Hydrological Sciences Journal
Hydrological Sciences Journal
Hydrological Sciences Journal
To cite this article: Hsiang-Kuan Chang, Yih-Chi Tan, Jihn-Sung Lai, Tsung-Yi Pan, Tzu-Ming Liu & Ching-Pin Tung (2013)
Improvement of a drainage system for flood management with assessment of the potential effects of climate change,
Hydrological Sciences Journal, 58:8, 1581-1597, DOI: 10.1080/02626667.2013.836276
Taylor & Francis makes every effort to ensure the accuracy of all the information (the “Content”) contained
in the publications on our platform. However, Taylor & Francis, our agents, and our licensors make no
representations or warranties whatsoever as to the accuracy, completeness, or suitability for any purpose of the
Content. Any opinions and views expressed in this publication are the opinions and views of the authors, and
are not the views of or endorsed by Taylor & Francis. The accuracy of the Content should not be relied upon and
should be independently verified with primary sources of information. Taylor and Francis shall not be liable for
any losses, actions, claims, proceedings, demands, costs, expenses, damages, and other liabilities whatsoever
or howsoever caused arising directly or indirectly in connection with, in relation to or arising out of the use of
the Content.
This article may be used for research, teaching, and private study purposes. Any substantial or systematic
reproduction, redistribution, reselling, loan, sub-licensing, systematic supply, or distribution in any
form to anyone is expressly forbidden. Terms & Conditions of access and use can be found at http://
www.tandfonline.com/page/terms-and-conditions
Hydrological Sciences Journal – Journal des Sciences Hydrologiques, 58 (8) 2013 1581
http://dx.doi.org/10.1080/02626667.2013.836276
Hsiang-Kuan Chang1 , Yih-Chi Tan1,2,3 , Jihn-Sung Lai1,2,3∗ , Tsung-Yi Pan2 , Tzu-Ming Liu1 and
Ching-Pin Tung1,2,3
1
Department of Bioenvironmental Systems Engineering, National Taiwan University, Taipei, Taiwan
jslai525@ntu.edu.tw
2
Center for Weather Climate and Disaster Research, National Taiwan University, Taipei, Taiwan
Downloaded by [George Mason University] at 20:18 24 December 2014
3
Hydrotech Research Institute, National Taiwan University, Taipei, Taiwan
Received 30 March 2012; accepted 15 February 2013; open for discussion until 1 May 2014
Citation Chang, H.-K., Tan, Y.-C., Lai, J.-S., Pan, T.-Y., Liu, T.-M., and Tung, C.-P., 2013. Improvement of a drainage system for flood
management with assessment of the potential effects of climate change. Hydrological Sciences Journal, 58 (8), 1581–1597.
Abstract Runoff discharge in the Tuku lowlands, Taiwan, has increased with land development. Frequent floods
caused by extreme weather conditions have resulted in considerable economic and social losses in recent years.
Currently, numerous infrastructures have been built in the lowland areas that are prone to inundation; the measures
and solutions for flood mitigation focus mainly on engineering aspects. Public participation in the development
of principles for future flood management has helped both stakeholders and engineers. An integrated drainage–
inundation model, combining a drainage flow model with a two-dimensional overland-flow inundation model is
used to evaluate the flood management approaches with damage loss estimation. The proposed approaches include
increasing drainage capacity, using fishponds as retention ponds, constructing pumping stations, and building
flood diversion culverts. To assess the effects on the drainage system of projected increase of rainfall due to
climate change, for each approach simulations were performed to obtain potential inundation extent and depth in
terms of damage losses. The results demonstrate the importance of assessing the impacts of climate change for
implementing appropriate flood management approaches.
Key words drainage; inundation; flood management; public participation; climate change
Amélioration des systèmes de drainage pour la gestion des crues, avec évaluation de l’effet de
différents scénarios de changement climatique
Résumé Les débits de surface dans la plaine de Tuku, à Taiwan, se sont accrus avec l’aménagement du territoire
et des activités humaines. Les crues fréquentes causées par des conditions météorologiques extrêmes ont engendré
des pertes économiques et sociales considérables au cours des dernières années. Actuellement, de nombreuses
infrastructures ont été construites dans la zone de plaine sujette aux inondations, les mesures et des solutions pour
l’atténuation des inondations se concentrant principalement sur les aspects d’ingénierie. La participation du public
dans les municipalités aide à la fois les parties prenantes et les ingénieurs à déterminer les principes de la gestion
des crues afin d’assurer une communication efficace. Pour évaluer les approches de gestion des crues et estimer les
dommages, nous utilisons un modèle intégré drainage-inondation combinant un modèle de flux de drainage avec
un modèle en deux dimensions d’inondation due aux écoulements de surface. Les approches proposées compren-
nent l’augmentation de la capacité de drainage, l’utilisant d’étangs comme bassins de rétention, la construction
de stations de pompage et la construction de canaux de dérivation des crues. Dans le cadre de l’étude des effets
de l’augmentation prévue des précipitations sur le système de drainage sous l’effet du changement climatique,
des simulations ont été effectuées pour traduire la hauteur et l’étendue potentielle de l’inondation en termes de
dommages, afin d’évaluer chaque approche proposée. Les résultats indiquent que la zone de plaine potentielle-
ment sujette aux inondations, dans laquelle le drainage est relativement difficile, doit être analysée en évaluant les
impacts du changement climatique afin de mettre en œuvre des approches de gestion des crues appropriées.
Mots clefs drainage; inondations; gestion des crues; participation du public; changement climatique
management usually depends on terrestrial character- indispensable for simulating the flood propagation
istics and local geography, and includes floodways, phenomena in wide lowland areas. Detailed hydrody-
and flood storage in foothill reservoirs (Roos 2006) or namic information of complex topography is neces-
rice fields (Chang et al. 2007). With the demand for sary for accurate evaluation of losses caused by flood
development, urban areas in lowlands and other flood- damage. Therefore, a drainage–inundation model is
prone regions, along coasts, river floodplains and in required to provide information on flood extent and
inland depressions, are growing worldwide (Vlotman depth of surface inundation. The 2D shallow water
et al. 2007). equations are used to describe the free surface flow
The common structural measure for flood protec- phenomenon for the analysis of distributed surface
tion in lowland or flood-prone areas is the building with adverse slopes and irregular geometry in flood-
of dikes along rivers or major channels (de Bruin plains (Guo et al. 2008, Lai et al. 2010a). The
2006). If water accumulates, because of heavy pre- 2D overland-flow model based on non-inertia wave
cipitation, where drains are lacking or their discharge surface flow dynamics, which usually neglects the
capacity is exceeded, flooding may occur due to water inertial terms but considers the backwater effect, is
overtopping dikes to produce widespread flood dam- applicable in practice to simulate regional overland
age over lowlands. To manage flood hazards, it is flow in lowland areas. The non-inertia wave model
vital to implement an effective flood risk manage- was first proposed by Cunge et al. (1976), and similar
ment concept. Although flooding cannot be elimi- approaches have been developed and applied by sev-
nated completely, the consequences of flooding can eral researchers (Vongvisessomjao et al. 1985, Hsu
be mitigated by appropriate actions in the broader et al. 2000, Bates et al. 2003, Lai et al. 2010b, Pan
context of integrated river basin management. In com- et al. 2011). Hsu et al. (1990) compared the per-
pliance with the concept of “room for river and formance of various numerical schemes on the 2D
people,” a systematic approach to flood risk manage- non-inertia wave equations. Among those schemes,
ment is to use the flood-prone lowlands efficiently. the alternating direction explicit (ADE) scheme per-
This leads to flood defence prioritisation for pro- forms well for predicting inundation processes with
tecting people and property, and also creating space various land use, irregular topography and pumping
for water storage and channel cross-section modifica- operations.
tion. Overall, implementation of appropriate actions In this study, a 2D diffusive-wave model, that is
to enhance flood security is both possible and nec- the 2D overland-flow inundation model with an ADE
essary to reduce the exposure and vulnerability of scheme, was used for flood inundation simulation for
people and property to flood hazards. However, long- the Tuku area, Taiwan. With the increasing demand
term strategies for flood mitigation and control issues for land development in the Tuku lowlands, located
must incorporate measures that are perceptibly inte- in southern Taiwan (Fig. 1), the runoff discharge has
grated with other aspects, such as socio-economics, increased due to the greater extent of impervious
culture, nature and the environment. Therefore, public areas, such as rooftops, roads and other infrastruc-
participation involving local communities represents ture. Extreme flood events in recent decades have
a crucial framework of consideration for the selection caused substantial losses in this area. The lowlands
Improvement of a drainage system for flood management with the effects of climate change 1583
Downloaded by [George Mason University] at 20:18 24 December 2014
in the Tuku drainage system are protected by levees with a 2D overland-flow inundation model, was
along the banks. The water stage in the main channel adopted to simulate flood inundation. Inundation sim-
is usually higher than in the lateral channels during ulation with the projected future rainfall increase was
a flood. Once water stages are higher than the top performed to determine the potential consequences
of the levees along lateral channels, the flood may in terms of damage losses to evaluate each proposed
overflow to the surrounding lowlands. The drainage approach.
system in the Tuku area is composed of drainage
channels and pumping stations at the outlets of lat-
eral channels, which are designed for rainfall less
2 SITE DESCRIPTION
than that of the 5-year return period. For engineer-
ing purposes, most flood prevention structures are Taiwan experiences an average of four typhoons per
built for a design flood of a certain return period, year, two of which make landfall. Heavy rainfall is
which is determined through frequency analysis based associated with the typhoons. In addition, the south-
on historical hydrological records. However, extreme west monsoons are responsible for the occurrence of
weather events have occurred more frequently in heavy rainfall in southern Taiwan, and often cause
Taiwan and worldwide in recent decades. Moreover, substantial flooding and result in loss of life and prop-
the reports issued in 2007 by the Intergovernmental erty. The Tuku area is located in the coastal lowlands
Panel on Climate Change (IPCC) indicate that the of southern Taiwan and includes the Alian, Lujhu
frequency and intensity of extreme events might and Gangshan townships, of which Gangshan has the
increase in the future (Pachauri and Reisinger 2007). largest population. The outlet of the drainage system
Thus the flood prevention structures may experience is the confluence with the Agoden River, which is
more challenges than previously. Therefore, this study tidal.
investigated the potential inundation losses caused The watershed area of the Tuku drainage system
by increasing the design flood discharge to reflect is approx. 76 km2 . The main land uses are agriculture,
climate changes projected for the next 30-years. aquaculture (mainly fishponds), building, forestry and
This study analysed appropriate flood manage- the road-network. Agriculture and aquaculture form
ment approaches for an inundation-prone lowland the primary sources of income. Diverse crops are
area with assessment of climate change impacts. grown in the agricultural areas, including rice, sugar-
The importance of public participation was addressed cane, maize, vegetables and fruit trees (guava, longan,
to help both stakeholders and engineers determine mango and Chinese date). Inundation often occurs
the principles of flood management. A drainage– because the terrain is lower than neighbouring areas
inundation model, combining a drainage flow model and so many lowland areas have been developed into
1584 Hsiang-Kuan Chang et al.
Main Channel
Main road
Watershed
Boundary
PA
PB
PC
Agriculture field
Fish pond
Building PA: Yuku pumping station
Forest PB: Tandi pumping station
0 2 km Other 0 2 km PC: Wujiawei pumping station
fishponds; the aquaculture is based mainly on rais- Its peak rainfall intensity was 66.5 mm/h, and total
ing milk-fish. Recently, the rate of urbanization of rainfall was 895 mm. The inundation area of the flood
the townships has accelerated. The industrial parks was about 2300 ha in Gangshan Township (Fig. 3(a)).
and commercial activities are prosperous and, conse- The average inundation depth was up to 2.65 m
quently, industries and factories have replaced culti- in the town, and the flood lasted for 7 days. Road
vated land over time. The land-use map of the Tuku traffic along a 600-m length of highway was com-
area for 2009 is shown in Fig. 2(a). pletely interrupted for 83 h when the inundation depth
The Tuku drainage system consists of one main reached 1.5 m. In the case of Typhoon Haitang, the
channel, which is the main tributary of the Agoden 24-h cumulative rainfall was 447 mm, the peak rain-
River, and 21 lateral channels (total length approx. fall intensity was 79 mm/h, and the total rainfall
70 km) that flow to the main channel (Fig. 2(b)). 887 mm. The main inundation area (∼1000 ha) was
The main channel has been renovated over many around Gangshan Township with an average depth
years, but the embankment collapses occasionally. of approx. 1.8 m. A heavy rainstorm on 9 June
Most lateral channels flow through farmland or vil- 2006, caused torrential rain with a 24-h cumulative
lages if their flood protection levees fail. A number rainfall of 316 mm; the peak rainfall intensity was
of lateral channels are poorly maintained, usually 59 mm/h and the total rainfall was 412 mm. The
weedy with substantial sediment and narrow so their inundation area was approx. 860 ha, and inundation
cross-sections are insufficient to drain flood water. depth increased to an average of 1.1 m, mainly in
As shown in Fig. 1, another reason for flooding in the Gangshan Township. The 24-h cumulative rainfall
Tuku area is that the water stage at the downstream recorded during Typhoon Fanapi was 463 mm, and
end of the main drainage channel connecting to the a rainfall hydrograph of peak intensity 77 mm/h was
Agoden River is relatively high because of a tidal recorded during the event. The inundation area was
effect. approx. 950 ha and occurred mainly in Yuku, Tandi,
and Jiaxing villages.
Frequency analysis using 56 years (1952–2008)
3 HYDROLOGICAL DATA
of rainfall data was conducted to generate the hyeto-
Extreme flood events, including Typhoon Doug graph of the 24-h rainfall pattern (Fig. 4). The Pearson
(August 1994), Typhoon Haitang (July 2005), 9 June type III distribution (Phien and Jivajirajan 1983) was
storm (2006) and Typhoon Fanapi (September tested and used to estimate the total rainfall depth
2010) caused substantial damage in recent decades, at each raingauge station for various return periods
particularly in the Tuku watershed; Fig. 3 shows their and durations. Horner’s equation was subsequently
impacts in Gangshan Township. Four raingauge sta- applied to set up the correlation of rainfall intensity,
tions measure nearby precipitation, and are operated duration and frequency. For 24-h design rainfall, total
by the irrigation associations. The 24-h cumulative rainfalls of various events for 5-, 10-, 25-, 50- and
rainfall recorded during Typhoon Doug was 539 mm, 100-year return periods, and their corresponding peak
which exceeded the 100-year return-period rainfall. rainfall intensity are listed in Table 1.
Improvement of a drainage system for flood management with the effects of climate change 1585
(a) (b)
Downloaded by [George Mason University] at 20:18 24 December 2014
(c) (d)
Fig. 3 Photographs of flood inundation in Tuku area: (a) Typhoon Doug, 13 August 1994; (b) Typhoon Haitang, 18 July
2005; (c) Storm, 9 June 2006; and (d) Typhoon Fanapi, 19 September 2010.
Table 1 Design 24-h rainfall and downstream-boundary water stage in the Tuku watershed.
Return Total rainfall Peak rainfall Downstream-boundary Average inundation Maximum
period (mm) intensity water stage depth inundation area
(years) (mm/h) (m) (m) (hm2 )
∂A ∂Q
+ =0 (1)
Table 2 Correlation coefficients between simulated GCM ∂t ∂
results and historical observations in the baseline period
for mean monthly precipitation.
∂Q ∂(Q2 A) ∂H
GCM Station + + gA + gASf = 0 (2)
∂t ∂ ∂
ChiShan Meinong
where A is cross-sectional area, Q is the conduit flow
CSMK3 0.58 0.58
GFCM20 0.93 0.94 discharge, g is the gravitational acceleration, H is the
GFCM21 0.91 0.91 hydraulic head, Sf is the friction slope, t is the time,
IPCM4 0.45 0.42 and is the distance along the pipe or channel.
INCM3 0.97 0.96
MPEH5 0.97 0.97 At each node, a special hydraulic situation is
MRCGCM 0.81 0.82 the occurrence of surcharge and overflow. Surcharge
NCCCSM 0.91 0.92 occurs when all pipes entering a node are full, or when
MIMR −0.11 −0.10
HADCM3 0.53 0.56
the water surface at the node lies between the crown
of the highest entering pipe and the ground surface.
Improvement of a drainage system for flood management with the effects of climate change 1587
Overflow is a special case of surcharge that occurs calculation procedures can be found in Lai et al.
when the hydraulic gradient intersects the ground sur- (2010b).
face and water is lost from the sewer node (channel
cross-section) to the overlying surface system. The
SWMM can simulate hydrographs at each overflow 5.3 Model linkage and verification
cross-section; however, it cannot manage detailed The flow chart of the linkage of the integrated model
information, such as the inundation zones and depths is shown in Fig. 5. First, the RUNOFF block of the
caused by overflow water. However, the information SWMM uses the rainfall as input to calculate the
on inundation is indispensable for engineering design surface runoff discharge hydrographs. The EXTRAN
and flood mitigation measures. The propagation of block of the SWMM calculates the hydrographs of
overflow water in the lowland is discussed in the fol- overflow flow rate at channel sections when the
lowing section using the 2D overland-flow inundation surface runoff exceeds the design capacity of the
model. drainage system. The overflow hydrographs are sub-
sequently used as sources in the 2D overland flow
Downloaded by [George Mason University] at 20:18 24 December 2014
60
(a)
Rainfall intensity(mm/h)
50
40
30
20
10
0
2006/6/9 2006/6/9 2006/6/10 2006/6/10 2006/6/10
11:00 19:00 03:00 11:00 19:00
Time
80
(b)
70
Rainfall intensity(mm/h)
60
Downloaded by [George Mason University] at 20:18 24 December 2014
50
40
30
20
10
0 Fig. 7 Simulated and surveyed inundation areas in Tuku for
2010/9/19 2010/9/19 2010/9/19 2010/9/19 2010/9/20 Typhoon Fanapi.
8:00 13:00 18:00 23:00 4:00
Time
with peak intensity 77 mm/h was recorded during the
Fig. 6 Rainfall hydrographs for: (a) the 2006 storm event,
and (b) Typhoon Fanapi, 2010. event (Fig. 6(b)). The Manning’s roughness values
and the grid size of the drainage–inundation model
were the same as those used for calibration. The
The flood event of 9 June 2006 is presented as
inundation-prone areas surveyed by the 6th River
an example for model calibration. Its highest cumula-
Management Office were plotted and used in this
tive rainfall within 24 h was 391.5 mm, and a rainfall
study. As shown in Fig. 7, it was also revealed that
hydrograph with a peak intensity of 59 mm/h was
most inundation situations were appropriately simu-
recorded (Fig. 6(a)). The Manning roughness in the
lated by the model.
inundation model was calibrated by applying land-use
information. The inundation information was sur-
veyed and delineated by the 6th River Management 6 FLOOD MANAGEMENT
Office. By using 40 m × 40 m grid size in the sim-
ulation zones, the simulated result at each grid for To evaluate proposed flood management approaches,
the maximal inundation depth greater than 0.5 m was hydrological analyses for various return periods were
considered as inundated. Comparison of the surveyed performed to obtain total rainfall depths and rain-
and simulated inundation area and depth revealed that fall intensities (Fig. 4). The downstream boundary
the inundation situations were properly simulated by conditions were related to the flood stages of the
the model (Table 3). Typhoon Fanapi, which had the Agoden River at the watershed outlet and influenced
highest cumulative rainfall, 463 mm within 24 h, was by the tide (Table 1). Using digital elevation model
used for model verification. A rainfall hydrograph (DEM) data of 40 m × 40 m grid size, inundation
Table 3 Comparison of surveyed and simulated results for the 2006 storm event.
Total Inundation depth (m)
inundation∗
area (hm2 ) Average Yuku area Tandi area Jiafeng area
N N N
W E W E W E
S S S
Drainage systems Drainage systems Drainage systems
Inundation depth (m) Inundation depth (m) Inundation depth (m)
0.5–1.0 0.5–1.0 0.5–1.0
1.0–2.0 1.0–2.0 1.0–2.0
2.0 + 2.0 + 2.0 +
0 2 km 0 2 km 0 2 km
N N N
W E W E W E
S S S
Drainage systems Drainage systems Drainage systems
Downloaded by [George Mason University] at 20:18 24 December 2014
0 2 km 0 2 km 0 2 km
Fig. 8 Inundation areas for 24-h design rainfalls representing the 2, 5, 10, 25, 50 and 100-year return periods.
areas and depths were simulated to investigate poten- long-term changes and investigation of the effective-
tial hazards in the Tuku area. The simulated results ness of each proposed approach.
of average inundation depth and maximal inunda-
tion extent were determined by imposing the 24-h
design rainfall hyetograph for various return periods,
6.1 Approaches to flood management with
as shown in Fig. 8. The average inundation depth
public participation
increased from 1.22 m for a 2-year return period
event to 1.58 m for a 100-year return period event. The ground elevation in Tuku watershed is relatively
Similarly, the inundation area increased from 221 hm2 lower in the central and southwest parts (Fig. 1);
for a 2-year return period event to 1231 hm2 for a the central lowlands comprise the Yuku, Tandi, and
100-year return period event (Table 1). Jiaxing villages. The main channel quickly accumu-
The success of flood management requires pub- lates water from the lateral channels during storms.
lic involvement to facilitate the understanding and Over time, the water level in the main channel rises
incorporation of local community opinions into the to the extent that lateral channels are no longer con-
designs for new infrastructure. In recent years, public ducive to water drainage as a result of the backwater
hearings have become a crucial tool for effective com- effect. Consequently, surrounding areas are inundated
munication by decision makers and engineers. In this as the lateral channels overflow.
study, public hearings were held at Yuku, Tandi and The public required assurance from engineers
Jiaxing villages to encourage agreement among the that flood water would not overflow the channel banks
engineers, decision makers and the public for estab- and inundate their dwellings during the typhoon sea-
lishing the principles of flood management. Flood son. Their interest in the simulated rainfall inundation
damage losses were estimated and compared with results led to a deeper understanding of the engineer-
those of the present situation using various flood man- ing background of the project. They understood the
agement approaches. The impact of climate change purpose of the project and were reassured as to the
on the projected increase in rainfall was assessed with safety of their lives and property. In addition, sev-
inundation simulations, which allowed estimation of eral useful suggestions were provided by the villagers
1590 Hsiang-Kuan Chang et al.
for revising the design, such as establishing a flood By using local community commentary to ensure
warning system and enhancing the aesthetics of the that the villages are within the 10-year return-period
riverside. Through public participation in the consid- flood protection standard, the proposed structural
eration of safety as well as scenery, the following measures included the requirement to:
principles were negotiated and agreed to reduce the
water stage in the main channel, and to drain the lat- (i) enlarge the capacity of the drainage system;
eral channel water smoothly into the main channel (ii) construct retention ponds by using fishponds; and
during a flood event: (iii) construct pumping stations and flood diversion
works.
(a) The drainage system must be designed so that If inundation situations exceed the 10-year return-
the 10-year return-period event does not cause period protection standard, non-structural measures
overflow; i.e. the 10-year return-period flood such as flood forecasting and warning systems must
protection standard. be implemented for the local government to plan
(b) If the drainage system fails to convey water
Downloaded by [George Mason University] at 20:18 24 December 2014
– Upstream reach: enlarge the channel cross- simulated inundation extent and depth. With the
sectional area by raising levees 50 cm higher. improvement of drainage capacity, the total inun-
– Midstream reach: modify the box culvert by dated area was reduced from the original 745 ha
increasing the cross-sectional areas on both sides to 350 ha, a significant 53% reduction (Table 4).
of the river. The simulations indicate that inundation extent and
– Downstream reach: rebuild the left bank as a ver- average depth are less than those due to the 5-
tical retaining wall and dredge sediment deposits year return-period flood event without improvements
to convey more flood flow. (Table 1). Most of the inundation areas typically
appear along riversides or confluences in the drainage
Additionally, the 21 lateral channels were system.
widened and rebuilt to meet the 10-year return-
period flood protection standard. At the request of Approach 2: Modifying fishponds to reten-
the local county government in the public hearing, tion ponds The Tuku drainage system does not drain
the height of the levee cannot exceed the nearby smoothly and usually causes inundation because of
Downloaded by [George Mason University] at 20:18 24 December 2014
ground elevation by 1 m. Figure 10(a) shows the the relatively high water stage in the main channel
Table 4 Simulated inundation depth and area for the flood management approaches.
Approach Description Average inundation Maximum inundation
depth (m) area (hm2 )
- Without (before) improvement 1.40 745
1 Improvement of drainage systems 1.12 350
2 Construction of detention ponds + Approach 1 0.77 34
3 Construction of pumping stations and diversion 1.03 304
box culverts + Approach 1
4 Integrated flood management approach 0.76 18
(combines approaches 1, 2 and 3)
1592 Hsiang-Kuan Chang et al.
during a flood. The area has also experienced stress in the upper Wujiawei drainage channel can divert
due to land development in recent years. In the 2.5 m3 /s (17% of its flood water) into the main
drainage system, fishponds were built adjacent to channel. Similarly, the proposed diversion channels
the upstream and midstream reaches of the main in the upper Tiancuo and Tandi drainage channels
channels. They were considered to be grouped and can divert 9.5 m3 /s (57% of their flood water) into
treated as six retention ponds, as shown in Fig. 9, the Agoden River. According to the simulated results,
and designed based on the agreement negotiated with the pumping stations and flood diversion works
the owners in the meetings. Incorporating Approach decrease the inundation area by only 46 ha, indicat-
1 and modifying fishponds to retention ponds is ing that Approach 3 is not a cost-effective approach
referred to as Approach 2 (Table 4). When storms (Table 4).
occur, flood water flows into these ponds decreasing
the flood peak discharge, and lowering water levels in Approach 4: Integrated flood management
the main and lateral channels. When the water stage approach The retention pond approach outperformed
in the drainage system is reduced as the rainstorm other approaches, considerably reducing the inunda-
subsides, water from the retention ponds is released tion depth and area. By integrating all the approaches
Downloaded by [George Mason University] at 20:18 24 December 2014
by gravity into the main or lateral channels. Most as Approach 4 (the integrated approach), the maxi-
water in these retention ponds should be released mum inundation area and average inundation depths
before the wet season to accommodate excess precip- decreased from the initial 745 ha to 18 ha and
itation. However, a minimum water level is necessary from 1.40 m to 0.76 m, respectively (Table 4 and
to maintain the aquatic system for rearing fish in the Fig. 10(d)).
ponds. Water quality monitoring during flood reten-
tion operations is requested. The simulated results
demonstrated that improving the drainage capacity 6.2 Estimation of losses from flood damages
with six retention ponds substantially reduced the Detailed records of past and recent flood damages rel-
inundation area by 95.4% for the present condi- evant to this study area do not exist. However, data are
tions of the Tuku drainage system (Table 4 and Fig. available on:
10(b)). Importantly, modifying the fishponds to reten-
tion ponds satisfies the concept of returning “room for (a) inundation depth versus loss of crop produc-
river”. tion (Fig. 11(a)) applicable to all crop areas in
southern Taiwan, and
Approach 3: Construction of pumping sta- (b) inundation depth versus property loss measured
tions and flood diversions works Initially, Yuku as New Taiwan Dollars (NTD) per average fam-
(0.6 m3 /s), Tandi (6 m3 /s) and Wujiawei (20 m3 /s) ily (Fig. 11(b)) specific to the Tuku area (Water
were the only pumping stations in the Tuku drainage Resources Agency 2008).
area (Fig. 2(b)). The total capacity of the three sta-
Using these relationships and other published reports,
tions was far below that required to drain water
the direct and indirect flood damages were estimated.
to the main channel during storms, and resulted
in overflow from the lateral channels. Thus, it was
essential to increase the pumping capacity of these 6.2.1 Direct loss
stations to direct excess water from the lateral chan- 6.2.1.1 Loss of rice production The Tuku drainage
nels to the main channel efficiently. Therefore, dur- area is predominantly used to grow rice crops. Data
ing simulation, the discharge capacity of Yuku and on rice production per hectare were collected from
Wujiawei were augmented to 11 and 30 m3 /s, respec- the county government. Because rice production was
tively. A new pumping station of 9 m3 /s capac- fairly stable over the years, an average productivity
ity was also added to pump flood water into the rate of 5776 kg per hm2 and rice price of 21 NTD
main channel (Fig. 9). In addition, the Wujiawei per kg (Water Resources Agency 2008) were used
(A) and Tiancuo (B) flood diversion box culverts to calculate the production value of rice (approx.
were proposed, as shown in Fig. 9. Figure 10(c) 121 300 NTD/hm2 ). Loss of rice production for a
shows the simulated inundation results after construc- range of inundation depths was calculated using the
tion of the pumping stations and two flood diver- relationship shown in Fig. 11(a). Rice production loss
sion box culverts, combined with Approach 1. The (Lc) regarding NTD was calculated as Lc = [(Vp ×
results demonstrated that the flood diversion works Pr) + Ci] × A, where Vp is rice production value
Improvement of a drainage system for flood management with the effects of climate change 1593
50 0.9
45
(a) (b)
0.8
40 0.7
million NTD/family
35 0.6
30
0.5
25
0.4
20
0.3
15 Inundation time (day)
10 0.2
1~2
5 2~3 0.1
3+
0 0.0
0.0 0.5 1.0 1.5 2.0 0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0 4.5 5.0
Inundation depth (m) Inundation depth (m)
Fig. 11 Relationship of inundation depth with: (a) % loss of rice production, and (b) loss of property at household level in
the Tuku area. (Source: Water Resources Agency 2008)
Downloaded by [George Mason University] at 20:18 24 December 2014
per ha; Pr is loss of production (%); A is inundation surplus labour because of job losses, the expense of
area (ha); and Ci is the increased cost of repeated land dealing with surplus offal and refuges, and costly res-
cultivation (75 000 NTD/ha) because of inundation. cue operations and waiting times in the safe area for
flood victims. The projections of indirect losses are
6.2.1.2 Loss of fishponds Fishponds cover 8.5% of numerous, but amount to about 25% of the direct loss
the total watershed in the Tuku drainage system. (Water Resources Agency 2008).
Average fishpond loss per inundation event was cal- Inundation losses estimated assuming implemen-
culated at approx. 99 800 NTD (Water Resources tation of the proposed flood management approaches
Agency 2008). The area under inundation, quantity are compared and plotted in Fig. 12. Approach 2, con-
of fish lost per pond, machinery and repair costs were structing of retention ponds, is the most effective of
the main factors used to calculate fishpond losses. them. Integrating all the approaches can substantially
reduce individual property and public facility dam-
6.2.1.3 Loss of individual property Individual prop- ages, and reduce the loss to only 2.3% of the total loss
erty loss in the residential areas includes elec- (303.5 million NTD) in the present situation without
trical appliances, furniture and upholstery. Other improvements (Table 5).
losses include infrastructure, cars and motorcycles,
agricultural machinery, fertilizer and food grain.
Figure 11(b) was used to calculate these losses in 6.3 Assessment with climate change impact
NTD for a range of inundation depths. The calcu- Structural measures in the proposed approaches were
lation used a standardized assumption of an average planned according to the historical data for vari-
of 25 resident families for each hectare of inundated ous return-period floods. The precipitation scenarios
area. produced by the MPEH5 model in three SRES sce-
narios for the 2020s period (2010–2039) were used,
6.2.1.4 Loss of public facilities Public facilities losses as described in Section 4, to assess climatic change
include damage to railroads, highways, power trans- impact on the inundation simulations in the Tuku
mission, telecommunications, gas lines, water supply, area. The uncertainty of climate change can result
sewer and hydraulic facilities, government buildings, in various scenarios. In southern Taiwan, the results
schools and park areas. Losses pertaining to pub- of scenarios A2 and A1B from the MPEH5 model
lic facilities accounted for approx. 20% of individual decrease rainfall in specific months of the wet period
property losses (Water Resources Agency 2008). (May–October). Conversely, scenario B1 from the
MPEH5 model tends to increase rainfall during the
wet period (Table 6). Conservatively, the outcomes
6.2.2 Indirect loss
of scenario B1 from the MPEH5 model indicated a
Indirect loss includes various losses caused by traf- 1.215-fold increase of design rainfall intensity in July,
fic interruption, the declining purchasing power of within a 30-year period in the Tuku watershed.
people because of increased market prices, temporary To assess climate change impacts on the drainage
failure of government utilities to offer public services, system in the Tuku area, inundation simulations under
1594 Hsiang-Kuan Chang et al.
350
300
200
150
100
50 Present
Approach 1
Approach 2
0 s
er tie d
Approach 3
rop on ion es
al p Fis
hp uct liti e Approach 4
u pro
d aci los
di vid lic
f ect l lo
se
ice ir present : Without (before) imporvement
Downloaded by [George Mason University] at 20:18 24 December 2014
In R ub Ind ta
P To approach1 : Improvement of drainage systems
approach2 : Construction of detention ponds + approach 2
approach3 : Construction of pumping stations and
diversion box culverts + approach 2
approach4 : Combined flood management approach
Table 5 Loss (in million New Taiwan Dollars, NTD) due to inundation for various flood management approaches.
Losses Present Approach 1 Approach 2 Approach 3 Approach 4
Table 6 Ratio of rainfall (average of period/average baseline) of the MPEH5 model under climate change in the 2020s
period.
Scenario Jan. Feb. Mar. Apr. May Jun. Jul. Aug. Sep. Oct. Nov. Dec.
A1B 1.214 0.869 0.746 1.184 0.980 0.984 0.840 0.844 1.110 0.861 0.777 0.995
A2 0.959 0.843 0.555 1.129 0.911 0.817 1.001 1.091 0.85 0.981 0.836 0.921
B1 1.486 1.010 0.837 1.140 1.006 0.866 1.215 0.946 1.212 1.174 0.701 1.276
Table 7 Impact of climate change on inundation for various flood management approaches.
Approach Average inundation Maximum inundation Average inundation Maximum
depth (m) area (hm2 ) depth ratio inundation area ratio
the 10-year return-period protection standard were 11% greater inundation depth and 55% greater inun-
performed, and the same modelling inputs were used, dation extent relative to the present drainage sys-
except for 1.215 times the design rainfall intensity. tem (Table 7). The results simulated under cli-
The increase of rainfall intensity induced approx. mate change were close in inundation extent and
Improvement of a drainage system for flood management with the effects of climate change 1595
350
300
200
150
100
50
Present
0
p r o p erties ond
Approach 4
idual Fishp ction ties Approach 5*
Indiv produ blic facili e
Ri e
c ct los lose
Pu Indire Total approach1 : Without (before) improvement
approach4 : Combined flood management approach
* : Climate change
Downloaded by [George Mason University] at 20:18 24 December 2014
Fig. 13 Effects of climate change on inundation losses (Approach 5: Approach 4 + climate change).
average depth to the 50-year return-period flood event 0.85 m, the total property loss was marked, 8.93 times
(Table 1). higher than that without the climate-change effect
Of the proposed flood management approaches, (Fig. 13).
Approach 1, in which improving flow conveyance of
the drainage system reduces inundation extent and
average depth to those of the 5-year return-period 7 SUMMARY AND CONCLUSIONS
flood event (Table 1), satisfied the request of the
local communities. Compared with the present sit- A drainage–inundation model was used to simulate
uation (Table 4), with future projected precipitation flood inundation induced by typhoon and storm rain-
increase, the maximal inundation area may expand falls in the Tuku lowland area of southern Taiwan.
1.71-fold within 30 years (Table 7). If Approach The model combines a drainage flow model with a 2D
2 is implemented by adding the retention ponds, the overland-flow inundation model to evaluate damage
inundation extent and average depth reduce consid- losses for proposed flood management approaches.
erably relative to the present situation. However, the For effective communication, public participation was
maximal inundation area expansion by 8.62-fold is encouraged in the township to help both stakeholders
more than for Approach 2, as shown in Table 4. and engineers outline the principles of flood man-
This indicates that the designed retention ponds may agement. By adopting local community commentary
not attenuate the high flood discharges effectively, to ensure that the villages were under the 10-year
and require well-defined operation rules and rain- return-period protection standard against flood dam-
fall forecast modelling. Comparing Approach 3 with age, four flood management approaches with struc-
Approach 1 in Table 7 demonstrates the only slight tural measures were proposed, including increasing
effectiveness of building pumping stations and diver- the drainage capacity, using fishponds as retention
sion works. Integrating all the proposed approaches, ponds, and constructing pumping stations and build-
Approach 4 demonstrates that the inundation area ing flood diversion culverts in the Tuku drainage
could expand 13.11-fold compared to that without system. The fourth, integrated flood improvement
climate-change impact (Table 7). approach, a combination of all proposed approaches,
Based on the simulated results, such large exhibited superior performance with regard to miti-
increases in inundation area and depth are attributed gating inundation extent and depth. Of the proposed
mainly to the impacts of climate change. The under- approaches, Approach 2 using existing fishponds as
sized flood protection structures of lowland agricul- retention ponds is consistent with the concept of
tural production areas will be the first to experi- “room for river” to store more water. The simulated
ence inundation. The inundation simulation based on inundation area of Approach 2 was reduced by 95.4%
Approach 4 indicated that, although the average inun- of that created in the present conditions with a 10-year
dation depth increased marginally from 0.76 m to return-period protection standard.
1596 Hsiang-Kuan Chang et al.
Based on the selected GCM scenarios, precipi- HR Wallingford Ltd., 1997. HydroWorks user manual. Wallingford
tation scenarios projected by the MPEH5 model for UK: HR Wallingford Ltd.
Hsu, M.H., Chen, S.H., and Chang, T.J., 2000. Inundation simulation
the 2020s period (2010–2039) were used to assess for urban drainage basin with storm sewer system. Journal of
climate change impact on the inundation simula- Hydrology, 234, 21–37.
tions. Using a conservative future projected rain- Hsu, M.H., Lai J.S., and Yen, C.L., 1990. Two-dimensional inunda-
tion model for Taipei city. In: Proceedings of fifth international
fall increment, the outcomes of scenario B1 from conference on urban storm drainage 23–27 July 1990, Osaka,
the MPEH5 model were analysed, and indicated a 169–174.
possible increase of 1.215 times the design rain- Huber, W.C. and Dickinson, R.E., 1988. Storm Water Management
fall intensity in July within a 30-year period in the Model, user’s manual version IV . Washington, DC: US
Environmental Protection Agency.
Tuku watershed. Compared with simulated results Kundzewicz, Z.W., Pińskwar, I., and Brakenridge, G.R., 2013. Large
under the present drainage conditions, the climate- floods in Europe, 1985–2009. Hydrological Sciences Journal,
change increase of rainfall intensity induced a 55% 58 (1), 1–7.
Lai, J.S., et al., 2010a. A well-balanced upstream flux-splitting
greater inundation extent and 8.93 times higher finite-volume scheme for shallow-water flow simulations with
total property loss, inundation consequences that are irregular bed topography. International Journal for Numerical
close to those of the 50-year return-period flood Methods in Fluids, 62, 927–944.
Downloaded by [George Mason University] at 20:18 24 December 2014
Vlotman, W.F., Wong, T., and Schultz, E., 2007. Integration of Taiwan: Water Resources Agency, Ministry of Economic
drainage, water quality and flood management in rural, Affairs.
urban and lowland areas. Irrigation and Drainage, 56, S1, Wrachien, D. de., Mambretti, S., and Schultz, B., 2011. Flood man-
S161–S177. agement and risk assessment in flood-prone areas: measures and
Vongvisessomjao, S., Tingsanchali, T., and Chaiwat, C., 1985. solutions. Irrigation and Drainage, 60, 229–240.
Bangkok flood plain model. In: 21st IAHR Congress, August, Yu, D. and Lane, S.N., 2006. Urban fluvial flood model-
Melbourne. ing using a two-dimensional diffusion-wave treatment, Part
Water Resources Agency, 2008. The planning of regula- 1: mesh resolution effects. Hydrological Processes, 20,
tion and environment rehabilitation of Tuku drainage. 1541–1565.
Downloaded by [George Mason University] at 20:18 24 December 2014