 8 shows a
detailed daily flood map for the City of Fredericton, with interactive graphs
providing the readings for water gauges along the hydrographic network.
While the flood maps are produced and updated on a daily basis [5],
hydrological modelling [6] provides the possibility for forecasts for the next two
days of the water levels along the Saint John River (shown on the table in
Figure 9).
A service for reports and warnings about ice jams is developed as well (see
Figures 10 and 11). The valuable knowledge about the past floods and their
impacts to the people living in the province of New Brunswick is compiled and
available via the utilization of the historical database. The database provides
search utilities and reports to the users (see Figures 12 and 13).
Figure 5: The roads condition in New Brunswick (from:

http://www1.gnb.ca/0113/en/traffic_advisories/flooding-e.asp).
Figure 6: Interactive flood warning map for the province of New Brunswick
[4].
Figure 7: Interactive flood warning map for the province of New Brunswick,
“zoom-in” for Saint John River watershed [4].
Figure 8: Interactive flood warning map for the province of New Brunswick,
“zoom-in” for the City of Fredericton [4].
Figure 9: Two days forecast for Saint John River (from: http://www2.gnb.ca
/content/gnb/en/news/public_alerts/river_watch/st_john_river_two
-dayforecast.html).
Figure 10: Ice movement warnings (from: http://www2.gnb.ca/content

/gnb/en/news/public_alerts/public_alert.2014.03.0266.html).
Figure 11: Ice jam in Stanley, NB (from: http://www2.gnb.ca/content

/gnb/en/multimedia/mmrenderer.2013.03.2013-18-03_3.jpg.html).
Figure 12: Detailed report about flooding in 2008 (from: http://www.elgegl

.gnb.ca/0001/en/Flood/Details/304).
Figure 13: Historical database of floods in NB (from: http://www.elgegl.gnb

.ca/0001/en/Flood/Search?LocationName=fredericton).
3 Conclusions
An overview of the flood prediction and mapping applications available on-line
from the “River Watch” web site has been presented. The online applications allow
access to flood forecast data and mapping services for ice jams, roads accessibility
or closure and near real time flood plain delineation. The historical database
provides the utilities for search and access to data about past flood events and
damages that were caused. The flood prediction and mapping applications and
other services provided by “River Watch” are accessible to the decision makers
and general public in order to assist them to comprehend the impacts and potential
damages of the flooding.
Acknowledgements
This project was financially supported, in part, by the N.B. Emergency Measures
Organization and by the Canadian Department of Natural Resources
Geoconnections program as well as by University of New Brunswick and New
Brunswick Innovation Foundation (NBIF). The IT Division of the City of
Fredericton and Geological Survey of New Brunswick provided datasets available
for this project. The New Brunswick Department of Environment has provided
data and expertise related to hydrological modelling, and the NB Emergency

Measures Organization helped with their expertise and additional funding for this
project.
References
[1] Hazards in New Brunswick, http://www.getprepared.gc.ca/cnt/hzd/rgnl/nb-
eng.aspx, accessed on-line, March 2014.
[2] Environment Canada, Flood damage reduction program,
https://www.ec.gc.ca/eau-water/default.asp?lang=En&n=B5349463-1,
archived information, accessed on-line, March 2014.
[3] Hazards in New Brunswick - Flood http://www.getprepared.gc.ca
/cnt/hzd/flds-eng.aspx#a1, accessed on-line, March 2014.
[4] River Watch, http://geonb.snb.ca/riverwatch/index.html#, accessed on-line,
March 2014.
[5] Mioc, D., Nickerson, B., Anton, F., Fraser, D., McGillivray, E., Morton, A.,
Tang, P., Arp, J.P. & Liang, G., Web-GIS application for flood prediction and
monitoring, International Conference on Flood Recovery Innovation and
Response, London, WIT Transactions on Ecology and the Environment
(ISBN: 978-1-84564-132-0), WIT Press, 2008, pp. 145-154.
[6] Mioc, D., Anton, F., Nickerson, B., Santos, M., Adda, P., Tienaah, T., Ahmad,
A., Mezouaghi, M., MacGillivray, E., Morton A. & Tang, P., Flood
Progression Modelling and Impact Analysis, Efficient Decision Support
Systems - Practice and Challenges in Multidisciplinary Domains, Chiang Jao
(Ed.), ISBN: 978-953-307-441-2, InTech, 2011, pp. 227-246.
Risk management and emergency response for

a 300 km2 sub-sea level area with a million
citizens against extreme storm surge and flood
due to the “Super Ise-Bay Typhoon”
T. Tsujimoto1, M. Igarashi2 & K. Kobayashi2
1
Department of Civil Engineering, Nagoya University, Japan
2
Chubu Regional Bureau,
Ministry of Land, Infrastructure, Transport and Tourism, Japan
Abstract
There is a land of 300 km2 lower than sea level with a million citizens facing Ise-
bay in the central part of Japan, which is located on a possible route of typhoons
and is exposed to a risk of serious storm surge and flood. This area was attacked
by storm surge by “Ise-bay Typhoon” in 1959 and more than 5,000 people were
killed. In spite of a protection infrastructure constructed in this half decade after
the event, recent climate change may cause extreme typhoons exceeding the
level of protection and response, and resilience against such an enormous
disaster has not yet been prepared. We have made efforts to prepare an action
plan of risk management and emergency response since 2005. Once an extreme
storm surge breaks the protection infrastructure, a wide area will be inundated
with various risks and drainage from there will take a long time. Meanwhile,
with recent progress in weather forecasting of magnitude and course of big
typhoons, we may have a lead time of 36 hrs. We introduce 4 phases: Risk
management before typhoon arrival (Phase 0), emergency response within 0–
72 hrs (Phase I) and successive stages (Phases II and III). In particular, we study
how to make a wide preliminary evacuation possible with proper operation in
Phase 0. We have organized a working group to support the authority including
all the stakeholders related to disaster mitigation to make an action plan of risk
management and emergency response.
Keywords: typhoon, storm surge, catastrophe management, risk management,
emergency response.
doi:10.2495/FRIAR140211
1 Introduction
There is a land of 300 km2 lower than sea level facing the Ise-bay in central
Japan. This area includes a part of Nagoya Metropolis whose population is more
than 2 million, and has developed as an industrial centre with high economic
activity. On the other hand, this area is located on a possible route of typhoons
and major rivers neighbouring this area cause huge flooding from mountain areas.
This area was attacked by a serious storm surge due to a typhoon (Ise-bay
typhoon, “Vera”) in 1959, and 500 km2 was flooded for a few months. More than
5,000 people were killed and the daily lives of a large number of citizens were
affected for a long time [1, 2]. Within around a half decade after this event, we
have completed protection infrastructures against that level of storm surge and
floods due to typhoons.
However, the level of typhoon may exceed the protection level because of
probabilistic phenomena, and recent climate change may cause such a super-
class typhoon. At the news of serious flooding of New Orleans by Hurricane
Katrina in 2005 [3], we learned of the emergency response there [4] and we
started to discuss risk and emergency management for the Ise-bay area. The
situation of this area such as the route of a typhoon and geographical
characteristics can cause high storm surge. There is a wide land below sea level
with a million citizens, and the large number of human activities is very similar
to New Orleans. Moreover, not only the area facing Ise-bay but areas facing
Tokyo-bay and Osaka-bay are in the same situation. In other words, 3 major
metropolises, Tokyo, Osaka and Nagoya, have similar risks of wide and long
term flooding due to storm surge to threaten a large number of lives and human
activities. Certainly the protection infrastructure has been accomplished during
this half decades, but catastrophic disaster may be estimated once a super
typhoon exceeding the level of our present protection level attacks those areas.
We have not prepared an appropriate risk management and emergency response
plan.
New Orleans Area Ise-bay Area Tokyo-bay Area

Nagara R.
Pontchartrain Lake
Kiso River Arakawa R.
City Center
Ibi River Shonai R.
Sumida R. Edo R.
Mississippi River
Tama R. Tokyo- Bay

Ise Bay
400km2, 0.66million 336km2, 0.90million 116km2, 1.67million

Area and Population in Area below Sea Level
Figure 1: Comparisons of some bay areas below sea level.
Just after the event of Hurricane Katrina in 2005, we started to prepare a

response plan against a super typhoon. We postulated “super Ise-bay typhoon” as
a possible maximum typhoon to case the worst storm surge in the Ise-bay, and
based on a scenario of this “super Ise-bay typhoon”, we started to discuss an
action plan against it. At first, we organized an authority named “Tokai
Nederland Regional Authority against Storm Surge and Flood” [5]. Various
stakeholders related to disaster mitigation of this area such as regional and local
governments, police, self-defence forces, Red Cross, water supply and sewage,
telecommunications, energy (electric and gas services) public transport, mass
media, and so on have joined as members of this “TNT authority”. We have a
system that heads of the respective organizations agree to the output of this
system [5]. In the process of making up an action plan (risk management and
emergency response), a working group was organized, and a plan has been made
and revised through extensive discussions and desktop exercises (DIG). In the
working group, parallel sessions for different functions for disaster mitigation or
different viewpoints have been prepared and driven by facilitators from academia
for each session, and a plenary session discusses the output from parallel
sessions to reach a tentative conclusion [6].
The case of a super-typhoon, which will attack this area with a very intensive
magnitude, can be predicted 36 hrs before its arrival. So, we postulate a response
of the action against the super typhoon at that time, and we divide phases as
follows [5–7]:
Phase 0 = 36 hrs before typhoon arrival;
Phase I = 72 hrs after typhoon arrival;
Phase II = 4th day–2nd week,
Phase III = ~1 month.
The main functions for disaster mitigation required for the respective phases
are: wide-area preliminary evacuation in Phase 0; rescue in Phase I; closure of
levee or dike, drainage from flooded area and elimination of obstacles on routes
of rescue and repair in Phase I–II; providing shelters, urgent recovery of life lines
(water, energy, access, telecom) in Phase II–III. Then, various restoration
programs will continue. Among some phases, Phase 0, risk management before a
disaster happens is characteristic in a storm surge and flood disaster due to a
super typhoon. In this phase we have no disaster-control headquarters though we
officially have “emergency response headquarters” after a serious disaster
happens. Most of the emergency responses after a disaster happens are common
among various types of disasters, though urgent closure of levees and drainage
operations are important in the case of flood disasters. Other emergency support
functions cannot be realized after Phase I without the closure of levees and
drainage.
2 Postulated wide area inundation due to super

Ise-bay typhoon
We postulated about a “super Ise-bay typhoon”, which is assumed to be 910 HP
in magnitude (the same as the Muroto typhoon in 1934 and the most intensive
record in Japan) and travels on a route that will cause the severest storm surge on
the Ise-bay. The time path of this typhoon is shown in Figure 2 [5].
Because risky points of coastal levee breach can be estimated based on the
numerical calculation of storm surge (rising of sea level due to astronomic tide,
lift up by pressure drop and waves), levee breaches are assumed there. In
addition, heavy rainfall with return period of 1,000 years is postulated. Then, we
assumed river levee breaches at several places along class A rivers. Such
assumptions are practically familiar in making a “flood hazard map” [5].
Figure 3 shows the flooded area [5] which is around 500 km2 (520 km2 with only
202 km2 caused by storm surge), and almost equal to the actual flooded area on
Super Ise-Bay Typhoon

Time Tidal level at Nagoya Port
Figure 2: Time path of “super Ise-bay typhoon”.
Nagara R.
Ibi R.
Kiso River
Maximum
inundation
Depth (m)
Shonai River
5.0 ~
4.0 ~ 5.0
3.0 ~ 4.0 Nagoya
2.0 ~ 3.0 City
1.0 ~ 2.0
0.9 ~ 1.0
0.8 ~ 0.9
0.7 ~ 0.8
0.6 ~ 0.7
0.5 ~ 0.6
0.4 ~ 0.5 Nagoya
0.3 ~ 0.4
0.2 ~ 0.3 Port
0.1 ~ 0.2
0.0 ~ 0.1 Ise Bay
Figure 3: Inundation due to “super Ise-bay typhoon”.
the occasion of an Ise-bay Typhoon in 1959 (531 km2 with only 310 km2 caused
by storm surge). In Figure 3, the grey-scale presentation cannot show a detailed
spatial distribution of maximum flood depth but from the legend of the figure
one can recognize the range of maximum flood depths and the resolution of the
simulation.
In this scenario, overflow due to storm surge begins 18:00, and the tidal level
shows the maximum at 22:00, while the levee breaches along rivers due to
flooding happens after 01:00 of the next day. On the other hand, storm with
stronger wind than 20 m/s begins at 18:00 [5].
3 Recognition of risk in Phase 0

This type of disaster is characterized by wide-area flooding which continues for a
long time. It expands to around 500 km2 expanding to the three prefectures and
including many communities (cities and towns). Against ordinary disasters,
evacuation is completed within communities and hence communities and their
heads are responsible for evacuation by issuing commands and preparing shelters.
However, against catastrophic floods discussed in this paper, we have to face the
problems of boundaries of communities. For example, few shelters are available
in flooded communities. Evacuation must be carried out over the community
boundaries.
Number of days
required for 3~7
levee closure
and unwatering days
7 ~ 14 days
3 ~ 7 days 7~14days
0 ~ 3 days
3~7
days
Figure 4: Period required for drainage.
Once the levees are broken, a wide area is flooded and immediately the levees
should be closed and drainage efforts are required in Phase I. Closure of levee
breaches and drainage have several technical problems (district division and
arranging pumping vehicles), but we roughly calculated the necessary terms of
drainage for individual districts, as shown in Figure 4 [8]. Some areas may
remain flooded for a few weeks and during this period people cannot live their
daily lives. Considering that the tasks to be done in Phases I and II are many,
there is less possibility to support life in such a flooded area. Thus, preliminary
evacuation from the area expected to be exposed to severe flooding is strongly
recommended and it should be achieved before the disaster happens. That is
“preliminary evacuation in a wide area” (over communities). If a disaster
happens, people staying in refuges inside the seriously flooded area must move
to the shelters in the dry areas because they cannot continue their daily lives
without lifeline services there (secondary evacuation), and they need special
transportation in the flooded areas.
The numbers of evacuees of respective cities or towns (or wards of Nagoya
city) who have no shelters within their communities were surveyed and the
results are shown in Figure 5 [5–7], where several neighbouring communities are
grouped as one block. The most important emergency support function of
Phase 0 is preliminary evacuation to a safe area (where flooding is not predicted)
and the evacuation destination must be different communities and such
evacuation necessitates a long journey.
balanced
Block 1
Block 2 balanced
Block 3
Block 4
5700
78400
Block 5
Block 6
25400 Block 7 Block 8
31700 50000
balanced
Figure 5: Numbers of wide-area evacuees.
Recent development of weather forecasting techniques can tell us with high

probability the attack of a super typhoon with an extremely strong magnitude 36
hrs before its arrival. Forecasting will be improved during the time that the
typhoon approaches, but the difficulties increase in the long-journey of
evacuation of many people within a limited time [6, 7].
Phase 0 is divided into several stages as follows and related to the “storm
surge warning level”, which will be issued from the meteorological service as
shown in Figure 6 [5]:
Stage 0: 36–24 hrs before landing;
Stage 1: 24–12 hrs before landing (storm surge warning leve1 1);
Stage 4: 6–0 hr before landing (storm surge warning leve1 4).
Usually, each warning level corresponds to each action for evacuation guidance
as follows:
Level 1: Recommendation of voluntary evacuation;
Level 2: Evacuation completed for handicapped persons;
Level 3: Issue of evacuation advisory by community head;
Level 4: Issue of evacuation order by community head.
However, in this case of a catastrophic typhoon, since evacuation requires long
travelling distances and time, the above guideline is not available or it may be
too late.
p g
名古屋
Nagoya
■18:00時点■
18:00
北緯33.4度,東経135.9度
Stage
■ステージ4 ４（(-6:00～0:00)
1 2 :0 0 -1 8 :0 0 ）
避難指示を発令する。間に合わない地区や時
Inundation
伊勢湾岸にて高潮越流
による浸水開始 Evacuation
間によっては緊急避難を command
指示する。
Stage 3 (-3:00～-6:00)
■ステージ３（ 9 :0 0 -1 2 :0 0 ）
Wide Range
避難勧告を発令する。Evacuation
広域避難を行う。
12:00
■12:00時点■
北緯31.0度,東経136.0度
Stage
■ステージ2
２（(-12:00～-9:00)
6 :0 0 -9 :0 0 ）
Evacuation準備・
要援護者の避難を of HC 開始・完了する。
09:00
■09:00時点■
北緯29.9度,東経136.4度 Stage
■ステージ1
１（(-24:00～-12:00)
1 日前1 8 :0 0 -当日の6 :0 0 ）
自主避難を呼びかける。精度の高い台風進路
Voluntary
予測、高潮予測が発表さ evacuation
れる。
Stage 0 (-36:00～-24:00)
■ステージ０（ 1 日半前～1 日前の1 8 :0 0 ）
情報共有本部を設立し、関係機関で情報を共
Set Pre-JFO
有する。
Figure 6: Stages in Phase 0 for super Ise-bay typhoon.
4 Preliminary evacuation in wide area

As mentioned in the preceding chapter, the most important issue in Phase 0 is
how to achieve preliminary evacuation in wide area with long travelling
distances and time. By considering the numbers of evacuees of various districts
shown in Figure 5, the evacuation direction of each district is indicated [5].
Though evacuees may select their own shelters (direction and route in
evacuation) in “voluntary evacuation”, somehow controlled evacuation becomes
necessary after the evacuation advisory is issued, otherwise many evacuees
cannot be saved during the limited time and with limited shelters. For the time
being, Figure 5 shows organized evacuation in one of the possible plans. If such
large scale evacuation is realized, it requires agreements between communities of
origin and the destination of the evacuation.
Such evacuations with large numbers of evacuees and long distance within
the limited time require some kind of mass transportation. In the case of ordinary
evacuation, the means of transportation should be limited to pedestrian (walking)
in Japan, but in this case the evacuation distance is too far. Automobiles, buses
(arranged and hired by communities) and trains are taken into account. In order
to realize the plan, respective agreements between transportation companies and
communities are necessary. We investigated the time required for evacuation
completion for each community (from the center of the origin community to the
center of the destination communities) for several sets of combinations of cars,
buses and trains. The capacity and travelling time on the main routes such as
highways and national roads to connect origin and destination communities are
investigated for transportation by cars and buses. The statistics of railroad
companies are taken into account in the calculations for transportation by train.
Three cases of combinations are tested:
Case 1: cars (70%), buses (10%), trains (20%);
Case 2: cars (40%), buses (40%), trains (20%);
Case 3: cars (10%), buses (40%), trains (50%).
The results are summarized in Table 1, where the required time for completion of
evacuation is balanced with the number of evacuees from the origin block and
capacity of destination community. Table 1 suggests that there is an unbalance
between numbers of evacuees and capacities for respective communities (as a
total the capacity is less than the number of evacuees) and that the advantages in
the required time for evacuation change depend on the cases (combinations of
transportation means are different from one another).
Though no detailed names of communities are indicated in Table 1, Nagoya
city lacks the capacity of shelters and most citizens and the ward governors
consider that taller buildings may become refuges. However, once the area is
flooded, it might be quite difficult to support daily life there. People should know
that their daily lives depend on various lifelines, which are at risk of damage and
require a long time for their repairs. On the other hand, in the Nagoya city area,
if evacuees change their evacuation means from cars to trains, the required time
for evacuation can be efficiently reduced. Bus transportation is most efficient for
some other communities. Evacuation by individual cars requires a long time
though many inhabitants would use their own cars because buses and trains are
not a convenience in their daily lives.
Table 1: Required time of preliminary evacuation: case study of

transportation means, combination among car, bus and train.
Block O-community No.of evacuee Capacity of shelter Time for Evacuation (hrs)
(D-Community) Case 1 Case 2 Case 3
2 A 79,400 70,900 40 24 41
B 50 30 35
4a C 53,300 59,300 5 3 3
D 27 16 7
4b E 57,000 51,700 28 17 42
F 6 4 2
4c G 160,800 58,400 13 8 10
H 41 25 8
5 I 22 13 5
8 J 50 30 9
7a K 15 9 11
L 3 2 1.5
6 M 26,100 37,100 25 15 11
N 7 5 2
7c O 1.3 1.2 1.1
It is necessary to educate citizens to use mass transportation under such

emergency conditions. In addition, it may be possible to control main roads more
efficiently for one-way directed evacuation (contra-flow). However, this area is
on the centre of east-west main route for all kinds of transportation and difficult
to control without strong rules particularly for early preliminary evacuation.
Furthermore, we can recognize the differences of difficulties in preliminary
evacuation among communities, and depending on such conditions respective
strategies should be investigated.
5 Headquarters for risk management for super typhoon

Once a disaster happens, headquarters for disaster control with its field operation
office is set up and they collect information, make a repair and restoration
program and manage various emergency support functions. In other words,
disaster mitigation actions after Phase I are controlled and managed by the
headquarters.
In the case of catastrophic disaster due to a super typhoon, we can have 36 hrs
lead time for preparation against estimated serious flooding based on the recent
advancement of meteorological forecasting of typhoons. The most effective
action to reduce disaster is a preliminary evacuation of a few hundred thousand
inhabitants to dry area far from the community of origin. As discussed in the
preceding chapters, without any control and support, the necessary evacuation
cannot be achieved [6, 7]. Furthermore, in order to make a wide
evacuation possible, the agreements with destination communities and perhaps
bus-companies should be set up by the communities of origin for evacuation.
Even if such agreements exist, some trouble caused by multiple bookings will
happen because of the wide scale disaster by a super typhoon. On the other hand,
some agreements cannot be realized because people are afraid such troubles may
be realized.
In this study, we propose to organize headquarters for risk management in
Phase 0 for information sharing, in particular among communities, disaster
mitigation organizations and other stakeholders (see Figure 7). Through
information sharing, strategies should be decided and improved over time [6, 7].
Fundamental time lines of functions to be carried out by respective organizations
must be preliminarily planned, but they should be adjusted with some
modifications by considering the imbalance of resources among stakeholders on
the real case. And the headquarters will become a centre of information sharing
and will recognize modifications of time lines of respective organizations to
advise them of a possible adjustment of resources. Actually, the headquarters
organized after a disaster happens play a role of such adjustment among
organizations and stakeholders related to the imbalance of resources for disaster
mitigation (repair, restoration and supporting victims). What this study would
demonstrate is as follows: Such headquarters in Phase 0 have not yet been
proposed in Japan, but as mentioned above, there are many issues to be settled
before a disaster happens if risk management is considered for a super typhoon.
Local
Prefecture town
Other inhabitants Municipality
Organizations
Local
Informations: Inhabitants
・weather & river information
・Institution and action of organizations
・Information for evacuation:
Numbers of evacuees
Refuge condition, evacuation route
・Information on Handicapped persons Pre Joint Field Office
supporting system
・Traffic information:.
Traffic control, traffic jam, etc.
・Number of evacuee, staying peoples, etc.
Set‐up of Pre FEO ：Flow of staffs
：Information Flow ：Staff of FEO
Negotiation among
Tops of organizations
：Information System ：TNT member
Belonging to TNT
Figure 7: Image of joint field office (headquarters) in Phase 0.
Furthermore, such headquarters will smoothly continue to operate in disaster

control after a disaster happens.
Headquarters in Phase 0 might be organized step by step. As soon as the
weather forecast proclaims that a possible major typhoon may attack the target
area, the river manager and meteorological service will set up a centre for
information sharing (Stage 0 in Chapter 3, 36–24 hrs before the arrival of the
typhoon). The 36 hrs before typhoon arrival must be a trigger to organize such a
preliminary form of headquarters. From these headquarters, various information
will be distributed to the members of the final headquarters. In stage 1 (24–
12 hrs before arrival), communities that are expected to be flooded will produce
preliminary evacuation plans with certificates of the agreements to support them
and headquarters will check necessary and/or possible arrangements. The
preliminary evacuation plan has to be performed immediately in this stage. In
stages 2 and 3 (12–6 hrs before arrival), the progress of preliminary evacuation
will be checked and necessary support will be arranged. In stage 4 (6–0 hrs), the
remaining functions should be checked and dangers of evacuation during
the approach of the typhoon should be assessed to prepare for a change of
strategy (wide-area evacuation to emergency evacuation to the nearest refuges).
In the tasks involved in the above, information sharing and possible necessary
arrangements are required and headquarters should show the initiative.
Furthermore, headquarters organized before a disaster happens can be
smoothly followed by the headquarters after a disaster happens (Phase I).
6 Continued functions in Phase I

After the super typhoon has landed, preliminary evacuation in a wide area
becomes impossible because of violent storms (wind and rainfall, and flooding
due to storm surge). The strategy should be changed to action to save lives. The
remaining inhabitants in the areas where serious flooding is expected should
evacuate to the nearest temporary refuges, such as high buildings. Communities
should prepare some preliminary agreements with building owners for
emergency refuges. However, if one understands that a super typhoon brings
long term flooding, the efforts to support daily lives there must be difficult
because of the failure of lifelines. Secondary evacuation from flooded areas
requires several techniques, and thus, preliminary evacuation is recommended as
a total system. However, actually perfect preliminary evacuation is extremely
difficult (some inhabitants will not evacuate in spite of evacuation advice and
remain where they are), and an appropriate separation between emergency
evacuation and preliminary evacuation as an actual action plan is a sensible issue.
In the case of serious flooding in a land below sea level, which is caused not
only by a typhoon but also is caused by a tsunami after an earthquake, necessary
emergency support functions in Phase I are clearly distinguished from those
against disasters without flooding. First, closure of levee breaches and then
drainage are the most important tasks. Without these closures and drainage,
rescue activities cannot be made successfully although helicopters and boats will
help.
7 Concluding remarks
The area facing the Ise-bay is a low land more than 300 km2 below sea level and
exposed to a risk of flooding by storm surge and floods due to super typhoons.
We postulated the “super Ise-bay typhoon” as a possible maximum one, and
discussed risk management and emergency response in this paper. From the view
point of emergency management against flooding of wide areas with failures of
lifelines, preliminary evacuation in a wide area was investigated as the key in a
risk management action plan. In this area, a few hundred thousand citizens are
considered as evacuees who must travel long distances within a day. Without a
plan and means for controlled evacuation, they cannot succeed. In this paper,
some model of the combination of transportation means was investigated by
using simulation. The numbers of evacuees and capacity of shelters are not
balanced, and depending on conditions of the respective communities the
appropriate combination of transportation means for evacuation are different
from one another. It is one of the difficulties in this problem but conversely this
point may give us a key to find an appropriate action plan. Furthermore, we have
proposed headquarters for information sharing and arrangements of resources
among different stakeholders, and it is expected to smoothly continue with
headquarters for disaster control after the disaster happens.
References
[1] Chubu Association of Regional Design, 50 Years after Ise-bay Typhoon,
(supervised by T. Tsujimoto), 109p., 2009 (in Japanese).
[2] Japan Water Forum, Typhoon Isewan (Vera) and Its Lessons, 60p., 2005.
[3] White House, The Federal response to Hurricane Katrina: Lessons
Learned, 228p., US White House, 2006.
[4] Tsujimoto, T., Field survey on repair and restoration process and
improvement of emergency response in US after Hurricane Katrina and risk
and emergency management of Ise-bay area below the sea level against
storm surge and flood, Jour. Hydroscience & Hydraulic Eng., JSCE,
Vol. 54, pp. 889–894, 2010 (in Japanese).
[5] Tokai Nederland regional Authority against Storm Surge and Flood, Risk
Management Action Plan, 2nd Ver., 177p., 2009 (in Japanese).
[6] Tsujimoto, T., T. Kohno and S. Tanaka, Action plan for risk management
against large scale inundation due to “super Ise-bay typhoon” – Tokai
Nederland action plan against extreme storm surge and flood, Abstract, 5th
International Conf. on Flood Management, Tokyo, 2011.
[7] Kobayashi, K., On the Authority against Storm Surge and Flood in Tokai
Nederland –TNT Risk Management Action Plan, Lecture Note, 49th Summer
Seminar Series on Hydraul. Eng., JSCE, A-1, 2013 (in Japanese).
[8] Chubu Regional Bureau, Ministry of Land, Infrastructure, Transport and
Tourism, Unwatering Plan in the Nobi Plain, 132p., 2013 (in Japanese).
Multi-robot system for disaster area

exploration
F. Burian2, L. Zalud1, P. Kocmanova2, T. Jilek2 & L. Kopecny1
1
LTR s.r.o., Czech Republic
2
CEITEC, Brno University of Technology, Czech Republic
Abstract
CASSANDRA robotic system developed at LTR s.r.o. company and Brno
University of Technology is described. The system contains an operator’s station
controlled with one operator and a couple of robots – small and big ground robots,
flying robots (quadrocopters), and mapping robot. The robots are primarily
controlled by the operator with an advanced user interface with visual telepresence
and augmented reality. Nevertheless, the robots include the possibility of semi-
autonomous operation based on self-localisation. The user interface consists of a
computer, joypad, head-mounted display with inertial head-tracker,
communication device, and Cassandra software developed by our team in
Microsoft .NET. Orpheus class robots are described in the text. The robots are
made to be reliable and to be able to work in extreme conditions, they are tested
by a series of MIL-STD military tests for environmental parameters, EMC,
vibrations and shocks, contamination/decontamination, etc. Orpheus-X3 is a
general US&R robot with enhanced victim search capabilities, Orpheus-HOPE is
made for water contamination measurements, Orpheus-AC2 is a ruggedized
version for environmental parameter measurement. Two flying drones developed
completely by our team are described, as well as EnvMap mapping robot for real-
time construction of spatial digital maps with texture mapping. All the robots can
be controlled with the help of visual telepresence and augmented reality – that
makes robot control much more intuitive, and lets the rescuer concentrate on the
mission itself. The control station may be used as a self-containing wearable
system. The fusion system with multispectral measurement containing tricolor
cameras, thermal imagers and TOF camera is described.
Keywords: robot, user interface, telepresence, augmented reality, data fusion.
doi:10.2495/FRIAR140221
1 Introduction
The reconnaissance of dangerous areas is one of the most challenging tasks for
today’s robotics. According to many indications, e.g. from the Robocup Rescue
League community where the DCI team is involved [1, 3], it seems that nowadays
the development of practical and usable reconnaissance robots is aimed at the
following tasks:
• A larger number of robots controlled by one operator, in such cases as when
the operator must concentrate on crucial tasks, such as victim identification,
while the robots perform basic tasks, like mapping, autonomously.
• Easy and intuitive human-to-robot interface should be optimized, since the real
operators will be rescuers rather than robotic specialists.
• For many kinds of reconnaissance missions it would be highly beneficial if the
user interface would somehow emphasize alive people – since they are often
the main objective (earthquake or floods victims, injured soldiers, criminals or
terrorists).
The remote robotic reconnaissance of dangerous areas is a very complex and
interdisciplinary task [7], and only well-tuned robotic systems, with good
software, hardware, communication and sensory subsystem, may succeed [17].
Mobility and the ability to work reliably in hard conditions are very important. It
also induces that mechanical construction and the hardware of the robots play a
very important role in this complex task [13].
The authors propose a possible solution of the abovementioned problems
through an advanced user interface program called CASSANDRA and show its
application on several reconnaissance robots developed by our team.
Although the technical features of individual robots are supposed to differ, the
robots can be divided into certain “classes” of robots that are capable of being
controlled with the control system. The classes are listed below with an emphasis
on their mapping and self-localization abilities.
• Bigger and more complex robots with sufficient mapping and self-localization
capabilities (e.g. Orpheus).
• Small robots with limited mapping and self-localization capabilities (e.g.
Perseus).
• Rotorcraft Unmanned Aerial Vehicles (UAVs) with self-localization only (e.g.
Uranus).
• Mapping robots with exceptional mapping and self-localization capabilities
(EnvMap).
At present, the reconnaissance robots and the operator’s telepresence control
system, the CASSANDRA, are completed; thus, each robot can be effectively
controlled by the system. Multispectral data-fusion for colour and thermal image
mixing with help of TOF camera, is finished as well. The current task consists in
enabling the automatic mapping and self-localization of the robots, both outdoors
and indoors, and implementation of enhanced reality mixing real telepresence data
with the data from the multispectral maps.
2 Orpheus robots
Orpheus robots have been developed at Department of Control and
Instrumentation (DCI) and LTR s.r.o. (spin-off Brno University of Technology)
since 2003. The first version was called simply Orpheus, and our team was quite
successful in Robocup Rescue 2003 world competition in Padova, Italy – we won
the competition (see [1]). In 2003–2006 we improved/rebuilt the robot to the
version Orpheus-X2 (see [2]). In 2006 we were asked to make a military version
of the robot. The prototype was finished in 2007 and named Orpheus-AC (Army
and Chemical). In 2009 we started development of second generation, based
on Orpheus-A2 platform. We decided to make two basic modifications – Orpheus-
AC2 for chemical and nuclear contamination measurements and
Orpheus-Explorer for more general reconnaissance missions and victim search.
Figure 1: Orpheus robots (from left): Orpheus-AC prototype, Orpheus-AC,

Orpheus-AC in snow, Orpheus-Explorer.
2.1 Orpheus-AC
Orpheus-AC (see Fig. 2) is a rugged robotic system made to reconnaissance highly

dangerous areas with chemical and nuclear risks. The main mission objective is
chemical and nuclear contamination measurement. The robot is equipped with
beta and gamma-radiation sensors as well as LCD 3.2 chemical contamination
probe. The robot is made for military purposes, so it fulfils military standards, it
successfully passed 17 MIL-STD STANAG tests, e.g. environmental, vibrations,
EMC, etc. The robot is equipped with two cameras – one zoom colour camera with
illumination (wide and narrow light beam) with both manual and automatic
parameter settings, and one rigid “rear” wide-angle camera – colour, highly
sensitive. The robot has one degree of freedom manipulator with sensors, while
other sensors are rigidly connected to the robot body. The robot base is rigid, has
low profile with high clearance because of big wheels. The robot may be operated
wirelessly or by wire. The robot is made to work in hard terrain; it is able to go
across obstacles up to 20cms high; it is able to work well during the night or in
bad visibility conditions (sensitive cameras, configurable illumination).
The robot itself is made to be easy to de-contaminate, the whole robot is
waterproof, painted by resistive paintings and the whole construction is made to
repel or at least not to keep liquids. Only several parts are marked as non-
decontaminable and have to be replaced – tires, antennas and two cables.
Figure 2: Orpheus-AC2 (from left) – Robot accessing the ramp to armoured

vehicle, Orpheus-AC2, User interface screen inside the armoured
vehicle.
decontamination process, it is newly equipped with two degrees-of-freedom

sensory manipulator with beta probe, chemical sorbent tube, distance
measurement and camera. Significant part of electronics is completely new,
internal communication system is newly based on both CAN and Ethernet, new
wireless communication modules working in licensed frequency spectrum are
used. The robot is a part of CBRNE armoured vehicle and aims to primary
contamination measurement in the areas with high contamination risks.
2.2 Orpheus-HOPE
Orpheus-HOPE (see Fig. 3) is a robot built on modified Orpheus-A platform and

its primary mission is water contamination measurement. It is a product of
research and development of our university together with Laboratory of
Metalomics and nano-technology and Laboratory of Microsensors and
Nanotechnology of CEITEC project.
Figure 3: (From left) Orpheus-HOPE controlled by wearable operators’

station, user interface screenshot during water heavy metal
measurement, sensory head detail.
Its main difference to the other Orpheus robots is the sensory head on 1 DOF
motorised manipulator with heavy-metal analysis probe and water dive sensor –
both the sensor and probe were newly developed by our teams. The robot, in its
current status is only a proof-of-concept, we are currently working on practically
usable device with peristaltic pump-based system of water sampling.
2.3 Orpheus-X3
The Orpheus-X3 (see Fig. 4) is an experimental reconnaissance robot based on the
Orpheus-AC2 model. It offers the same drive configuration as its predecessor,
namely the four extremely precise AC motors with harmonic gears directly
mechanically coupled to the wheels; this configuration makes the robot very
effective in hard terrain and enables it to achieve the maximum speed of 15 km/h.
The main difference consists in the chassis, which is not designed as completely
waterproof but consists of a series of aluminium plates mounted on a steel frame
of welded L-profiles. This modular structural concept makes the robot markedly
more versatile, which is a very important aspect in a robot made primarily for
research activities. Furthermore, the device is equipped with a 3DOF manipulator
for the sensory head. The manipulator, again, comprises very powerful AC motors
combined with extremely precise, low backlash harmonic drive gearboxes by the
Spinea Company. The presence of such precise gearboxes can be substantiated by
several reasons, mainly by the fact that the robot will be used not only for
telepresence but also for mobile mapping and SLAM [9, 10]. As currently planned,
the robot’s only proximity sensor will be the TOF camera.
1 3 1
2 2
Figure 4: (From left) Orpheus-X3, multispectral sensory head. 1 – the tricolor

CCD cameras, 2 – the thermal imagers, 3 – the TOF camera.
2.3.1 Multispectral data-fusion
The aim of the data fusion is to facilitate remote reconnaissance of previously
unknown areas under a wide variety of visibility conditions, including fog, smoke,
complete darkness, or high illumination dynamics with point light sources. It also
visually emphasizes alive people (usually victims).
It represents a technique for the alignment of visual spectrum data and thermal
imager data, utilizing the information provided by a TOF camera. The TOF camera
measures the distance of an object, while corresponding pixels are found on the
applied color camera and thermal imager. Each of the sensors has to be calibrated
for geometrical errors; mutual position and orientation are found and used to
secure the corresponding calibrations [6].
The sensory head is shown in Fig. 4 right. It contains:
• Two tricolor CCD cameras (see 1 in Fig. 4). The Imaging Source DFK23G445
with 1280x960 pixels resolution, max refresh rate 30Hz, and GiGe Ethernet
protocol. Computar 5mm 1:1.4 lens is used.
• Two thermal Imagers (see 2 in Fig. 4). MicroEpsilon TIM 450 with a wide
lens, 382x288 pixels resolution, temperature resolution of 0.08K.
• One TOF camera (see 3 in Fig. 4). A Mesa Imaging SR4000 with the range of
10m, 176x144 pixels resolution. The field of view is 56°(h) x 69°(v).
The scheme of the presented system is indicated in Fig. 5, right.
THERMO L EMBEDDED PC
THERMO R EMBEDDED PC
TOF
ETHERNET
CCD L SWITCH
CCD R
OPERATOR STATION PC
ETHERNET
USB
Figure 5: Calibration pattern for TOF camera, CCD camera and thermal
imager (left), scheme of multispectral sensory head connections [4]
(right).
The technique was already studied by our team in the past (see [14]), but as the
sensory prices decreased rapidly and TOF cameras further developed, the method
may be improved to reach a significantly more advanced stage.
Image transformations are applied for data fusion. The range measurements of
the TOF camera can be displayed into images of CCD cameras and thermal imagers
using spatial coordinates. The procedure is outlined in the diagram on Fig. 6. The
input data include the range measurement, the image coordinates of all sensors, and
the results of the previous calibration.
The spatial coordinates X, Y, Z are computed from Eqs (1) and (2), where d is
the measured distance, xc, yc are the calibrated TOF image coordinates, and f is the
focal length of the TOF camera. The homogeneous transformation is determined
by Eq. (4), where R[3×3] is the rotational matrix, t[3×1] is the translation vector, and
X', Y', Z' are the spatial coordinates of the second sensor. The image coordinates of
the TOF camera in the next frame xc',yc' are computed according to perspective
projection (see Eq. (4)), where f' is the focal length of the second sensor.
   


Z  d cos arctan
yc   cos arctan xc   
  2    f 
  f 2  xc     
Zxc Zy
X  Y  c 
f f
Range measurement and image coordinates of TOF camera
Spatial coordinates
Homogeneous transformation
Perspective projection
Correction of principal point
Displaying overlapping images
Figure 6: Image transformation scheme.

Measured Intrinsic and
TOF, CCD FUSION
Range
spatial 3D extrinsic
image + +
points [X, Y, Z] parameters
Range image projected to CCD image
TOF, TERMO FUSION
Measured Intrinsic and
Range
spatial 3D extrinsic
image + +
points [X, Y, Z] parameters
Range image projected to thermal image
CCD, TERMO FUSION
ID point in CCD ID point in
image thermal image
Figure 7: Scheme of data fusion: up – TOF and CCD data fusion; centre –
TOF and thermal data fusion; down – CCD and thermal data fusion.
 X ' X 
Y '   
     R t   Y  
 
 Z '   0 1  Z 
   
1 1
f 'X' f 'Y '

 xc '   yc '  
Z' Z'
According to the identical (ID) points of the TOF camera transformed into the
frames of the CCD camera and the thermal imager, the thermal image can be
displayed into the CCD image and vice versa.
Figure 8: (Left) complete fusion for one eye; (right) top right: the range
image; bottom: fusion of the range and thermal images by the
described algorithm.
This is performed for two stereo pairs of cameras, and thus the resulting image
may be presented to a head-mounted display with a stereovision support [16]; the
operator therefore receives a very good spatial representation of the environment
under any visibility conditions.
Figure 9: Data-fusion evidence grid with colour only (left) with thermal
imaging (right).
It has to be pointed out that the sensors on the sensory head are not used only
for this technique; simultaneously, we are also developing a SLAM technique and
similar texture-mapping algorithms [11] with robot evidence grids and octree [15].
Both of these maps contain color information and thermal information [12], so e.g.
alive humans can be easily emphasized in the image – see Fig. 8. The octree map
has the advantage of great loseless data-compression (up to 1:512 for the scene on
Fig. 9 and resolution 1.28 cm), while the evidence grids are easy-to-modify.
Currently we are able to combine both of them in one image.
3 EnvMap mapping robot

The present status of the robot being developed under the name EnvMap is shown
in Fig. 10. The final design of the robot is not expected to be similar to the
prototype, because the currently used drive configuration is unsuitable for hard
terrain operation.
Figure 10: EnvMap robot indoors (left) and outdoors (right).
Precise digital autonomous mapping of a previously unknown environment [5]

forms a crucial part of the entire robotic reconnaissance system. A typical activity
requiring a faithful map of the environment is victim rescue planning, where the
rescuers need to recognize the exact position of the victim, know the dimensions
of the passages, and plan the rescuer passage through the area. Maps built by the
robot from the surrounding of our building are on Fig. 11.
Figure 11: Spatial robot evidence grid (left) and height map (right) scanned by
EnvMap robot.
4 Other robots
A couple of other robots were developed by our team as a part of CASSANDRA
system. Uranus is our own multicopter [19] system. It currently contains two
quadrocopters – Uranus-ALU with 350 g payload capacity, and Uranus-CARB
with approx. 1500 g payload capacity (see Fig. 12).
Scorpio is an indoor robot based on Dr. Robot drive system similar to iRobot
Packbot. The robot is intended for indoor operation, it is able to climb up-stairs.
Our team only developed the electronics and camera manipulator. Perseus is one
of our small robots capable of operation in hard terrain (see Fig. 12).
Figure 12: From left – Uranus-ALU quadrocopter, Uranus-CARB

quadrocopter prototype, Scorpio indoor robot, Perseus mini-robot.
5 CASSANDRA software
All the mentioned robots may be controlled by CASSANDRA software,
developed by our team. It is basically a universal user-interface program
developed in Microsoft .NET 4.5, WPF. It has many displaying capabilities (see
Fig. 13). The most important central part is filled with main camera image, while
the corners can be covered by configurable virtual head-up displays containing
video from other robot cameras or video from other active robots, as well as, other
data from robot sensors or depicting system status. The system can work with
variety of head mounted displays equipped with head movement sensors.
Figure 13: CASSANDRA software screenshot with description.
One of the main advantages of the whole CASSANDRA system is, that since
all of the parts (i.e. all the robots, software, communication protocols) are
developed by our team, it is possible to control all of the robots by one operator’s
station equipped with CASSANDRA software (see Fig. 14). So the operator has
the possibility easily switch among the robots and select the most appropriate one
for the task.
Figure 14: CASSANDRA system scheme.
6 Conclusion
The presented CASSANDRA system represents work-in-progress, rather than
completely finished system. The telepresence part of the system is considered
finished, currently the team works on integration of semi-autonomous and
autonomous functions, like self-localisation and autonomous real-time map
building. Several parts of the system are currently practically usable, e.g. Orpheus-
AC2 robot, that is in active military service in the Czech Army.
Acknowledgements
This work was supported by CEITEC – the Central European Institute of
Technology (CZ.1.05/1.1.00/02.0068) utilizing the European Regional
Development Fund.
This work was also supported by VG 2012 2015 096 grant named Cooperative
Robotic Exploration of Dangerous Areas by the Ministry of Interior, Czech
Republic, program BV II/2-VS.
References
[1] Zalud, L., (2004). Rescue Robot League – 1st Place Award Winner. In:
RoboCup 2003: Robot Soccer World Cup VII, Springer, Germany, ISBN 3-
540-22443-2.
[2] Zalud, L., (2001). Universal Autonomous and Telepresence Mobile Robot
Navigation. In: 32nd International Symposium on Robotics – ISR 2001,
pp. 1010-1015, Seoul, Korea.
[3] L. Zalud, “Integration of 3D Proximity Scanner to Orpheus Robotic
System”, in 16th IFAC World Congress. Prague, Czech Republic: 2005,
pp. 1209-1215.
[4] Kocmanova, P., Zalud, L., Spatial Calibration of TOF Camera, Thermal
Imager and CCD Camera. In Mendel 2013: 19th International Conference
on Soft Computing. Brno: Brno University of Technology, 2013, pp. 343-
348. ISBN 978-80-214-4755-4.
[5] Hartley, R., Zisserman, A., Multiple View Geometry in Computer Vision.
Cambridge, Cambridge University Press, 2003, ISBN 05-215-4051-8.
[6] Zhang, Z., Flexible camera calibration by viewing a plane from unknown
orientations, In Computer vision, Vol. 1, 1999, pp. 666-673.
[7] Lundberg, C., and Christensen, H. I. Assessment of man-portable robots for
law enforcement agencies. In PerMis (Gaithersburg, MD, Aug 2007), R.
Madhavan and E. Messina, Eds., ACM/IEEE.
[8] Leonard, J.J., Durrant-Whyte, H.F. (1991). “Simultaneous map building and
localization for an autonomous mobile robot”. Intelligent Robots and
Systems’ 91. ‘Intelligence for Mechanical Systems, Proceedings IROS’91.
IEEE/RSJ International Workshop on: 1442–1447. doi:10.1109
/IROS.1991.174711. Retrieved 2008-04-08.
[9] Karlsson, N., Di Bernardo, E., Ostrowski, J, Goncalves, L., Pirjanian, P.,
Munich, M. (2005). “The vSLAM Algorithm for Robust Localization and
Mapping”. Int. Conf. on Robotics and Automation (ICRA).
[10] Ju, X., Nebel, J.-C., Siebert, J. P., 3D Thermography Imaging
Standardization Technique for Inflammation Diagnosis. In Proceedings of
the SPIE, Vol. 5640, 2005, pp. 266-273.
[11] Prakash, S., Pei Yean Lee, Caelli, T., 3D Mapping of Surface Temperature
Using Thermal Stereo, In Control, Automation, Robotics and Vision, 2006,
pp. 1-4.
[12] Tournas, E.-Tsakiri, M., Distance Error Estimation for Range Imaging
Sensors. In: Proceedings of the ISPRS Commission V Mid-Term
Symposium “Close Range Image Measurement Techniques”,
Vol. XXXVIII, 2010, Part 5, pp. 581-585. Newcastle upon Tyne, United
Kingdom, 21.-24. 6. 2010.
[13] Wise, E., (1999). Applied Robotics, Prompt Publications, USA, ISBN: 0-
7906-1184-8.
[14] Henning Eberhardt, Vesa Klumpp, Uwe D. Hanebeck, Density Trees for
Efficient Nonlinear State Estimation, Proceedings of the 13th International
Conference on Information Fusion, Edinburgh, United Kingdom, July,
2010.
[15] Posted on (19 December 2011). “Understanding Requirements for High-
Quality 3D Video: A Test in Stereo Perception”. 3droundabout.com.
Retrieved 29 March 2012.
[16] Robin Ritz, Markus Hehn, Sergei Lupashin, and Raffaello D’Andrea,
“Quadrocopter Performance Benchmarking Using Optimal Control”,
IEEE/RSJ International Conference on Intelligent Robots and Systems,
pp. 5179-5186, 2011.
[17] “CBRN Defence Market Forecast 2014-2024”.
http://www.visiongain.com/Report/1206/CBRN-Defence-Market- Forecast
-2014-2024 London, U.K. 14 February 2014. Retrieved 24 March 2014.
[18] Zalud, L., Burian, F., Kopecny, L., Kocmanova, P. (2013). Remote Robotic
Exploration of Contaminated and Dangerous Areas, International
Conference on Military Technologies, pp 525-532, Brno, Czech Republic,
ISBN 978-80-7231-917-6.
Section 8
Adaptation to flood risk
Floating houses: an adaptation strategy for

flood preparedness in times of global change
P. Strangfeld & H. Stopp
Department of Building Physics,
Brandenburg Technical University, Germany
Abstract
The rising sea-level and the frequency of devastating floods have already
increased in a considerable way. At the same time, the population is
continuously rising, along with the demand for adequate housing and sufficient
space. In this context the so-called floating houses are a future-oriented solution
for settlements along coastlines and river districts or on little islands.
In highly industrialized countries which export products or the associated
licenses it is an opportunity for the development and construction of floating
houses. Up to now in most cases the floating objects are built on pontoons as
usual buildings on a fixed ground. The special boundary conditions caused by
water waves, water chemistry and climate components should be considered in
order to prevent damage. Concomitantly the floating objects bring chances for
mobility and use of alternative energies due to the water environment.
In Lusatia, a landscape in the eastern part of Germany southeast of Berlin, a
lot of former lignite open-cast mines were filled with water and the worldwide
largest artificial lake landscape was created among others by the assistance of the
International Building Exhibition “Fürst-Pückler-Land”. Different types of
floating houses have already been built. The department Building Physics of the
university BTU-CS has carried out a lot of investigations with regard to
materials, energy use and climate boundary conditions by means of
measurements and numerical simulations. Besides, water as a building ground
must be cost effective and exhibit an affordable floating architecture.
Keywords: floods, urban strategies, floating house, heat exchanger.
doi:10.2495/FRIAR140231
1 Present global situation

1.1 Population
Contrary to the German and European situation, the global population is growing
(fig. 1). Above all, the aspiration level increases unrestrained in our society also
with regard to the demand for living room. For instance in Germany 70 ha of
land per day are used additionally for building in spite of the decreasing
population in this country.
Figure 1: Prognosis of the increase in millions of the worldwide population.
1.2 Climate change
Independent of the knowledge of the reasons for climate change, the sea level is
rising and endangers the infrastructure of settlements in many regions (fig. 2).
Figure 2: Examples for the effect of sea level rising in the Netherlands and Asia.
2 Floating houses
2.1 General situation
Floating houses have a long history [1]. The technique and architecture of these
buildings depend on climate boundary conditions, culture and raw materials
which were available in various places. Nowadays one can find exquisite
examples of floating buildings all over the world (fig. 3).
Figure 3: Seoul, floating amusement park opened 2011, Hangang river at night.
2.2 Examples in Germany
The following figures represent examples of buildings in Germany floating on

lakes of former opencast mines.
Figure 4: Floating houses type “Ar-che” in the Lusatian landscape.
Figure 5: Floating research station on a former opencast gravel mining (left);

Floating church on a former lignite opencast mining (right).
3 Adaptation
3.1 Material and construction
3.1.1 Materials
Building materials used especially for pontoons are highly dependent on the
quality of the surrounding water in addition to economic reasons (e.g. the pH
value of mining lakes plays a major role in the corrosion of steel and concrete).
3.2 Construction
Floating houses are built on mobile ground. The buoy in fig. 9, installed on the
lake of a former opencast lignite mine, records the water waves data and other
sensors the effects to the construction of floating houses.
Figure 6: Installation of a buoy for recording the parameters of water waves

by GPS -technology.
Another considerable load for a construction is the frost load of the pontoons
and piles (fig. 7a). By means of so-called heat pipes alternative energy is used to
avoid or at least to reduce the frost action upon the piles. Fig. 7b displays such a
manufactured heat pipe for a pile and figure 8 shows its installation with the help
of a crane.
a b
Figure 7: a: Piles subjected to freezing conditions. b: Prefabricated heat pipe.
Figure 8: Installation of the heat pipe by means of crane work.
3.3 Self-sufficiency
3.3.1 Water supply, sewage and waste disposal

Autonomous building is an important prerequisite for a largely economic
implementation of floating houses. Drinking water supply and wastewater are
part of a field with major recent advances. Decisive impulses have been shaped
by space technology.
3.3.2 Energy supply

 Spiral heat exchanger
The virtually unlimited amount of surrounding water provides new opportunities
for the use of alternative energies. Figure 9 displays images during experiments
for a spiral heat exchanger and the thermal response method. Figure 10 depicts
results for the hygrothermal characteristics of the spiral heat exchanger by using
numerical simulation. Assembly of the heat exchanger into the pontoon of the
“Ar-che”-type floating house is shown in fig. 11. In this case the exchange of
heat energy is reduced.
Figure 9: Investigation of the performance parameters of a heat exchanger

with the thermal response method.
Figure 10: Calculated temperature distribution of a spiral heat exchanger

(left), velocity distribution of the exchanger caused by buoyancy
(right).
Figure 11: Assembly of the heat exchanger into the pontoon by crane.
Figure 12: Scheme of the arrangement of six heat exchangers, valves and
heat pump.
Figure 13: Water-temperature course at different measuring points.
 Compact heat exchanger

In contrast to the spiral heat exchanger, a compact shape requires less space for
the same heat power. By means of numerical simulations an optimal
arrangement is to be found with regard to heating in wintertime and cooling in
the summer.
Figure 14: Heat exchanger of the Fa. Frank GmbH in Germany.
Figure 15: Temperature distribution in a heat exchanger. Heating mode in

wintertime, surrounding water: 15°C, sole temperature: 5°C.
Figure 16: Temperature distribution in a heat exchanger. Heating mode in

summer, surrounding water: 15°C, sole temperature: 25°C.
Figure 17: Measuring results of the thermal response method. Left: spiral heat
exchanger, right: compact heat exchanger – temperature of
surrounding water: 9.8 °C.
 Heat storage
Solar energy can be stored in the pontoon’s space in connection with the classic
solar thermal energy. It is also possible to collect the solar energy directly
through a transparent, insulating cover in a floating storage box (fig. 18) [2]. In
the latter one the detection of temperature distribution is of interest (fig. 19).
Figure 18: Floating box with a cover of heat-insulating glass for experiments.
The temperature distribution is calculated with and without the convection

influence upon the sea water.
Figure 19: Figure 19: The influence of the asymmetric solar radiation on the
water surface (left) is eliminated by convection within the water
(right).
Figure 20: Investigation of heating and cooling of building envelopes by

means of water flowing through the structure caused by different
densities due to the asymmetric effect of solar direct/ diffuse
radiation and long-wave emission.
4 Outlook
Figure 21 depicts the potential of floating houses. Thanks to its mobility
principle, a floating object can be transported to another location after fulfilling
its purpose. For instance a floating stadium is used for another purpose after a
football championship. Nowadays amphibious buses already travel among
islands or between the canals of Amsterdam and the Schiphol airport (fig. 22).
Another possibility is buildings floating up in the case of floods if water
occurs. It could be an adaptation strategy for flood preparedness near river
districts in the future.
Figure 21: Design of a floating soccer stadium for the World Cup 2022 in
Qatar, Architectural Office Düsseldorf, Peter Knoebel.
Figure 22: Amphibious buses in Budapest and Amsterdam as rational means

of public transport, without the need for a ferry [3].
References
[1] Stopp, H.; Strangfeld, P.: Schwimmende Wohnbauten, Beuth Verlag,
Berlin-Zürich-Wien 2012.
[2] Harnath, M., Heating and water supply of floating houses in compliance
with energy issues, Master thesis, University of Appl. Sciences, Hochschule
Lausitz (FH), 2011.
[3] www.floatingdutchman.nl, www.toursales.com/Floating-Bus.
Design as a negotiation platform:

new deals and spatial adaptation in
flood-prone areas
F. Rossano1 & L. Hobeica2
1
Institute of Landscape Architecture,
Federal Institute of Technology – ETH Zurich, Switzerland
2
Institute for Interdisciplinary Research, University of Coimbra, Portugal
Abstract
In current measures taken in Europe to cope with growing flood risks, various
elements characterize the strategic and practical choices involving anticipation,
protection or mitigation. One crucial element in all flood-related projects is
space. In quantitative and qualitative aspects, most flood adaptation strategies
imply a morphological transformation of city and landscape, as well as the
redefinition of land use and status, which in its turn can lead to new deals among
territorial players. These multi-scale interplays can eventually put financial,
political and social status-quo under unknown pressure, and transform the role of
urban and landscape design, which gains in importance but also in complexity.
The nine contemporary flood-related projects reviewed reveal that the fluctuating
conditions and multiple interests in which they evolve require, in addition to
creative approaches, openness, perseverance and diplomatic skills. Landscape,
urban or architectural design becomes then an open and dynamic platform for
spatial renegotiation and adaptation, challenging design practices in flood-prone
areas as well as democratic structures.
Keywords: flood risk, urban and landscape design, adapted spatial design,
negotiation platform.
1 Introduction
Flood-related riverine projects involve two precious resources for urban
civilizations: water (the river) and space. In fact, they all imply physical and/or
cognitive redefinitions of space: local public space and civil works can become
doi:10.2495/FRIAR140241
regional defence infrastructures; land that seemed suitable for building can
become junk bond for investors if declared risk zone; purely agrarian areas
can turn into water storage. Along with these transformations, city and landscape
negotiate a new potential damage distribution, building up inter-linkages and
engaging into a reflexive redefinition of their respective roles. Besides, within
the city, riverine spaces generate simultaneously growing fears, waterfront
development ambitions and new functional combinations. These multi-scale
interplays, the economic and social pressures linked to them, and the diversity of
territorial players involved represent an extra layer of complexity in the remit
of spatial designers. Yet, some contemporary European flood-related projects do
recognize both space and water dynamics as crucial variables of flood adaptation
strategies. Our methodological approach was thus to review nine of these
projects, aiming to identify how spatial design has fulfilled its task of negotiation
platform. Case study was adopted as the research method, as it allows to gain a
comprehensive view of the targeted projects, thanks to its simultaneous attention
to “the complex relationships between context, product and process that govern
every design process” [1].
After introducing an overview of the multiple stakes involved in most flood
adaptation strategies (Section 2), the paper will pinpoint to how the notion of
‘river space’ has taken spatial design as a new dimension of flood management
(Section 3), implying for flood-prone territories a New Deal generated by design
(Section 4). We will then summarize some of the main roles played by spatial
design (acting as a dynamic negotiation platform) to shape these new deals in the
analysed cases (Section 5), before concluding with general implications and
prospects for future developments in flood-prone territories.
2 Flood proneness: one among several territorial constraints

Coping with riverine flood risk usually involves multiple conflicts, despite the
overall characteristic of the river or the territorial scope of the adaptation
alternatives. When embedded in the urban scale, a first dilemma can be accepting
the very existence of flood risk in this setting, a well-known cognitive conflict.
In fact, the lives of urban dwellers are increasingly disconnected from natural
variations, and people living in flood-prone zones, through a heuristic
mechanism, tend to perceive their homes as inherently safe places [2]. Despite
the concentration and value of assets exposed to floods in cities, this risk is
usually made invisible by the existence of structural flood defences, which
promote (or at least do not discourage) a less precautious attitude towards flood
proneness. This is even more blatant when urban regeneration or development is
at stake. Here, several demand conflicts are added to the perceptive one: shorter-
term urban needs (like housing or economic development) tend to exert –
justifiably, one could argue – high pressure on flood-prone areas, as the
experience of everyday problems by local populations is more direct than that of
floods, with their extraordinary (but potentially devastating) character [3].
In their turn, other sustainability issues (e.g. urban compactness and mobility)
transform traditional “bad places” (floodable, polluted etc.) into valid options for
urban development [4]. In recognition of such site strengths as location, scenery

or existing infrastructure, instead of simply banning redevelopment, some
authors advocate a more pragmatic approach (for example Barroca and
Hubert [5]), whereby existing site constraints and strengths are weighed against
each other, in a more horizontal decision process. In this condition, damages and
responsibilities (and also benefits) can be recognized in advance and shared
between all stakeholders, in an attempt to maximize gains and minimize regrets.
But this is far from a straightforward process; the reality on the ground is much
more complex, especially in cases where the best flood adaptation solutions have
a regional scope, exceeding usual administrative boundaries and competencies.
As space is the real arena where the conflict between rare hazardous events
and more tangible human interests is made visible, designing space has logically
to deal with all idiosyncrasies involved in a given river and the wider space
around it. Similar baseline conditions can lead to various conflicts and divergent
results. For the Scheldt Quays in Antwerp and the Isarplan in Munich, floodplain
function and urban life had to be combined into one single design. In Antwerp,
the competition brief highlighted (potential) conflicting issues within the design
task: raising the flood defence scheme (final height 2.25 metres), not obstructing
the city view of the river and eliminating the urban barrier effect of the wall. The
selected design proposal finally accommodated these demands by merging them
in a single urban “civil-civic structure” [6], which is altogether a levee and a
belvedere, and adapts to the site’s local circumstances. In Munich, the main
goals of the initiative seemed well established by the city and the State of
Bavaria, namely combining urban recreation with environmental restoration and
a functional hydrological system. However a controversy emerged after the
results of the design competition were made public, opposing partisans of an
outspoken urban space design and supporters of a nature-like project. The final
solution was a compromise that respects infrastructural constraints but suggests
natural freedom by creating artificial islands, pebble paths and curved shores (a
nature forged by the designer in order to meet the public’s aesthetic
expectations). In both cases of Antwerp and Munich, spatial design was
challenged by conflicting and combinatory expectations, becoming a
fundamental dimension of the pursued flood management strategy.
3 Spatial design: a new dimension to flood management

3.1 From river to river space
Many of the recent flood adaptation projects associate to traditional interventions

(such as river bed dredging or levees enforcement) horizontal solutions such as
river widening, floodplain restoration or the creation of controlled flood areas.
The Netherlands, renowned for its dikes and sea walls, is now implementing the
2,300-million programme Room for the River that consists, among other
interventions, in widening rivers in both cities (e.g. the Waal River in Nijmegen)
and countryside (e.g. the Merwede River in the Noordwaard area) [7]. The
country today officially promotes a “multi-layer safety” approach [8], based on
three dimensions: flood prevention, sustainable spatial planning and disaster

control, which include both mitigation and adaptation measures (such as the
so-called ‘calamiteitenpolders’, agricultural polders that allow temporary
flooding to avoid greater damage in urban areas). Comparable expansion and
diversion strategies are currently being developed along the Isère River (France)
and the upper Rhône Valley (Switzerland).
Although the “space for the river” approach could wrongly be presented as a
new concept [9], the increasing interest for horizontal answers to riverine flood
risk, together with the growing acceptance of occasional flooding as an
inevitable hazard to be dealt with rather than eliminated, logically reinforce the
spatial aspect into flood management. Subsequently, contemporary official
documents elaborated to communicate on flood adaptation projects often refer to
the river not only as a stream but more frequently as a space: Ruimte voor de
Rivier (the Netherlands), Isarraum (Bavaria) or Espace Rhône (Swiss Valais), all
suggest the necessity to consider not only the stream and its edges, but a wider
area that includes all surfaces that can be potentially affected by the river’s
fluctuations. Furthermore, post-World War II urban densification and sprawl
have changed radically the context of flood management. Water retention and
flood diversion areas cannot be solely implemented within the natural
environment, generally too reduced or fragmented to assume this function; in
fact, they compete today with farmland, infrastructure, recreation space,
ecological restoration or urbanization. Thus, the expansion of the river space
initially meant to accommodate higher discharges and prevent flooding,
combined with a strong land shortage within urban areas, calls for an integrative
design to blend all parameters into an altogether attractive, ecologically valuable,
resourceful and safe living environment. Long seen as infrastructures or threats,
rivers and their fluctuations are now by necessity being reintegrated into the
public physical and cultural realms, raising new questions in regards to the space
needed, its perimeter, status, accessibility and still-to-be-defined aesthetics.
3.2 A New Deal for flood-prone territories
Traditional European planning regulations long defined flood zones on the basis
of previous events and/or flood models, to then apply limitations in land use and
construction. Although still essential to most planning practices, this passive
method shows today its limits, especially in densely built flood-prone
environments. Numerous constructed obstacles have modified the contours and
behaviour of flooding. Contemporary flood zones are thus no longer determined
only by natural elements, but increasingly by the effects of man-made civil
works, earthworks, buildings or planted vegetation, as a result of past political
decisions – if not the sum of faits accomplis. Furthermore, when potential flood
space covers all or large parts of the living territory, the question cannot be
solved in simple terms of building limitations or natural floodplain restorations,
but also involves a crucial negotiating aspect to define what needs to be
floodable in order to accommodate higher discharges and to protect the most
valuable assets. Permanent river widening, as applied to the upper Rhône River,
or the creation of temporary flood spaces, such as the Dutch calamiteitenpolders
and the French champs d’inondation contrôlée (controlled flood fields) currently
implemented along the Isère River, are successful examples of diversion
strategies.
Contemporary flood adaptation programmes therefore suppose a notion of
acceptable loss of safe ground and often a notion of acceptable damage. These
two notions are both dynamic (as they cannot be exactly predefined) and
reflexive (as by accepting flooding in certain areas of a land or city, other parts
of the territory will be spared). The newly designated ‘river space’ thus
encompasses much more than the surface of the stream, but refers to the
necessary space of fluctuation, whose contours are not fixed but rather gradual
(from permanent stream, seasonal riverbed, foreshore, retention areas, flood
zones); each level of permeability allowing different activities to take place, as
long as primary hydraulic functions are guaranteed. In this context, the
redefinition, expansion or transformation of flood-prone areas exclude any
purely objective, unique and final configuration, but imply complex negotiations,
painful arbitrages and dynamic designs to reach optimal risk-safety distribution
and land valorisation. In the investigated projects (listed in Table 1), this
redefinition resulted in a new territorial deal among owners, users and
beneficiaries of the adaptation project, involving both material elements (land
and infrastructure) and immaterial ones (value of areas and degree of risk
allocated to them), all merged into a new territorial structure. The studied cases
have shown that this new deal has clear implications on, among others, the limits
of flood-prone areas, land statuses, related rules, as well as on practices of
riverine users, as presented below.
Table 1: The nine flood-related projects studied.
Flood adaptation
Country River Location Spatial type
intervention
BE Scheldt Antwerp Floodplain expansion Intra-urban park
CH Rhône Valais River widening Mixed-use valley

Floodplain
DE Isar Munich, Bavaria Intra-urban park
restructuration
Floodable urban Intra-urban
FR Garonne Bordeaux
development development
FR Isère Isère, Rhône-Alpes Controlled flood fields Mixed-use valley
Groningen, Emergency retention
NL Eemskanaal Peri-urban extension
Meerstad lake
High-water floodplain
NL Maas Overdiepse polder Agricultural polder
extension
High-water diversion Agricultural polders
NL Merwede Noordwaard
channel and nature area
Floodplain
PT Mondego Coimbra Intra-urban park
consolidation
4 Design as negotiation platform

4.1 New negotiation frameworks
Contrary to traditional vertical flood protection, adaptive approaches and

horizontal interventions often generate strong resistances from land owners
and users, as they imply radical changes onto public and private property, and
affect the global land-use distribution of the area. In several cases, ontological
discussions arose, questioning the legitimacy of each of the competing land uses
and, more generally, the priority that society as a whole should give to each of
them, opposing productive functions (such as farming) to functions considered
unproductive (nature and recreation). A more symbolic dimension plays as well
a role in the negotiations: giving back to water a space that has been gained on
rivers and marshlands through centuries of land reclamation can be (wrongly)
interpreted as a regression from a cultivated or otherwise explored territory to a
natural state. Yet, the extensive investigations and technical means needed to
implement such spaces show that even the new ‘space for the river’ is primarily
the result of a design intervention [3]. It appears thus difficult but crucial for
local authorities to articulate the different scales and terms of the equation in
order to install a positive climate for negotiations.
4.2 New perimeters
While river space has long been defined as a negative space, progressively
reduced to maximize productive areas and expand building lands, the current
shift from flood defence to flood adaptation implies a reverse approach that first
defines the space needed to accommodate expected high waters, and
subsequently seeks to adapt the surrounding areas to provide the needed
capacity. Contrarily to the passive definition of flood zones, the definition of
adaptive measures, though elaborated with scientific tools, remains in essence a
political choice in its spatial translation, which implies a consensus on the
principle of the intervention and its perimeter. The interventions decided within
the Room for the River programme, located along river courses, were motivated
by the raise of national norms for river capacities that followed the 1993–1995
near flooding along the Rhine and Meuse rivers. For each measure, the type of
intervention, the financial means and expected effects on water level were set,
while the precise definition of river space and flood areas was left to regional and
local players, in collaboration with the national water authorities. The definition
of the new flood zones within the Noordwaard area and Overdiepse polder
eventually incorporated various elements: efficiency in hydrological terms, cost
targets, spatial quality, ecological value and sustainability of remaining farms
and dwellings. In the Swiss upper Rhône Valley, the decision of widening the
riverbed in order to increase its capacity was set by the Canton authorities at
the turn of the century, but various options can still be implemented locally,
including reinforcing the existing dikes or dredging the river. There again,
agreeing on the general means and objectives sets a discussion frame, but merely
opens the negotiations that will ultimately modify the area where water will be
allowed to fluctuate with more or less freedom.
Figure 1: Spatial adaptation in the Noordwaard area – Room for the River,
the Netherlands (Rossano).
4.3 New status
Territorial flood adaptation implies in most investigated cases a change in land

status. This change is not necessarily binary – from protected land to floodable
area – but more often combinatory: for example, urban public space embraces
river expansion zones (Antwerp, Coimbra), or farmland is used as emergency
storage area (Overdiepse polder, Isère Amont). Introducing a flood-related
function in a given site often goes with a loss of its value: safety is reduced
locally in order to increase in a wider area. In the article “Who likes to live in the
calamiteitenpolder”, the Dutch newspaper NRC echoed the debate following the
proposal of the Luteyn governmental commission to designate several
emergency retention polders, where waters from the Meuse and Rhine rivers
could be diverted to in case of threatening high waters [10]. House owners
complained that their properties had lost in value, even though the proposal was
still at an early stage. In the French Isère Valley, opposite protests were heard
when regional authorities announced the creation of 16 champs d’inondation
contrôlée closed for construction: local representatives saw land prices soar
around one of the designated flood zones, threatening municipal housing
policies. In both cases, local economies were influenced by the mere eventuality
of a status change that would turn (potential) building land into designated
flood area.
However, status changes can have positive global effects, and should
therefore not only be seen in terms of risk catchers and beneficiaries, but also in
terms of combinatory opportunities. For example, farmland used for flood
adaptation is itself, by essence, located in a flood-prone area, and can thus
benefit from explicit agreements with local authorities. As it appeared to farmers
of the Groningen province during the negotiations held with the Water Board,
they were actually better off if their land was identified as “emergency polder”
and covered by a compensation guarantee, than not insured and still in a flood
zone [11]. Cities are not left aside from major status modifications induced by
changes in flood adaptation strategy. Open public spaces along the river are then
the most obvious urban land use to absorb flood adaptation projects. However,
status change can also take place in the opposite direction, from flood area to
building land within new conditions (Bordeaux), or from restricted floodplain
to open public recreation space (Munich), which in its turn increases quality of
life and values up the immediate surroundings. However, as the Munich case
shows, new status generates new practices, which are not always foreseen, as the
new combinatory land uses overlap various regulations and mores. Local
authorities welcome the success of the new Isar River space, but simultaneously
struggle to control crowds’ behaviour in what is altogether a new kind a public
park, a nature area and still a floodplain.
Figure 2: From floodplain to urban beach: Isarraum, Munich,

June/September 2013 (Rossano/Kuenzel).
4.4 New rules and practices
Flood-prone spaces officially acknowledged as such not only undergo changes in

perimeter and status, but also in the way various activities can take place and be
regulated, bringing new challenges for local authorities and citizens. The robust
and simple design of the Isarplan facilitates the maintenance and post-flood
restoration, but also introduces a new freedom within the city, allowing
behaviours that are generally banned from historical parks and squares. Vast,
informal and less regulated, the flood-prone public spaces offer freedom of use
and the thrill of finding oneself in a risk area. The downside of this new freedom,
as it appeared there in recent years, is the difficulty to offer basic facilities and to
protect ground and vegetation in a space that can host more than 30,000 visitors
on a sunny weekend and be covered by high waters a week later. The Isar space
is altogether loved for but also victim of its dynamic nature, spatial simplicity
and low regulation, illustrating the need for a new balance between control and
laisser-faire – for the river and for its visitors. Within the built area, the
Bordeaux Brazza case is perhaps the one that better illustrates changing rules,
since it is a typical urban regeneration project within a flood-prone zone. Here,
land-use regulations were fully reviewed, for example to make possible the
conciliation between elevated ground-floors and accessibility for disabled
people, or to guarantee that every new building is as much as possible
hydraulically transparent. Finally, outside the city, where extreme water
discharges are temporarily directed towards farmlands, status change also
implies new rules and agreements. Once defined as diversion stream or

emergency retention area, hydraulic functionality becomes an extra constraint in
farming and nature areas, as no obstacles should hinder the expected effects of
controlled inundation. Depending on the predicted flood frequency, costly
infrastructures such as irrigation systems or glasshouses are to be avoided to
limit potential damages, but also uncontrolled vegetation growth that diminishes
permeability. Flood adaptation in agricultural areas furthermore brings along
new recreational functions, which are not always welcomed by local farmers and
dwellers, and need again careful design negotiations to combine recreation, risk
and productive activities.
5 Design roles within flood adaptation

An adaptive perspective, looking at the territory from the ‘point of view’ of
the river, implies an important shift in the planning process: besides the
indispensable knowledge of hydrologists and civil engineers, active investigation
into the social, economic and cultural characters of an extensive area is needed,
in order to identify in the concerned territory the best adaptation strategy
potentially embedded in it. This implies a good understanding of its morphology
(seen as the materialization of functions and interests assembled into a dynamic
spatial structure), as well as an ability to mentally manipulate this structure and
envision transformation possibilities. These abilities, developed in architectural
practice, appear useful to address the layered and spatial nature of this territorial
adaptive approach. In each negotiation process analysed, the design of the river
space has fulfilled various roles. Three rough categories emerged from the
collected data so far; their polishing is being pursued while the authors deepen
their analyses.
5.1 Design as eye-opener
Projects today implemented or under construction show that sketches and

practical spatial proposals facilitate the appropriation by local players, even at an
early stage of development. In this sense, the process that took place in Sion, the
capital city of Valais, is an exemplary illustration. Short after the launch of
the Rhône 3 programme, the city’s urban planning department made a proposal
to commission a design study to investigate potential changes in the city’s
relation to the river, but this was rejected by the city council, which considered
Rhône 3 a strict flood defence project (thus a prerogative of the Canton and not
eligible for municipal funding). Yet, both City and Canton welcomed, in 2009,
an initiative from the Chair of Landscape Architecture of Prof. C. Girot
(ETH Zurich) to organize a landscape design studio on the same theme. With
their support, the students’ visionary works were shown in an exhibition in Sion,
in 2010. Mr. Gross, one of the persons in charge of urbanism within the
municipality, recalls that “through the students’ projects, we could raise
awareness for the potentiality of this project, that it was not just a problem of
security but also a formidable opportunity to bring quality to the city along this
river” [12]. He also stressed that “the exhibition was a success, and people really
appropriated the term Sion-sur-Rhône”, title given to the design studio and to the
following publication [13]. A design competition for the reconfiguration of the
public spaces along the Rhône was eventually organized a year later, and is, right
now, in the process of further detailing by the winning design team. Although the
initial design envisioned by ETH Zurich’s students was not literally adopted, it
bridged the gap between City and Canton at a crucial moment in the planning
process (namely when urban ambition and flood mitigation could be connected
for the benefit of both), and it mobilized a population that hitherto had shown no
interest in what they saw as an abstract and purely technical issue.
5.2 Design as clarifier
The design competition held in Munich for the most central segment of the
Isarplan shows how spatial design can reveal latent expectations and oppositions.
Differently from the strictly internal process that supported the restoration of the
southern part of the river, the project commissioners decided in 2006 to organize
a landscape design competition on the Isar segment crossing the city centre. The
winning design envisioned a central linear sculptural element separating
the main stream and the new recreational open space, acknowledging the existing
technical constraints that would make impossible to set the river free. The second
prize was awarded to a completely different proposal, with an organic, informal
design. Yet, a public quarrel followed, showing that a significant part of the
population had expected a more spectacular ‘renaturation’ project and rejected
the urban aspect of the winning design. A period of intense and often emotional
discussions followed, involving city, local districts, water board and citizens,
giving the opportunity to express wishes and constraints, and eventually leading
to a consensual proposal that guaranteed the safety of urban infrastructures and
still suggested a certain natural freedom, most wanted inside the city. In the
words of the former Head of the city planning department, “it was important to
show, on the one hand, how little freedom there is when the river is so important,
but, on the other, to speak with people about this limited freedom we have,
because people perhaps expected something much more impressive. (...) These
competitions were more an education project. It was necessary to communicate”
[14]. Through an intense debate that could only have been ignited by concrete
proposals, the Isarplan left the secluded world of environmental and
technological expertise, on the one hand, and the realm of romantic dreams, on
the other. The design was thus not the result of a predefined image, but initiated
new perspectives leading to alternative trajectories.
5.3 Design as matchmaker
The opening of the design process to a wider panel, in the programmatic phase,
offers the opportunity for participants to match more general expectations and
possibilities with specific options of spatial configuration. Shared scenarios can
then be created through the discussion about the distribution of land
and investments, and the elaboration of the envisioned spatial framework and
physical interventions. This matchmaker role was well illustrated by the process
of elaboration of the Meerstad project: during the planning workshops,
participants received basic programme elements (represented by on-scale pieces
of coloured paper, proportional to the land requested for water, wetlands, woods,
housing and industry), and by playing with them onto the area’s map, they could
quickly elaborate spatial distribution scenarios. Apart from allowing all present
players to envision their preferred options, this scenario-based participative
approach made them conscious of the difficulty of combining various elements
into a legible and attractive spatial framework. The community participation in
the design process that took place in the Overdiepse polder was even more
radical, as local inhabitants actually anticipated the planning process, right after
the area was identified by the government as a suitable floodplain extension.
They had to deal with a relatively simple equation (but a sensitive matter), as it
was clear that not all 16 existing farms could sustain their activity in the area.
They grasped the chance to develop their own plan, with support from the
national and regional authorities, choosing the most convenient project from
their point of view (the reconfiguration of the whole polder into a temporary
expansion space for high waters with nine heightened platforms for the future
farms). In this particular case, the design allowed the building of a consensus not
only between authorities and farmers, but first of all among local players
themselves, who could better deal with the economic and human aspects of the
project, and translate them into an agreed and shared framework.
6 Final considerations
Territorial design, by nature, is a complex task of organizing multiple collective
intentions, uses, desires, possibilities and constraints in a balanced, sensitive and
also inspiring spatial arrangement. Yet, when the existing constraint is linked to
riverine flood risk, the designers’ task is made even more challenging, as the
possibility to turn flood proneness into a great spatial opportunity is latently
offered. The analysed projects showed that despite all its complexities, the
design of flood-prone spaces can be performed as an open negotiation platform.
As a dynamic process, design is allowed to evolve: it can take into consideration
natural fluctuations as well as ever-changing sociocultural aspects, and can also
orchestrate the interdisciplinary approach needed to balance (apparently)
concurrent objectives with different time horizons. As an open platform,
designing river spaces comprises two complementary characteristics: by
fostering a wider participation, it promotes a valuable interchange of inputs
between stakeholders and designers, where unforeseen combinatory options can
actually emerge. On the other hand, by reintroducing the free will that
characterizes the practice of spatial design [15], it facilitates the emergence of
collective choices and consensual territorial visions beyond problem solving. Not
only this can eventually increase players’ sense of project ownership but surely
acts as a powerful sensitization tool, bringing flood risk closer to people’s daily
lives, and helping democratic societies build positive and shared answers to
flood risk challenges.
Acknowledgements
This study was funded by the Swiss National Science Foundation and the
Portuguese Foundation for Science and Technology, through PhD grants. The
authors are thankful to all designers and territorial players who generously
provided information to both PhD researches.
References
[1] Foqué, R., Building knowledge in architecture. UPA: Brussels, p. 174,
2010.
[2] Harries, T., Feeling secure or being secure? Why it can seem better not to
protect yourself against a natural hazard. Health, Risk and Society, 10(5),
pp. 479-490, 2008.
[3] Klijn, F., Bruin, D., Hoog, M.C., Jansen, S. & Sijmons, D.F., Design
quality of room-for-the-river measures in the Netherlands: role and
assessment of the quality team (Q-team). International Journal of River
Basin Management, 11(3), pp. 287-299, 2013.
[4] Viganò, P., Extreme cities and bad places. International Journal of
Disaster Risk Science, 3(1), pp. 3-10, 2012.
[5] Barroca, B. & Hubert, G., Urbaniser les zones inondables, est-ce
concevable? Développement Durable et Territoires.
http://developpementdurable.revues.org/index7413.html
[6] De Meulder, B. Personal communication, 5 October 2013,
Architect/Urbanist, WIT Architecten, Louvain, Belgium.
[7] Ruimte voor de rivier. Samen werken aan een veilig en mooi
rivierengebied.
http://www.ruimtevoorderivier.nl/media/78773/corporate_brochuredefdoo
rlopend.pdf
[8] Ministerie van Verkeer en Waterstaat, Ministerie van Volkshuisvesting,
Ruimtelijke Ordening en Milieubeheer, Ministerie van Landbouw, Natuur
en Voedselkwaliteit. Nationale Waterplan 2009-2015, pp. 6, 71-73, 2009.
[9] Warner, J., Edelenbos, J. & van Buuren. A., Making space for the river:
governance challenges (Chapter 1). Making space for the river, eds. J.
Warner, A. van Buuren & J. Edelenbos. IWA: London, pp. 1-13, 2012.
[10] Schreuder, A. Wie woont er graag in de calamiteitenpolder?, Rotterdam,
02.12.2002, http://vorige.nrc.nl/binnenland/article1554360.ece
[11] Van Hall, A. Personal communication, 13 March 2013, Dike Master,
Water Board Hunze & Aa’s, Veendam, the Netherlands.
[12] Gross, D. Personal communication, 24 July 2013. Municipality Urbanist,
Sion, Switzerland.
[13] Girot, C., Rossano, F. & Duner, I., Sion-sur-Rhône, un nouveau paysage
pour la vallée du Rhône à Sion. GTA: Zurich, 2010.
[14] Tahlgott, C. Personal communication, 25 June 2013. Former Head of the
planning department, Munich, Germany.
[15] Sijmons, D., Mind the gap (Chapter 2). Rising waters, shifting lands, eds.
C. Girot & F. Rossano, GTA: Zurich, pp. 9-12, 2012.
Author index
Adegoke P. B. ............................ 23 Joseph P. .......................... 175, 201
Ahilan S. ............................ 13, 113 Jukrkorn N. ................................ 75
Al Khaddar R. M. ...................... 23
Allen D. ................................... 113 Kang J. ....................................... 87
Anton F. ................................... 239 Kilsby C. .................................. 113
Ardagna D................................ 215 Kim H. ....................................... 89
Arias C. .................................... 215 Kobayashi K. ........................... 251
Arthur S. .................................. 113 Kocmanova P........................... 263
Atherton W. ............................... 23 Kopecny L. .............................. 263
Atun F. ..................................... 215
Lamond J. ................101, 113, 125,
Beddoes D. W. ......................... 151 ................................. 163, 175, 187
Berg H...................................... 227 Lawson E. ................................ 113
Bhattacharya-Mis N. ................ 163
Booth C. A. .............................. 151 McGillivray E. ......................... 239
Burian F. .................................. 263 Mant J. ..................................... 113
Maskrey S. ............................... 113
Cameron D. .............................. 139 Mazuran M. ............................. 215
Cameron R. .............................. 201 Mezouaghi M........................... 239
Carnacina I. ................................ 49 Minucci G. ............................... 215
Chen J.-C. .................................. 35 Mioc D. .................................... 239
Choi C. ....................................... 87 Mofford L. ............................... 239
Chuang M.-R. ............................ 35 Mohssen M. ................................. 3
Coates G. ................................... 13 Molinari D. .............................. 215
Mount N................................... 113
Ebeltoft M. ............................... 227
Everett G. ......................... 101, 113 Nielsen J. ................................. 227
Fenner R. ................................. 113 Panya O. .................................... 75

Proverbs D. ...................... 139, 175
Glenis V. .......................... 113, 125
Guan D..................................... 113 Rose C. .................................... 125
Rossano F. ............................... 287
Hawe G. I. .................................. 13
Hoang L. .................................. 113 Sachdev H. ................................. 75
Hobeica L. ............................... 287 Sleigh A. .................................. 113
Smith L. ................................... 113
Igarashi M. ............................... 251 Stephenson J. ........................... 201
Stopp H. ................................... 277
Jemberie A. ................................ 49 Strangfeld P. ............................ 277
Jeng C.-J. ................................... 35
Ji J. ............................................. 87 Tang P...................................... 239
Jilek T. ..................................... 263 Thomas M. ................................. 61
Thorne C. ................................. 113 Wright N. G. .............................. 13

Tsujimoto T. ...................... 61, 251 Wright N. ................................. 113
Vaganay M. ............................. 201 Yi J. ........................................... 87

Yu M.......................................... 87
Wang J.-S................................... 35
Wilkinson S. J. ......................... 125 Zalud L. ................................... 263
...for scientists by scientists
Flood Early Warning Systems

Knowledge and Tools for their Critical Assessment
D. MOLINARI, S. MENONI, F. BALLIO, Politecnico di Milano, Italy
This book presents the results of an ambitious research activity designed to understand
why Early Warning Systems (EWSs) fail. However, from the beginning, the objective
of the research proved to be challenging for two reasons. First, as yet there is not
a shared understanding of what an EWS is among either research or practitioner
communities. Second, as a consequence, it is equally unclear when an EWS can be
considered successful or not. Because of this, the research needed first to define EWS
and identify its components, functions, peculiarities, and weak points. Only at that
point was a first attempt to evaluate EWSs performance possible.
Flood Early Warning Systems Performance is organised according to the conceptual
steps required by the research. In part I the “open questions” about the definition and
the role of EWSs are handled, the aim being the identification of how to evaluate
EWSs effectiveness/performance. Part II focuses on the real aim of the research,
providing concepts and tools to assess EWSs performance; suggested tools are also
implemented in a case study to describe how they can be applied in practice. The
sections are independent of each other to allow readers to focus only on the content
they are most interested in.
The book is designed for a wide audience. The book can serve as a sort of manual
for EWSs designers, managers, and users, but also has appeal for general readers
with an interest in the subject. While the focus of the book is flood risk in mountain
regions, most of the results can be applied to other hazards as well.
Traditionally early warning systems (EWSs) have been identified with monitoring and
forecasting systems and their assessment has therefore focused only on the accuracy of
predictions. The authors propose a shift in thinking towards the more comprehensive
concept of total warning systems, where monitoring and forecasting systems are
coupled with risk assessment, emergency management and communication aspects.
In line with this, a new approach to assess EWSs is proposed that is based on system’s
capacity of reducing expected damages, with the hope that improved EWSs will result.
ISBN: 978-1-84564-688-2 eISBN: 978-1-84564-689-9
Published 2013 / 196pp / £84.00
Tsunami
From Fundamentals to Damage Mitigation
Edited by: S. MAMBRETTI, Universidade Estadual de Campinas, Brasil
A tsunami is a series of water waves caused by the sudden displacement of a large

volume of a body of water, typically an ocean. Earthquakes, volcanic eruptions and
other underwater explosions (including detonations of underwater nuclear devices),
landslides, glacier calving, meteorite impacts and other disturbances above or below
water all have the potential to generate a tsunami.
These waves are very different from normal sea waves, because their wavelength is
far longer. Large events can generate wave heights of tens of metres and therefore,
although the main impact of tsunamis is to coastal areas, their potential destructive
power is enormous and they can affect entire ocean basins; the 2004 Indian Ocean
tsunami was among the deadliest natural disasters in human history with over 230,000
people killed in 14 countries bordering the Indian Ocean.
Tsunami: From Fundamentals to Damage Mitigation comprises seven chapters,
dealing with the different aspects of the field. The first chapter deals with the different
types of tsunami and their historical data. Chapter 2 describes an inverse type solution
to find a posteriori of the tsunami waveform. One of the main problems with tsunamis,
described in Chapter 3, is how to assess the flooding they produce. Chapter 4 deals
with the very important topic of Early Warning Systems. Chapter 5 not only studies
the behaviour of RC buildings under the 2011 Japanese Tsunami but puts forward
a series of recommendations. One of the most damaging aspects of tsunamis is the
damage to infrastructure and building systems. Chapter 6 discusses this along with
providing guideline measures to take in the future. Finally, Chapter 7 studies the
important problem of health and related issues due to tsunami disasters.
Series: Safety & Security Engineering
ISBN: 978-1-84564-770-4 eISBN: 978-1-84564-771-1
Published 2013 / 168pp / £76.00
Flood Risk Assessment and Management

Edited by: S. MAMBRETTI, Politecnico Di Milano, Italy
This volume is the first in a new series that covers various aspects of Safety and Security
Engineering with the aim of developing a comprehensive view on risk mitigation. This
volume is devoted to floods, since one-third of annual natural disasters and economic
losses, and more than half of the victims of natural disasters are flood-related.
The risk from flooding, and the demand for protection from it, has been growing
exponentially as a result of a burgeoning global population and growing wealth,
climate change and urban development. These factors make it imperative that we
change the way flood risk is managed.
Knowledge and scientific tools play a role of paramount importance in the strain of
coping with flooding problems, along with capacity building in the context of political
and administrative frameworks. Therefore, governments need to establish clear
institutional, financial and social mechanisms and processes for flood risk management
in order to ensure the safety of people and property and, thereby, contribute to both
flood defence and sustainable development.
The present volume contains selected papers presented at Conferences organised by
the Wessex Institute of Technology. The papers have been revised by the Authors to
bring them up to date and to integrate them into a coherent understanding of the topic.
It covers: Risk Assessment; Mathematical Models for Flood Propagation; Effect of
Topographic Data Resolution; Social and Psychological Aspects; Decision Making
and Management; Legislations and Directives; Alternatives in Flood Protection;
Response and Recovery; Damages and Economic-related Problems; Case Studies
The quality of the material makes the volume a most valuable and up-to-date tool
for professionals, scientists, and managers to appreciate the state of the art in this
important field of knowledge.
Series: Safety & Security Engineering
ISBN: 978-1-84564-646-2 eISBN: 978-1-84564-647-9
Published 2012 / 160pp / £65.00

Water Front

Uploaded by

Copyright:

Available Formats

Water Front

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Water Front

Uploaded by

Copyright:

Available Formats

Flood Recovery, Innovation

WIT Press publishes leading books in Science and Technology.

International Scientific Advisory Committee

B Abersek University of Maribor, Slovenia G Belingardi Politecnico di Torino, Italy

For USA, Canada and Mexico

Computational Mechanics Inc

British Library Cataloguing-in-Publication Data

A Catalogue record for this book is available

© WIT Press 2014

The present volume contains papers presented at the Fourth International

Flooding is a global phenomenon that claims numerous lives worldwide each

The conference provided a forum for researchers, academics and practitioners

Section 1: Flood modelling

A new approach for flood forecasting of river flows

Agent-based modelling and inundation prediction to enable

A novel simple method for measuring the velocity of dam-break flow

Numerical simulation of the inundation area for landslide-induced

Section 2: Risk assessment

A practical approach to floodplain mapping for large-scale

Vulnerability to flood risks in Japanese urban areas: crisis management

Section 3: Flood management

Community-based flood risk management: lessons learned from the

Section 4: Considering ‘Blue-Green’ approaches to

A conceptual framework for understanding behaviours and attitudes

Delivering and evaluating the multiple flood risk benefits in

Modelling a green roof retrofit in the Melbourne Central Business District

Section 5: Property-level flooding and health consequences

Improving the uptake of flood risk adaptation measures for

Waterproofing basement apartments: technical insights of a

An investigation of patterns of response and recovery

Resilient reinstatement: what can we learn from

The role of flood memory in the impact of repeat flooding

Section 6: State-of-the-art flooding-damage survey and assessment

Implementing tools to meet the Floods Directive requirements:

Flood damage survey after a major flood in Norway 2013:

Section 7: Emergency preparedness and response

An overview of the applications for early warning and mapping

Multi-robot system for disaster area exploration

Section 8: Adaptation to flood risk

Floating houses: an adaptation strategy for flood preparedness

Design as a negotiation platform: new deals and spatial adaptation in

Author index .................................................................................................. 299

A new approach for flood forecasting of

2 Flood modelling of the Leith River

2.1 Model development and formulation

For to have the minimum distance “difference” from Qt, Qt - should be

< Qt - , R >=0, where <X,Y> = E [XY] in Hilbert space (2)

< Qt - ∑ , R > =0, j = L1, L1+1, …, L2 (3)

Figure 1: Locations of rainfall and flow sites in the Leith catchment.

2.2 Model calibration

Figure 4: Results of the calibration process.

account for hydrologic abstractions, or losses from rainfall before it becomes

Table 1: Observed vs. forecasted Leith River peak flows.

Event Peak Observed Forecast R2 FC %

Agent-based modelling and inundation

3 Agent-based modelling and flood modelling

Virtual Geographic Agent Interaction

Flood modelling Agent-based simulation

Figure 1: Overview of modelling and simulation framework.