Multi Spectral
Multi Spectral
Multi Spectral
net/publication/315797574
Multispectral satellite imagery and airborne laser scanning techniques for the
detection of archaeological vegetation marks
CITATIONS READS
9 13,863
2 authors:
Some of the authors of this publication are also working on these related projects:
Restauration strategies of the deteriorated peatland ecosystems from Romania (PeatRO) View project
All content following this page was uploaded by Anamaria Roman on 01 April 2018.
Editor
CORIOLAN HORAȚIU OPREANU
ROMANIAN ACADEMY – INSTITUTE OF ARCHAEOLOGY AND HISTORY OF ART CLUJ-NAPOCA
LANDSCAPE ARCHAEOLOGY
ON THE NORTHERN FRONTIER
OF THE ROMAN EMPIRE
AT POROLISSUM
An interdisciplinary research project
Edited by
Coriolan Horațiu Opreanu and Vlad-Andrei Lăzărescu
ISBN 978-606-543-787-6
© Editors, 2016
List of Illustrations 7
List of Tables 13
Authors and Contributors 15
Acknowledgements 17
Introduction (Coriolan Horațiu Opreanu, Vlad-Andrei Lăzărescu) 19
I
DEFINING THE ROMAN LIMES OF DACIA
II
INTERDISCIPLINARY RESEARCH METHODS AND THE STUDY OF THE
ROMAN LIMES IN DACIA POROLISSENSIS
Dan Ștefan
Geophysical surveys and the reconstruction of the ancient landscape 115
Sorina Fărcaș, Ioan Tanțău, Roxana Grindean, Andrei Marian Panait & Andrei Cosmin Diaconu
Palaeoenvironmental research methods 121
Ioan Tanțău, Roxana Grindean, Andrei Marian Panait, Andrei Cosmin Diaconu & Sorina Fărcaș
Palaeoenvironmental reconstructions since 5000 BC 153
Anamaria Roman, Tudor-Mihai Ursu, Vlad-Andrei Lăzărescu & Coriolan Horațiu Opreanu
Multi-sensor surveys for the interdisciplinary landscape analysis and archaeological feature
detection at Porolissum 237
Bibliography 309
Authors and Contributors
Sorina Fărcaș PhD is Senior Researcher and Director of the Institute of Biological Research
Cluj-Napoca (IBRC), branch of the NIRDBS. Her scientific areas of competence include
Quaternary palynology, past and present vegetation ecology and conservation.
Coriolan Horațiu Opreanu PhD is Deputy Director of the Institute of Archaeology and
History of Art and Associate Proffesor at ‘Babeș-Bolyai University’ Cluj-Napoca, Director
of the excavations at Porolissum, Director of the project “Seeing the Unseen. Landscape
Archaeology on the Northern Frontier of the Roman Empire at Porolissum” and Editor of
the archaeological yearbook “Ephemeris Napocensis”. His main research interest is focused
on archaeology of the Roman frontiers, Roman urban archaeology, Late Roman and Early
mediaeval archaeology, relations between the Roman and barbarian worlds, Roman mili-
tary history.
Anamaria Roman PhD is Researcher at IBRC, branch of the NIRDBS, and member in the
Executive Council of the Romanian Plant Phytosociological Society (SFR). Her research
interests focus on vegetation, landscape ecology and technologies such as Remote Sensing,
enabling ecosystem assessment in the context of past, present or future environmental
changes due to human activities.
Ioan Rus PhD is Assistant Professor and researcher at Babeș-Bolyai University of Cluj-
Napoca, Faculty of Geography with the scientific areas of competence in Geology,
Photogrammetry and Remote Sensing.
Dan Ștefan PhD is Senior Researcher at the National Museum of Eastern Carpathians
Sfântu Gheorghe with the scientific areas of competence in near surface Topography,
Geophysics and Remote Sensing
Ioan Tanțău PhD is Assistant Professor and researcher at Babeș-Bolyai University of Cluj-
Napoca, Department of Geology, with the scientific areas of competence in Geology and
Quaternary palaeoenvironments.
Tudor-Mihai Ursu PhD is Scientific Researcher at IBRC, branch of the NIRDBS, and mem-
ber of the SFR. His research interests include landscape ecology, plant taxonomy, studies of
past and present vegetation and Remote Sensing.
T his volume summarizes the research supported between 2012–2016 by the proj-
ect ‘Seeing the Unseen. Landscape Archaeology on the Northern Frontier of the
Roman Empire at Porolissum’ of the Romanian National Authority for Scientific Research,
CNCS-UEFISCDI (project number PN-II-PT-PCCA-2011-3.1-0924). As director of the
project I am grateful to the colleagues who were members of the interdisciplinary research
group for their professional approach, but also for their team spirit (Sorina Fărcaș, Ioan
Fodorean, Roxana Grindean, Vlad-A. Lăzărescu, Dănuț Petrea, Anamaria Roman, Ioan Rus,
Ioan Tanțău, Tudor Ursu, Iuliu Vescan). I appreciate at higher level the scientific accomplish-
ments of the young researcher colleagues and I consider their performance as a key-factor
of making a go of the project. Many thanks go also to our collaborators Andrei Diaconu,
Monica Gui, Andrei Panait, Raularian Rusu, Dan Ștefan and Cristian Ștefan.
Coriolan H. Opreanu
Multispectral satellite imagery and airborne
laser scanning techniques for the detection
of archaeological vegetation marks
1. Introduction
424
WESTHOFF/VAN DER MAAREL 1978.
425
WILSON 2000.
426
BECK ET ALII 2007.
427
EVANS/JONES 1997; HEJCMAN/SMRZ 2010.
428
BENNETT 2011.
Multispectral satellite imagery and airborne laser scanning techniques for the detection 143
including vegetation, absorb and reflect light of different wavelengths depending on their
physical, physiological or chemical properties429. Understanding the spectral language of
plant reflectance (spectral signature) in the context of the electromagnetic spectrum is very
important in environmental studies analyzing, for instance, vegetation stress responses or
land cover change, being also employed in many studies for the comprehensive analysis of
archaeological landscapes430.
Fig. 88. The electromagnetic spectrum, reproduced (after TEMPFLI ET ALII 2009 and BERGMAN ET ALII 2011).
Healthy leaves absorb 70–90% of incident visible radiation, particularly in the blue and red
wavelengths (centered on 450 nm and 670 nm respectively), and reflect most of the green
light (centered on 533 nm) which is why leaves appear green to the human eye431. Green light
is reflected back by the chlorophyll pigment within the chloroplasts in the palisade cells.
Blue and red light are absorbed and used in photosynthesis by the chlorophyll pigment in the
palisade cells. However, wavelengths in the NIR region are mostly reflected and transmitted
through the leaves432, being scattered by the cell interfaces in the mesophyll tissue (Fig. 89).
Fig. 89. Cell structure of a green leaf and interactions with the electromagnetic radiation,
adapted from SUMMY ET ALII 2003 and VERHOEVEN 2011. The cross section trough the leaf
of Ligustrum vulgare is provided by the courtesy of Dorina Podar and Zoltan Balazs.
In healthy leaves 40–60% of the NIR light is reflected, although at canopy level the situ-
ation is more complex, because a range of effects such as additive reflectance, incidence
angle, leaf orientation, shadow and soil background reflectance. A typical spectral reflec-
tance curve for healthy vs. stressed vegetation is given in Fig. 90. Plant stress considerably
changes the spectral properties of vegetation, as the chlorophyll pigment rapidly decays
and loses its absorption properties433, increasing reflectance in the green to reddish part
of the visible region (from around 530 nm to 670 nm) of the electromagnetic spectrum.
Moreover, stress often results in leaf chlorosis: a yellowing discoloration due to chlorophyll
losing dominance over the carotenoids434. Stressed plants reveal, therefore, a different spec-
tral signature which can be observed both in visible light (VIS) (they become yellowish), and
as a generally lower reflectance in the NIR region of the electromagnetic spectrum (Fig. 91).
While images captured in the visible spectrum can be visually interpreted, the ones involv-
ing bands from the invisible wavelengths or their combinations with visible light require a
mathematical expression and intervals to make them interpretable. Vegetation indices are
mainly derived from reflectance data within the Red and NIR bands. They are quantitative mea-
sures, based on vegetation spectral properties that attempt to measure biomass or vegetative
431
KNIPLING 1970.
432
SLATON/HUNT/SMITH 2001.
433
KNIPLING 1970; CARTER 1993.
434
HENDRY/HOUGHTON/BROWN 1987; STRZALKA/KOSTECKA-GUGALA/LATOWSKI 2003.
Multispectral satellite imagery and airborne laser scanning techniques for the detection 145
Fig. 91. Reflectance curves for healthy vs. stressed vegetation (left) and the stressful
effect of positive subsurface archaeological structures on vegetation (right).
vigor435. They operate by contrasting intense chlorophyll pigment absorption in the red against
the high reflectance of leaf mesophyll in the near infrared. The simplest form of vegetation
index is the ratio between two digital values from the red and near infrared spectral bands.
Theoretical analyses and field studies have shown that vegetation indices are directly related
to light-dependent physiological processes (e.g. photosynthesis) of photosynthetically active
plants, predominant in the upper canopy436. Their use is based on the principle that combin-
ing spectral bands can highlight specific characteristics of vegetation regarding growth and
health that are otherwise not visible from individual bands or true/false colour RGB images.
Vegetation indices aid the identification of contrasts in plant quality, vigor and stress, all of
which could be related to the presence of upstanding or buried archaeological features. More
than 150 such indices are presented in the environmental remote sensing literature. The
AGAPIOU/LYSANDROU 2015.
435
most widely used index for archaeological analysis is the well-known Normalized Difference
Vegetation Index (NDVI437) obtained by using the following formula:
Vegetation indices are usually employed in order to monitor seasonal or even long-term
variations of structural, phenological and biophysical parameters of land surface vegeta-
tion cover438. Additionally, on the basis of remotely sensed data, crop marks may be suitably
identified by exploiting vegetation indices that are spectral combinations of different bands.
Although the number of published indices is high, apart from Traviglia439, that compared a
simple Red/NIR ratio, the NDVI and Modified Soil-Adjusted Vegetation Index (MSAVIS2),
there are very few studies that tested vegetation indices systematically as tools for vegeta-
tion mark detection. The vegetation indices used for any archaeological study need to be
selected carefully so that they have a substantial biophysical (as opposed to purely numeri-
cal) basis. This is crucial to the aim of understanding the physical and biological parameters
that influence the representation of archaeological features in the data. Further comparison
tests of various indices are needed to establish their individual effectiveness and their use-
fulness for landscape and land cover analysis.
detection, however, is more recent449. In the last decade, the interpretation of ALS-derived
DTMs (Airborne Laser Scanning-derived Digital Terrain Models) proved to be a very useful
tool for the extraction of anthropogenic features belonging to different historic environ-
ments450. Although there are several papers that deal with this subject, most of them are
focused primarily on Mediterranean landscapes451, on alluviated landscapes452, and on grass-
land environments453, with just a few examples of steep terrain forests. Nevertheless, even in
these cases at least some archaeological features are above the soil surface, being obscured
only by the forest canopy454.
The fundamental concept of a LiDAR measurement is to send a laser pulse towards
a target and to measure the timing and amount of energy that is scattered back from the
target (Fig. 92). The return signal timing (t) provides measurement of the distance between
449
DONEUS ET ALII 2008; BENNETT ET ALII 2012; LASAPONARA/COLUZZI/MASINI 2011; STEWARTA/
LASAPONARA/SCHIAVONA 2014.
450
CHALLIS ET ALII 2008; DONEUS 2013; BENNETT/COWLEY/DE LAET 2014; CARLSON/BAICHTAL
2015; HERRMANN 2016.
451
POIRIER ET ALII 2013.
452
CHALLIS/FORLIN/KINCEY 2011.
453
BENNETT ET ALII 2012; BENNETT ET ALII 2013.
454
DONEUS/BRIESE 2006; DONEUS ET ALII 2008; ŠTULAR 2011.
148 Anamaria Roman & Tudor-Mihai Ursu
the instrument and the scattering object (d), where c is the speed of light (1.07925285
x109 km/h).
t = 2 d/c
If the top of the vegetation canopy is measured (t1) along with a measurement of the
local ground height (t2), then the height of the canopy (h) can be calculated:
h = d1- d2= (c/2) t1- t2
The airborne sensor itself comprises the laser and projection mechanism, a Global
Positioning System (GPS) and an Internal Measurement Unit (IMU) along with a control
system that also records the data. The GPS provides location data throughout the survey,
while the IMU records the pitch, yaw and roll of the aircraft during survey. Together, these
measurements are used to process the ALS data and to enable correction for the movement
of the aircraft during survey and alignment to real-world co-ordinates. Although airborne
systems were known to be able to record height to less than 1 m accuracy in the 1970s, the
more recent advancements in GPS and IMU technology enabled the sensor to be used for
topographic mapping455.
ALS data are generally measured by two factors: point density (average number of points
per square meter) and point distance (average separation of points). As part of the process-
ing, error images can be generated to highlight areas where the data are inconsistent, such
as at the overlap of flight lines. Processing and analyzing LiDAR data involves the applica-
tion of different filtering, interpolation and visualization algorithms in order to generate
very high resolution terrain models such as: Digital Terrain Model (DTM), Digital Surface
Model (DSM), Slope Model, Canopy Height Model (CHM), Hill-Shade Models etc.
enable the investigations of more wavebands and wavelengths from the electromagnetic
spectrum. Nevertheless, these aerial unmanned platforms cannot cover large investigation
areas like the satellite multi-sensors. In principle, it is clear that remote sensing techniques,
such as airborne laser scanning (ALS) and airborne digital spectral imaging, can signifi-
cantly enhance our understanding of archaeological features within a landscape. ALS allows
greater and more precise measurement of the topography of the ground surface than any
other technology at landscape scale, while digital spectral imagery captures the nature of
vegetation and soil changes not just in the visible wavelengths but also in the near and short-
wave infrared (NIR and SWIR) and thermal regions of the spectrum458.
The multi-sensor approach of archaeological surveys has the advantage of being com-
plex and variable, thus overcoming the inherent weaknesses of every remote sensing tech-
nique used by itself. Archaeological features are very variable in what regards their topology,
topography and structure. Sometimes they can be directly visible at the soil surface or they
can appear only as proxy changes to soil and vegetation caused by sub or near surface fea-
tures. In some cases, they are completely masked by soil, vegetation or other environmental
conditions. In most cases they have been altered to a certain degree by the environmental
factors. It is clear that no single sensor could detect such a wide range of characteristics.
Thus, the advantage of multi-sensor surveys lies in the complementarity of the data that
can be collected by multiple sensors, allowing the detection of multiple characteristics of
the archaeological features. This complementarity not only improves the detection rate but
also promotes the understanding and interpretation of the detected features and their sur-
roundings. The availability of more complex information allows, in the end, for more cor-
rect interpretations and better decisions.
The increased spectral resolution of satellite images and spatial resolution of laser scan-
ning systems, as well as the improved analytical methods available led us to believe that such
terrain images, if generated and analyzed properly, may be a valuable data source to assist
archaeological surveys in the open-grassland or the forested area from Porolissum. The joint
usage of both high-resolution multispectral and airborne laser scanning for archaeological
purposes has been, to our knowledge, only rarely approached before in this environmental
context at European level, and never at national level.
In Romania, the remote sensing techniques are still underused in archaeology: aerial
photography was first used in the 1930s, and although some studies on this subject were
published during the 1970s, the technique is still primarily used only as a tool for study
site illustration. The lack of a broader understanding of the potential of this remote sensing
technique prevented its large scale use for discovering and monitoring archaeological sites
in Romania. A notable exception is the work of I. Oltean and W. Hanson from 2013459, who
have been studying, since 1998, the history and development of the western lowlands of
Romania by the application of archaeological aerial reconnaissance. In what regards ALS
technology, it had only been minimally used for archaeology, mostly in the area of public
awareness at the UNESCO site Sarmizegetusa Regia, as commissioned by the BBC.
In the case of the Roman archaeological site from Porolissum, the first stage in approach-
ing the spectral properties of the vegetation cover was to identify the plant communities
growing above the presumed subsurface archaeological remains. This step was necessary
since in our research area spectral properties are variable among different archaeological fea-
tures/sites but also within the same site. Therefore, we have relied mainly on the geometrical
OLTEAN/HANSON 2013.
459
150 Anamaria Roman & Tudor-Mihai Ursu
properties of the vegetation cover (particularly tree height) which were extracted from high
spatial resolution airborne laser scanning data (LiDAR data). Also, we tested the suitability of
WorldView-2 (WV2) multispectral satellite imagery for archaeological reconnaissance and
interpretation. The WV2 instrument is a “push broom” imager launched in October 2009
by DigitalGlobe and was the most technologically-advanced high-resolution satellite ever
launched, before its twin, WorldView-3, reached orbit in 2014. The WV2 satellite constructs
an image one row at a time as the image of the Earth, focused through the telescope, moves
across the linear detector arrays, which are located on the focal plane. The WV2 satellite
carries two types of sensors that collect, respectively: panchromatic images (0.450-0.800
μm) with cell size 0.46 m at nadir (0.50 m for commercial uses) and multispectral images
(Coastal Blue: 0.400–0.500 μm; Blue: 0.450–0.510 μm; Green 0.510-0.580 μm; Yellow
0.585–0.625 μm: Red 0.630–0.690 μm; Red Edge 0.705–0.7 45 μm; NIR1 0.770–0.895 μm;
NIR2 0.860–1.040 μm) with cell size 1.84 m at nadir (2 m for commercial uses). Operating
at an altitude of 770 km with an inclination of 97.2°, this sun synchronous satellite has an
average revisit time of 1.1 days and is capable of collecting up to 1 million km2 of 8-band
imagery per day. Commercially available products are resampled to 0.5 m (Panchromatic)
and 2.0 m (Multispectral). The nominal swath width is 16.4 km460.