Natural Hazards 33: 283–301, 2004.

283
© 2004 Kluwer Academic Publishers. Printed in the Netherlands.

Application of Remote Sensing in Flood

Management with Special Reference to Monsoon
Asia: A Review

JOY SANYAL
and X. X. LU

Department of Geography, Faculty of Arts and Social Science, National University of Singapore, 1
Arts Link, Singapore 117570

(Received: 20 June 2003; accepted: 4 November 2003)

Abstract. The conventional means to record hydrological parameters of a flood often fail to record
an extreme event. Remote sensing technology along with geographic information system (GIS) has
become the key tool for flood monitoring in recent years. Development in this field has evolved
from optical to radar remote sensing, which has provided all weather capability compared to the
optical sensors for the purpose of flood mapping. The central focus in this field revolves around
delineation of flood zones and preparation of flood hazard maps for the vulnerable areas. In this
exercise flood depth is considered crucial for flood hazard mapping and a digital elevation model
(DEM) is considered to be the most effective means to estimate flood depth from remotely sensed or
hydrological data. In a flat terrain accuracy of flood estimation depends primarily on the resolution
of the DEM. River flooding in the developing countries of monsoon Asia is very acute because of
their heavy dependence on agriculture but any flood estimation or hazard mapping attempt in this
region is handicapped by poor availability of high resolution DEMs. This paper presents a review of
application of remote sensing and GIS in flood management with particular focus on the developing
countries of Asia.

Key words: flood, remote sensing, geographic information system, digital elevation model, hazard
mapping, monsoon Asia

1. Introduction
Among all kinds of natural hazards of the world flood is probably most devastating,
wide spread and frequent. In the humid tropics and subtropical climates, especially
in the realms of monsoon, river flooding is a recurrent natural phenomenon. Floods
resulting from excessive rainfall within a short duration of time and consequent
high river discharge damage crops and infrastructures. They also result in siltation
of the reservoirs and hence limit the capacity of existing dams to control floods.
For formulating any flood management strategy the first step is to identify the
area most vulnerable to flooding. This step is even more critical for the developing
countries in the monsoon Asia as the funding available for developmental activities
is very limited. Thus these funds need to be utilized optimally for the areas that

Author for correspondence. E-mail: g0202381@nus.edu.sg
284 JOY SANYAL AND X. X. LU

suffer from river flooding most frequently. With the equipment currently installed
at river gauging stations it is sometimes difficult to record an extreme flood event
having a very high return period. In the developing counties, the density of gauging
stations is very low and thus any flood prediction or risk assessment model tested
in the developed countries faces acute shortage of ground data when applied. Re-
mote sensing is a reliable way of providing synoptic coverage over a wide area
in a very cost effective manner. It also overcomes the limitation of the ground
stations to register data in an extreme hydrological event. In addition multi-date
imageries equip the investigators with an additional tool of monitoring the change
or reconstruct progress of a past flood.
For the last two decades advancement in the field of remote sensing and geo-
graphic information system (GIS) have greatly facilitated the operation of flood
mapping and flood risk assessment. It is evident that GIS has a great role to play in
natural hazard management because natural hazards are multi dimensional and the
spatial component is inherent (Coppock, 1995). The main advantage of using GIS
for flood management is that it not only generates a visualization of flooding but
also creates potential to further analyze this product to estimate probable damage
due to flood (Hausmann et al., 1998; Clark, 1998).
Smith (1997) reviews the application of remote sensing for detecting river in-
undation, stage and discharge. Since then, the focus in this direction is shifting
from flood boundary delineation to risk and damage assessment. Therefore, there
is a need to review the current literature with a holistic view of dealing with vari-
ous prospects and constraints of using the technology of remote sensing and GIS
in flood management. This paper presents recent developments on delineation of
flooded areas and flood hazard mapping using remote sensing and GIS. In particu-
lar this paper draws attention to some of the issues associated with application of
remote sensing in combating floods in extremely flat flood plains of monsoon Asia.
Our review includes three aspects. First, we focus on the development of remote
sensing as a tool of flood delineation. Second, we emphasize the assessment of the
intensity of flood hazards and damage. Third, we highlight some of the issues in
the application of the technology with particular reference in the monsoon Asia, an
agricultural area with lack of high resolution spatial data bases.

2. Remote sensing as a tool of flooded area delineation

2.1. APPLICATION OF OPTICAL REMOTE SENSING
In the initial stages of satellite remote sensing the data available was from Landsat
Multi Spectral Scanner (MSS) with 80 m resolution. The pioneering investigations
in the field of application of remote sensing in flood mitigation were predominantly
concentrated on the flood prone regions of USA. MSS data were used to deal with
the flood affected areas in Iowa (Hallberg et al., 1973; Rango et al.,1974), Arizona
(Morrison et al., 1973), and Mississippi River basin (Deutsch et al., 1973; Deutsch
et al., 1974; Rango et al., 1974; McGinnis et al., 1975; Morrison et al., 1976). MSS
APPLICATION OF REMOTE SENSING IN FLOOD MANAGEMENT 285

band 7 (0.8–1.1 µm) has been found particularly suitable for distinguishing water
or moist soil from dry surface due to strong absorption of water in the near infrared
range of the spectrum (Smith, 1997).
From the early 1980s, Landsat Thematic Mapper (TM) imageries with 30 m
resolution became the prime source of data for monitoring floods and delineating
the boundary of inundation. Special attention was given to dealing with monsoon
flooding in the developing countries like West Africa (Berg et al., 1983), India
(Bhavsar, 1983) and Thailand (Ruangsiri et al., 1984). Landsat TM band 4 proves
to be very useful in discriminating water from the dry land surface because it is a
near equivalent of MSS band 7. But Landsat TM NIR band cannot be used optim-
ally in developed land use areas such as downtown commercial or industrial areas.
The main reason is that the NIR band reflects very little energy from asphalt areas,
appearing black in the imageries. Therefore makes it easy to confuse developed
areas with water. Wang et al. (2002) successfully solved this problem by adding
Landsat TM band 7 to the NIR (band 4) band to delineate the inundated areas.
In TM band 7 (2.08–2.35 micrometre) images the reflectance from water, paved
road surfaces, and roof tops differ significantly and therefore in the Band4+Band7
image, it becomes easier to choose the density slice for extracting the flood water.
But in some cases a simple density slice or supervised classification is not enough
to identify the inundated area accurately.
During later stages SPOT multi spectral imageries, were also used for flood
delineation with the similar assumption that water has very low reflectance in
the near infrared portion of the spectra. SPOT imageries, for example, were used
along with a DEM for delineation of monsoon flood in Bangladesh (Brouder, 1994;
Oberstadler et al., 1997; Profeiti et al., 1997; Sado et al., 1997).
Apart from these medium resolution imageries, coarse resolution imageries like
Advanced Very High Resolution Radiometer Radiometer (AVHRR) data have been
also found useful for floods of a regional dimension ( Wiesnet et al., 1974; Huh et
al., 1985a–c; Ali et al., 1987; Islam et al., 2000a–c, 2001, 2002). Although AVHRR
imageries are coarse in resolution and frequently contaminated by cloud cover their
merit lies in their high temporal resolution. This advantage enables us to monitor
the progress of a flood in near real-time.
To use the capability of the near infra red band more effectively to detect water
Normalized Difference Vegetation Index (NDVI) can be used to monitor river in-
undation from AVHRR images. It is well known that water has a unique spectral
signature in the near infrared which is very different from other surface features.
Therefore, when a surface feature is inundated its NDVI value changes consid-
erably from the normal situation. Wang et al. (2002) observed that in the lower
reaches of the Yangtze River, the NDVI value for inundated surface features re-
mains negative while the value for non inundated surface is commonly greater than
0. But choice of this threshold is critical because natural condition of river flooding
varies greatly from place to place. The main difficulties of selecting an appropriate
threshold arise from two facts. Firstly the albedo of water bodies increases signi-
286 JOY SANYAL AND X. X. LU

ficantly due to high concentration of sediment in the flooded water and secondly,
albedo of bare soil decreases considerably due to its high moisture content during
the monsoon season. These two factors collectively reduce the difference in NDVI
value between inundated and dry surface. In some studies, NDVI values of flood
water were found to be significantly positive (Barton et al., 1989) Thus, a straight
forward approach of using simple NDVI values might not be universally effective
in delineation of inundated area. Moreover, many other factors such as atmospheric
condition, cloud cover and satellite viewing angle also influence NDVI values and
attempts should be made to minimize these effects before calculating the NDVI.

2.2. APPLICATION OF MICROWAVE REMOTE SENSING

The existence of cloud cover appears as the single most important impediment to
capture the progress of floods in bad weather condition (Rango et al., 1977; Lowry
et al., 1981; Imhoff et al., 1987; Rashid et al., 1993; Melack et al., 1994). The
development of microwave remote sensing, particularly radar imageries, solve the
problem because the radar pulse can penetrate cloud cover. Currently the most
common approach to flood management is to use synthetic aperture radar (SAR)
imagery and optical remote sensing imagery simultaneously in one project (Honda
et al., 1997; Liu et al., 1999; Chen et al., 1999). Apart from its all weather capabil-
ity the most important advantage of using SAR imagery lies in its ability to sharply
distinguish between land and water.
Thresholding is one of the most frequently used techniques in active remote
sensing to segregate flooded areas from non flooded areas in a radar image (Liu
et al., 1999; Townsend et al., 1998; Brivio et al., 2002). Commonly, a threshold
value of radar back scatter is set in decibel (dB) and a binary algorithm is followed
to determine whether a given raster cell is ‘flooded’ or not. Radar back scatter is
computed as a function of the incidence angle of the sensor and digital number
(DN) (Chen et al., 1999). The threshold values are determined by a number of
processes depending on the study area and overall spectral signature of the imagery.
Change detection can be used as a powerful tool to detect flooded area in SAR
imagery. It is generally performed by acquiring two imageries taken before and
after the flood. Coherence and amplitude change detection techniques are widely
applied in SAR domain. In the amplitude approach, areas are estimated as flooded
where the radar back scatter is observed to be in considerable decline from be-
fore flood to after flood imagery. In the coherence approach areas are generally
identified as flooded where the coherence or correlation of radar backscatters from
before and after flood imagery are very low (Nico et al., 2000). Multi-date SAR
scenes for the same area can be projected to red, green and blue channels to create
a colour composite. Long et al. (2001) used three ERS SAR scenes to produce
this kind of composite image. The composite image effectively depicts progress of
a flood during a specific time period. This methodology is simple to execute and
APPLICATION OF REMOTE SENSING IN FLOOD MANAGEMENT 287

provides an opportunity to readily identify the area that remains water logged for a
maximum period of time.
The existing studies pointed out some common problems encountered in accur-
ately extracting the flood affected area from SAR imageries. A major problem is
associated with the relation between radar wave length and roughness of the terrain
and water body. Normally pure and calm water acts as a specular reflector to the
radar signals. Thus the radar antennae receive no backscatter and the water appears
in dark tone in the SAR imageries. Rough water surface appears in brighter tone
in the SAR imageries than the calm water (Yang et al., 1999). During floods, bad
and windy condition usually prevails over the affected area. Wind induced ripples
in the water surface frequently creates problems for the interpreter to determine the
threshold value to delineate the flooded area.
Forest cover also poses an obstacle to accurately identify inundated areas from a
SAR image (Hess et al., 1990). The key to identify the inundated areas under forest
cover lies in the fact that flooded forests produces a bright radar back scatter in
contrast to non flooded forests due to a double bounce effect (Kundus et al., 2001),
whereas the flooded areas without a forest canopy appear dark in SAR imageries.
The flooded forests reflect 2dB as radar back scatter in the L-band (Pope et al.,
1997). This particular property has been utilized to separate the inundated areas
from the non inundated ones by applying the thresholding technique (Townsend
et al., 1998; Ali et al., 2001; Rosenqvist et al., 2002). A correct separation of
the flooded and non-flooded settlements is also problematic. Normally the high
back scatter of the buildings overlay the back scatter of flood water within the
settlements. Rural settlements, especially in monsoon Asia are often surrounded
by trees. Therefore, due to the effect of trees, inundation within the settlements is
very difficult to detect (Oberstadler et al., 1997).
Establishing a universal threshold value for detecting flood is not justified. A
particular algorithm of deriving the threshold for differentiating inundated areas
from the dry land may not work with the same efficiency in varied natural settings.
Jin (1999) used DMPS SSM/I (Specific Sensor Microwave/Imager) data for this
purpose and came to the conclusion that a particular algorithm developed on the
basis of a wet land environment produces spurious results for inundated areas un-
der a forest canopy. He emphasized the importance of knowledge in the regional
geography of the area under investigation in setting any threshold value.
Combination of wave length, incidence angle and polarization play a major role
in influencing the interpreter’s ability to segregate flood areas from the non-flooded
ones under a forest canopy. Given the same wave length and incidence angle the
ratio of back scatter from flooded forests to the non-flooded ones is higher in ho-
rizontal polarization than at vertical polarization. It can be further demonstrated
that while wave length and polarization parameters are held constant, the aforesaid
ratio is larger under a small incidence angle than large (Wang et al., 1995). Radar
back scatter also depends on the orientation of the rough surface (e.g., furrows
of a ploughed field). As radar signals are directional in nature, the same surface
288 JOY SANYAL AND X. X. LU

may produce different tonal signatures depending on the relative orientation of the
rough surface to the radar antenna.
Radar incidence angle and the consequent variation in back scatter also pose
difficulties in the delineation of inundated areas from SAR imageries. Generally,
it is found that dealing with forested area imageries taken at a lower incidence
angle prove more useful compared to imageries taken at a higher incidence angle.
Townsend et al. (1998) recommended the use of JERS SAR imageries to ERS for
identification of flooded areas under forest canopy due to the smaller incidence
angle of the former as compared to the latter. This satellite has some advantages
over other earth observation satellite systems to detect flooding under forest as the
L-Band signal of JERS1 is particularly sensitive to standing water below the forest
canopy (Rosenqvist et al., 2002). Radarsat SAR imageries is also preferable to ERS
for flood mapping as the former can rotate its sensor disseminating the radar signal
at different incidence angles. This manoeuvring capability proves very useful in
locating the flood affected area in different types of terrain and land cover (Andre
et al., 2002).

2.3. A COMBINED APPROACH

In recent years, flood mapping efforts synthesize the advantages of both optical
and microwave remote sensing technologies for better results. In some occasions
this approach also leads to the formulation of better flood management strategy. In
mountains, slopes positioned perpendicular to the radar beam only appear bright
and all other areas appear as dark or shaded. This poses an obstacle to effectively
identify the flooded areas in the mountains. Due to its shaded appearance it is very
common to erroneously identify the mountainous areas as inundated. Yang et al.
(1999) came out with a solution to this problem by fusing Landsat TM imageries
with SAR imageries. After extracting the mountain regions from a Landsat TM
data it was overlaid on the SAR imagery. To get rid of the spuriously demarc-
ated ‘flooded’ area in the SAR imagery, the portion identified as mountain in the
TM image was eliminated from it. This process led to accurate delineation of the
inundated area. However, the authors realized that adoption of this method may
result in removal of some existing water bodies from the actual hydrological layer
of the mountainous area. To rectify this error, the TM data was consulted and the
water bodies in the mountain region have been restored in the final output. This
paper illustrates that the method of using data captured by different sensors is not
straight forward and often requires iterative experiments to arrive at an optimal
result.
In this context, the utility of generating land cover maps is realized as it provides
information about permanent water bodies in normal hydrological conditions and
rescues the interpreter from possible confusion of including permanent water bod-
ies as inundated area (Tholey et al., 1997). But while dealing with the monsoon
flood one should consult the imagery of a wet season to carve out the natural
APPLICATION OF REMOTE SENSING IN FLOOD MANAGEMENT 289

drainage. Use of an imagery captured in dry season would lead to underestimation

of the natural drainage, which in turn would lead to an overestimation of the flooded
area (Islam et al., 2002).
The efforts towards delineation of inundated area are ultimately aimed at assess-
ing the impact of flood on our economy and livelihood. Land use maps obtained
primarily from optical remote sensing are overlaid on the flood maps for assessing
the degree of damage on different kind of land uses. Varieties of land use clas-
sification strategies have been adopted by researchers to estimate flood risk. An
analysis of most of the papers show predominance of classifying land use pattern of
the flood prone areas into conventional categories like cropland, urban area, barren
land etc (Islam et al., 2001). However, these conventional land cover classifications
do not serve the purpose optimally. Since damage inflicted over areas of high value
land use is of greater concern a more detailed classification of the economically
intensive land use is recommended. It can be done meaningfully by clubbing low
value land use areas like barren land, forest, marshy land into one category and
applying a rigorous classification of arable land and urban areas (Honda et al.,
1997). It is also realized that a data archive for selected satellites should be built for
the developing countries. This kind of archive would depict the background land
cover condition for the study area over different seasons and facilitate the process
of change detection. Only with such a data set can remote sensing of flooded areas
become a useful tool in flood management and mitigation.

3. Flood Hazard & Risk Mapping with GIS and Remote Sensing
Another primary issue for flood management is to identify the area having higher
hazard potential. Hazard can be defined as some threat, natural, technological, or
civil to people, property and environment and risk is viewed as the probability that
a hazard will occur during a particular time period. Flood is a natural hazard and
flood risk is defined in terms of hundred-year flood (Godschalk, 1991). The issue
of preparing a reliable hazard map is one of the latest concerns within the subject
of flood management.
Different approaches have been taken to map the potential hazard. Rejesk
(1993) introduces three different methods for hazard zoning. His first method de-
scribes a binary model which evaluates whether the hazard is present or not in a
particular raster cell. The second method involves ranking different locations of an
area depending upon the intensity of the hazard present. In the last approach some
‘hazard’ values have been assigned to each of the raster cells based on the results
of a multivariate model which were built up on a host of variables related to river
flooding and associated hazards. But the quantification of the severity of the hazard
in some units or numbers as proposed by Rejesk (1993) in his last approach was
criticized by some authors (Wadge et al., 1993).
Flood depth is considered as the most important indicator of the intensity of
the hazard (Islam et al., 2001, 2000; Townsend et al., 1998; Wadge et al., 1993).
290 JOY SANYAL AND X. X. LU

Flood risk maps are prepared depending on the estimated depth of inundation. The
estimation is commonly derived from various hydrological and remotely sensed
data. For identifying flood depth it is important to classify the phenomenon of river
flooding into two categories, namely ‘non-source flood’ and ‘source flood’. ‘Non-
source flood’ is defined as inundation caused by well distributed rain storm over
a large area while ‘source flood’ describes inundation caused primarily by over
bank flow, predominantly affecting areas near the river channel (Liu et al., 2002).
These two basic characteristics of overland flow have significant implications for
formulating the GIS model. In the case of ‘non-source’ floods, all the raster cells or
vector points having an elevation below the water level are considered as ‘inund-
ated’ while for source floods, it is necessary to simulate the path of over bank flow
from the main channel to the adjacent flood plain to accurately estimate the flood
affected area.
The concept of topographical convergence or wetness index (Beven et al., 1979;
Moore et al., 1991; Wolock et al., 1995) was used to measure the depth of in-
undation (Townsend et al., 1998). The concept of wetness index is based on the
assumption that the accumulation of water in a particular cell of a raster depends
upon the area of the upslope region contributing water to that particular cell. The
main problem of using this index is that when a slope tends to zero the wetness
index becomes undefined. Thus this index is not very useful for modelling in
extremely flat flood plains. Townsend et al. (1998) also used another model for
simulating ‘source flooding’. The model assumes that the potential for any site to
be inundated is directly related to the difference in elevation between that site and
the river at its nearest hydrological link.
Perhaps the most innovative, simple and cost effective study regarding flood
hazard management has been conducted by Islam et al. (2001). He assessed the
flood depth from NOAA AVHRR imageries simply by the tonal difference of the
flood water. In this study, the flood affected area was subdivided into different flood
depth zones using supervised classification. To accurately identify the training sets,
the AVHRR data was superimposed over a DEM. Flood hazard has been assessed
by calculating a weighted score for each land use, physiographic and geologic
division of the country. The highlight of this methodology is that it assigns greater
weight to the categories of deeper flood depth in an exponential manner. In other
words, for the shallow depth, the weight increases in a progressive manner but
beyond a certain depth the increment of weight is much higher than the previous
depth categories. This process ensures that areas having higher depth of inundation
will be assigned a high ‘hazard denomination’. The rationale is that after a certain
depth, the flood water becomes very destructive and identifying this critical depth
is crucial for mapping hazard zone. This depth of water, or more precisely the
critical river stage, is likely to vary from region to region depending on the local
topography, building materials, settlement pattern etc.
Population density and hazard ranking have been multiplied to produce land de-
velopment priority maps. Zones of this type of maps indicate development priority
APPLICATION OF REMOTE SENSING IN FLOOD MANAGEMENT 291

for flood counter measures (Islam et al., 2002). This method is heavily dependent
upon the skill of the analyst who chooses the training sites for different depth of
inundation. The process of choosing the training sets is likely to be hampered by
forest cover and varying sediment load of the flood water which may alter the nat-
ural relationship of depth of the water and its tonal appearance. The main limitation
of this approach is that AVHRR imageries are too coarse to be used for analyzing a
local flood and cloud free images are very rare, especially in the monsoon season.
Boyle et al. (1998) used an even more diverse dataset to estimate flood risk.
Land use, hydraulic characteristics and human activities like demography, property
values and land ownership have been taken into consideration to determine the
type, location and severity of the hazard produced by the floods. Maps have been
prepared showing the properties under potential risk in the event of floods having
500, 100 to 5–2 years return period.
Flood hazard maps derived from coarse to moderate resolution imageries is
misfit in the developed countries. This kind of flood hazard maps can hardly make
any improvement to the existing flood insurance infrastructure in North America
or Western Europe. But in contrast, these maps are extremely useful for the devel-
oping countries of monsoon Asia, countries like China, India and Bangladesh etc.
These countries often suffer from devastating monsoon floods and a large propor-
tion of the population in these countries live in the flood prone areas. In this part
of the world flood insurance maps are often unheard of. Here flood hazard maps
are primarily required for saving the lives and livelihood of millions of marginal
farmers. Thus, these flood hazard maps can be used meaningfully by planners to
formulate strategies to combat this natural hazard.
Developed countries also need to evaluate the accuracy of flood risk mapping.
A very accurate flood risk map will exclude those individuals, found to be located
at a very high flood risk area, from a flood insurance scheme. On the other hand,
the insurance sector would lower the premium for those individuals who would be
found at minimum flood risk zone. Thus, a very high resolution flood hazard map
is likely to reverse the basic principle of insurance that loss of few should fall upon
many (Clark, 1998). Therefore, generalizing the risk zones to a certain extent is
likely to be more beneficial to the community as a whole.
Hazard maps should have a good visual effect so that the end user can have an
overview of the entire situation at a glance. In recent times hazard maps have be-
come so technical in nature that they convey very little information to the planners
and policy makers (Rocha et al., 1994). Unless the experts of hydrology and GIS
overcome this handicap, hazard maps cannot be used optimally by all the relevant
users.
292 JOY SANYAL AND X. X. LU

4. Some Issues of Remote Sensing Applications with Special Reference to

Monsoon Asia
Having briefly given an overview of the development and methods of the applic-
ation of remote sensing in flood management, this section tries to highlight the
constraints in applying this technique. Some of the problems, for example, agricul-
tural damage assessment, are typical for monsoon Asia which is heavily dependent
on agriculture

4.1. DEPENDENCY OF DIGITAL ELEVATION MODELS IN FLOOD MANAGEMENT

In a majority of the studies dealing with the application of remote sensing in inund-
ated area delineation and flood risk assessment, digital elevation models (DEM)
are used to visualize the interface of flood water with the terrain. Flood depth is
normally calculated by subtracting the elevation of each cell in a raster from its
water level (Brouder, 1994; Townsend et al., 1998; Ali et al., 2001; Islam et al.,
2002). DEMs are also used to simulate the flood depth from discharge data and
very often the result is compared with actual flooded area derived from satellite
imageries. As the spatial extent of inundation is subjected to a method of cross
checking this methodology is likely to yield a more accurate flood map as com-
pared to one derived from pure hydrological modelling. The main drawback of this
approach is that it is fully dependent on the accuracy of the DEM (Jones et al.,
1998). In the largely featureless plains of monsoon Asia, the accuracy required of
the DEMs is almost unattainable.
In a flat flood plain, where a vertical error of 1 m in the DEM may lead to an
error of 100s of square kilometres in flood estimation, recognition of the magnitude
of errors in the DEM is important in hydrological modelling. This issue has been
addressed from the view point of its significance in flood plain mapping (Lee
et al., 1992; Hunter et al., 1995). In the near absence of high resolution DEMs
required for flood mapping in a very flat terrain, multi-date SAR imageries can
serve as a very potent alternative data set in monsoon Asia. For example, multidate
Radarsat imageries were used in monsoon Asia to depict progress of a flood from
its inception to the peak (Liu, 1999; Chen, 1999). This particular operation can
create a visualization of the course of inundation from the river channel to the
adjacent low lying areas of the flood plain and complement the method of flow
direction simulation. Although multi-date imageries can serve as an alternative to
flow direction simulation, it has severe limitation of determining the flood depth.
A very accurate terrain data base regarding the local flood plain morphology is an
essential prerequisite for such an operation. The flood water intersecting with the
slope is taken as the primary indicator of determining flood depth. Thus for the
gently sloping topography, the resolution of the terrain data actually controls the
accuracy of the estimated flood depth (Brakenridge et al., 1998). High resolution
satellite imageries or aerial photographs are needed for preparing an accurate DEM
which can meet the precision level of a flood depth investigation.
APPLICATION OF REMOTE SENSING IN FLOOD MANAGEMENT 293

One of the recent developments in the application of remote sensing to flood

related problems is the use of LIDAR (Light Detecting and Ranging) sensor. In the
developed countries, especially in USA, this technology has become very popular
for creating DEMs for flood prone areas. LIDAR sensors can readily identify the
vertical differences in the landforms and can be exploited as powerful instruments
to create DEMs of exceptional accuracy. This sensor can also detect the flood depth.
Although LIDAR sensor can attain the vertical accuracy of 5 cm or better, it is
difficult to map it in that resolution. Due to the limitation of GPS systems to locate
an aircraft/sensor functionally LIDAR generated DEMs are released at an accuracy
of 15 to 25 cm RMSE. Moreover the accuracy decreases gradually with increase
in the density of vegetation cover of the ground (Hodgson et al., 2003). Although
LIDAR data is even more expensive than the SAR imageries, sometimes it provides
the only appropriate option to do flood mapping in the extremely flat flood plains.
Fowler (2002) maintained that the resolution of the LIDAR data depends upon the
intensity of the laser pulses and any attempt to make the survey more intensive by
increasing the laser pulses would increase the cost of the data exponentially.

4.2. AGRICULTURAL DAMAGE ASSESSMENT

Boyle et al. (1998) classified flood damages into two categories; tangible and intan-
gible. Tangible damage occurs due to direct contact with the flood water whereas
intangible damage is exemplified by the loss of historical monuments, heritage sites
etc. A disease assuming a form of an epidemic due to flood is also categorized as
an intangible damage. In the agricultural landscape of monsoon Asia, estimation
of agricultural damages requires special attention. Asia’s population is predomin-
antly rural and its economy is heavily reliant on agriculture. Monsoon flood often
creates havoc on the economy by damaging the standing paddy. This problem is
unique for monsoon Asia because in the developed world damage to urban area
and infrastructure facilities is of concern to government authorities. Hence, in this
section we shall emphasize studies dealing with application of remote sensing in
assessment of agricultural damage.
Erosion of top soil due to a flash flood and deposition of flood borne coarse
sand reduce the fertility of soil very severely and thus have a negative impact on
agricultural economy. The process of change detection is found useful to monitor
this kind of damage to agricultural land. The most widely used procedure is to
monitor the change in brightness value (VB) at a particular wave length or different
bands to identify the erosion caused by a flood. Several change detection techniques
like Spectral Image Differencing (SID), Tasseled Cap Brightness Image Differen-
cing (TCBID), Principal Component Analysis (PCA), and Spectral Change Vector
Analysis (SCVA) are employed for the purpose of detecting the erosions due to
flooding, but for Landsat TM data, SCVA is found to yield most accurate results
(Dhakal et al., 2002).
294 JOY SANYAL AND X. X. LU

In the flood plains of monsoon Asia rice is the dominant crop. Landsat TM-
4 band (Near Infrared) has been extensively used to estimate the damage to rice
crops due to flood (Shibayama et al., 1989; Okamoto et al., 1996) while the use of
band 3 is also considered to be very effective in assessment of paddy field damage
(Yamagata et al., 1988) due to its very high reflectance in turbid or muddy water
typically associated with flooded paddy field in the riverine plains of monsoon Asia
(Miller et al., 1983; Patel et al., 1985). The studies in this direction even attempted
to estimate the quantum of crop damage from satellite data. For example, a DN
value of 84 (in TM-4) indicates a yield of 3.0 metric tons/ha, whereas a value
around 55 corresponds to no crop in North Korea (Okamoto et al., 1998). The
main shortcoming of formulating a universal range of DN number is that it might
fail to extract accurate information for different varieties of rice crops.
Traditional supervised classification by maximum likelihood method generally
perform well in flood affected agricultural area where the instantaneous field of
view (IFOV) is less than the inundated segment of land and the land use is domin-
ated by crop lands (Jensen et al., 1995; Hugenin et al., 1997). The mean value
of Normalized Difference Vegetation Index (NDVI) derived from JERS-1 OPS
data was used to differentiate damaged crops from undamaged one in North Korea
(Choen et al., 1998) because vegetation condition or biomass over a portion of land
is highly correlated with NDVI (Singh, 1989; Tappen et al., 1992; Gamon et al.,
1995; Michener et al., 1997; Lyon et al., 1998).

4.3. PROBLEM OF TEMPORAL RESOLUTION IN FLOOD MANAGEMENT

Devastating floods are generally low frequency, high magnitude natural phenom-
ena. Flash floods occur within a very short interval of time and the peak stage
remains only for a couple of hours, but the most extensive and severe damage takes
place during that time. With the current Radarsat resources it is very difficult to
capture the spatial extent of a flood at its peak. Thus, attempts have been made to
extrapolate the extent of inundation at the peak of a flood from an image acquired
at a later stage of the event. Some GIS algorithms in ARC/INFO are promising to
perform this extrapolation from an image that captures some standing water only at
a time when the flood peak had already passed. The method of ‘least accumulation
cost distance’ can provide a viable solution to this problem (Brivio et al., 2002).
This principle simulates flow direction from the river channel to the flood plain
based on the assumption that water flows through the path where the work done in
doing so is least.
This methodology yielded a remarkable accuracy of 96.7 percent when com-
pared with the aerial photographs of the peak of the particular flood event which
occurred in November 1994 in Italy. The main weakness of this approach is that
the ‘least accumulation cost distance’ operates on certain values of the raster cells
which represent the roughness of the terrain causing frictional drag to the overflow-
ing flood water. Since roughness is a function of a host of other geomorphic and
APPLICATION OF REMOTE SENSING IN FLOOD MANAGEMENT 295

lithological factors, it is very difficult to control the parameters of the experiments

in an area having diverse lithology and land use. Apart from the DEMs and remote
sensing data, field work conducted in recently flood affected area can prove useful.
Wang et al. (2002) used high-flood level marks of a recent flood on houses as a
supplementary data set to the DEMs for estimating flood depth. This kind of data
serve very well to reconstruct dimension of a past flood peak.
General studies of flood geomorphology also provide some insights to resolve
this problem. Many very high magnitude floods may leave permanent imprints in
the flood plain morphology rather than creating some transient features (Brunsden
et al., 1979). When the morphology of the basin perfectly suites with the magnitude
and frequency of the flood event, the flood created features attain a state of perman-
ency. Since the frequency of such an event is high in the humid tropics, the interval
between two high magnitude floods gets reduced here. It indirectly contributes to
the permanency of the flood created characteristic morphological features in the
flood plain (Gupta, 1983). The permanency of this kind of topographic features
would be more prominent in areas of high relief and coarser valley sediments
(Gupta, 1988). Hence in the humid tropics there is enough scope to use relatively
inexpensive high resolution optical remote sensing not necessarily at the time of
flooding to demarcate the areas vulnerable to river flooding. But the analyst has to
be trained enough to identify the morphological features in the flood plain typically
associated with high magnitude floods.

5. Conclusion and Prospective

The preceding discussion regarding various facades of the prospects and con-
straints of using remote sensing and GIS for flood management unveils some
significant facts. A majority of the researchers favoured multi date radar imageries
to observe a particular flood event and considered different image processing tech-
niques to overcome the limitations of remotely sensed data in flood delineation.
The main weaknesses of this type of approach are manifold. Most of the investig-
ations mentioned in this paper are heavily dependent on the availability of satellite
data, which is not always guaranteed for the time of peak flood. Rather than moving
towards a comprehensive flood management strategy, these papers concentrated on
some specific issues like delineation of flooded area. Most of these projects are on
a very high budget and no attempts were made try to keep the cost low. Monsoon
floods affect the developing countries more acutely than the developed ones. There-
fore, while framing the methodology one should be aware of its feasibility in the
operational area. Most of the studies in this field cannot achieve its desired level
of accuracy without a very high resolution DEM, therefore, we need to develop
alternative methodology to shed our dependence on high resolution terrain data.
A detailed hydro-geomorphological mapping depicting the trace of past floods
may help us in this direction. The hazard of monsoon flood and the destitution
of people associated with it are very different from the industrially developed
296 JOY SANYAL AND X. X. LU

countries. Hence, we should adopt an improvised but effective methodology of

risk and damage assessment to come out with meaningful flood hazard maps for
this region.
Application of remote sensing and GIS is convincingly a very efficient and cost
effective way of flood management. Use of very high resolution imageries like
IKONS or SPOT 5 have not been very popular yet in the field of flood management
due to its high price, but it is likely that with these imageries would be available
at a reasonable price and would be widely used for flood mapping. In the age
of internet, GIS has assumed new dimensions, especially for coping with natural
disasters like river flooding. The most important advantage of using internet based
GIS is that it has opened the door of GIS technology to the end users who would not
like to install expensive GIS software. One of the numerous examples is Arc IMS
technology. This technology has been used to develop a web enabled application
named Map Action Processing Digital Interactive Geo Resource (MAPDIGR) for
providing very recent information regarding flood risk to an analyst via internet
(Smith et al., 2002). This technology is at present at an embryonic stage of devel-
opment but has great potential for expand the user base of GIS technology for flood
management by substantially reducing the cost of operation.
Since the problem of flood is very acute in the developing countries of monsoon
Asia, special attention should be given to deal with this problem in the regional
context. GIS models having low cost and simple data requirement are likely to
attract the local authorities in the developing countries to adopt this technology
as an essential input towards a comprehensive flood management system. In the
age of all embracing flood plain management, these sophisticated technologies can
be very useful for the planners to formulate effective strategy for combating the
perpetual natural disaster of river flooding.

Acknowledgement
This study has been done as part of an ongoing project titled “Application
of Remote Sensing and GIS in Flood Hazard Management: A Case Study of
West Bengal, India”, funded by National University of Singapore research grant
R-109-000-049-112.

References
Ali, A. and Quadir, D. A.: 1987, Agricultural hydrologic and oceanographic studies in Bangladesh
with NOAA AVHRR data, International Journal of Remote Sensing 8(6), 917–925.
Ali, S., Hassan, A., Martin, T. C., and Hassan Q. K.: 2001, Geospatial tools for monitoring flood plain
water dynamics, In: M. Owe, K. Brubaker, J. Ritchie, and A. Rango, A. (eds), Remote Sensing
and Hydrology 2000, IAHS Publication, Oxford, pp. 465–471.
Andre, G., Guillande, R., and Bahoken, F.: 2002, Flood mapping using spatial radar and optical
imagery and digital elevation model: Limits and capacities, Houille Blanche-Revue Int. De LEAU
1, 49–54.
APPLICATION OF REMOTE SENSING IN FLOOD MANAGEMENT 297

Barton, I. and Bathols, J.: 1989, Monitoring floods with AVHRR, International Journal of Remote
Sensing 10(12), 1873–1892.
Berg, A. and Gregiore, J. M.: 1983, Use of remote sensing techniques for rice production forecast-
ing in West Africa, (Mali and Guinea: Niger-Bani project), ESA Satellite Remote Sensing for
Developing Countries, Ispra, Italy, pp. 161–168
Beven, K. J. and Kirkby, M. J.: 1979, A physically based, variable contributing area model of basin
hydrology, Hydrol. Sci. Bull. 24, 43–69.
Bhavsar, P. D.: 1984, Review of remote sensing applications in hydrology and water resource
management in India, Advances in Space Research 4(11), 193–200.
Blyth, K.: 1995, A telenetwork for acquisition, processing and dissemination of Earth Observation
data for monitoring and emergency management of floods, In: Proc. First ERS Thematic Working
Group Meeting on Flood Monitoring. European Space Agency/ESRIN, Frascati, Italy, 26–27
June.
Boyle, S. J., Tsams, I. K., Manber, A. S. C. E., and Kanaroglu, P. S.: 1998, Developing geographical
information system for land use impact assessment in flooding conditions, Journal of Water
Resource Planning and Management 124(2), 89–98.
Brakenridge, G. R., Tracy, R. T., and Know, J. C.: 1998, Orbital SAR remote sensing of the river
flood waves, International Journal of Remote Sensing 19(7), 1439–1445
Brimicombe, J., and Bratlet, J. M.: 1996, Linking geographical information system with hydraulic
simulation modeling for flood risk assessment: The Honk Kong Approach, In: M. F. Goodchild
(eds), GIS and Environmental Modeling, Oxford University Press, New York, pp. 165–168.
Brivio, P. A., Colombo, R., Maggi, M., and Tomasoni, R.: 2002, Integration of remote sensing data
and GIS for accurate mapping of flooded areas, International Journal of Remote Sensing 23(3),
429–441.
Brouder, J. A. M.: 1994, Flood study in the Meghna-Dhonagoda Polder, Bangladesh, In: Proc. Asian
Institute of Remote Sensing, Bangalore, India, 17–23 November.
Brunsden, D. and Thorns, J. B.: 1979, Landscape sensitivity and change, Trans. Inst. Br. Geogr., New
Series 4, 463–484.
Chen, P., Liew, S. C., and Lim, H.: 1999, Flood detection using multitemporal Radarsat and ERS
SAR data, In: Proc. 20th Asian Conference of Remote Sensing, Hong Kong, 22–25 November.
Choen, K. and Kwang-Hoon, C.: 1998, Flood damage mapping in North Korea using multi-sensor
data, In: Proc. 19th Asian Conference of Remote Sensing, Manila, Philippines, 16–20 November.
Clark, M. J.: 1998, Putting water in its place: A perspective on GIS in hydrology and water
management, Hydrological Processes 12, 823–834.
Collins, S. H. and Moon, G. C.: 1981, Algorithms for dense digital terrain models, Photogrammetric
Engineering and Remote Sensing 47, 71–76.
Coppock, J. T.: 1995, GIS and natural hazard: an overview from a GIS perspective, In: A. Carrara
and F. Guzzetti (eds), Geographical Information System in Assessing Natural Hazard, Kluwer
Academic, Netherlands, pp. 21–34.
Deutsch, M.: 1976, Optical processing of ERTS data for determining extent of 1973 Mississippi River
flood, USGS Professional Paper 929, ERTS-1, A New Window on Our Planet, pp. 209–222.
Deutsch, M. and Ruggles, F.: 1974, Optical data processing and projected application of the ERTS1
imagery covering the 1973 Mississippi River Valley floods, Water Resource Bulletin 10(5), 1023–
1039.
Deutsch, M., Ruggles, F., Guss, P., and Yost, E.: 1973, Mapping the 1973 Mississippi floods from
the Earth Resource Technology satellites, In: Proc. International Symposium on Remote Rensing
and Water Resource Management, American Water Resource Association, No. 17, Burlington,
Ontario, pp. 39–55.
Dhakal, A. S., Amda, T., Aniya, M., and Sharma, R. R.: 2002, Detection of areas associated
with flood and erosion caused by a heavy rainfall using multi temporal Landsat TM data,
Photogrammetric Engineering and Remote Sensing 68(3), 233–239.
298 JOY SANYAL AND X. X. LU

Fowler, R. A.: 2002, LIDAR for flood mapping, Earth Observation Magazine 9(7)
(http://www.eomonline.com/Common/Archives/July00/robert.htm, true as par 30th May, 2003)
Gamon, J. A., Field, C. B., Goulden, M. L., Griffin, K. L., Hartley, A. E., Joel, G., Penuela, J., and
Valentini, R.: 1995, Relationship between NDVI, canopy structure and photosynthesis in their
Californian vegetation types, Ecological Applications 5, 28–41.
Godschalk, D. R.: 1991, Disaster mitigation and hazard management, In: T. E. Drabek and G. J. Hoet-
mer (eds), Emergency Management: Principles and Practice for Local Government, International
City Management Association, Washington, D.C. pp. 131–160.
Gumley, L. and King, M. D.: 1995, Remote sensing of flooding in the US Upper Midwest during the
summer of 1993, Bulletin of American Meteorological Society 76(6), 933–943.
Gupta, A.: 1983, High magnitude floods and stream channel response, Special Publication of
International Association of Sedimentology 6, 219–227.
Gupta, A.: 1988, Large floods as geomorphic events in the humid tropics, In: Victor R. Barker, Craig
R. Kochel, and Peter C. Paton (eds), Flood Geomorphology, John Wiley and Sons, pp. 301–321.
Hallberg, G. R., Hoyer, B. E., and Rango, A.: 1973, Application of ERTS1 imagery to flood inund-
ation mapping, NASA Special Publication No. 327, Symposium on Significant Results Obtained
from the Earth Resources Satellite 1, Vol. 1, Technical presentations, section A, pp. 745–753.
Hausmann, P. and Weber, M.: 1988, Possible contributions of hydroinformatics to risk analysis in
insurance, In: Proc. 2nd International Conference on Hydroinformatics, Zurich, Switzerland,
9–13 September, Balkema, Rotterdam.
Hess, L. L., Melack, J. M., and Simonett, D. S.: 1990, Radar detection of flooding beneath the forest
canopy – a review, International Journal of Remote Sensing 11(7), 1313–1325.
Hodgson, M. E., Jensen, J. R., Schmidt, L., Schill, S., and Davis, B.: 2003, An evaluation of LIDAR
and IFSAR derived digital elevation models in leaf-on conditions with USGS level-1 and level-2
DEMs, Remote Sensing of Environment 84, 295–308.
Honda, K. C., Francis, X.J., and Sah V. P.: 1997, Flood monitoring in central plain of Thailand using
JERS-1 SAR data, In: Proc. 18th Asian Conference of Remote Sensing, Malaysia, 20–24 October.
Horritt, M. S. and Bates, P. D.: 2001, Predicting flood plain inundation: Raster based modeling versus
the finite element approach, Hydrological Processes 15(5), 825–842.
Huguenin, R. L., Karaska, M. A., Blaricom, D. V., and Jensen, J. R.: 1997, Subpixel classification of
bald cypress and tupelo gum trees in Thematic Mapper imagery, Photogrammetric Engineering
& Remote Sensing 63, 717–725.
Huh, O. K., Ali, A., and Quadir, D. A.: 1985a, Mapping of green leaf biomass over Bangladesh
with NOAA satellite AVHRR data, Manuscript Report prepared for UN Flood and Agriculture
Organization, Coastal Studies Institute, Louisiana State University., Baton Rouge, LA.
Huh, O. K., Ali, A., and Quadir, D. A.: 1985b, Observations on the surface waters of the Bay
of Bengal with NOAA satellite AVHRR data, Manuscript Report prepared for UN Flood and
Agriculture Organization, Coastal Studies Institute , Louisiana State University, Baton Rouge,
LA.
Huh, O. K., Ali, A., and Quadir, D. A.: 1985c, Use of NOAA satellite AVHRR data to monitor
river flood hydrology in Bangladesh, Manuscript Report prepared for UN Flood and Agriculture
Organization, Coastal Studies Institute, Louisiana State University, Baton Rouge, LA.
Hunter, G. J. and Goodchild, M. F: 1995, Dealing with error in spatial databases – a simple case-
study, Photogrammetric Engineering & Remote Sensing 61(5), 529–537.
Imhoff, M. L., Vermillion, C., Story, M. H., Choudhury, A. M., Gafoor, A., and Polcyn, P.: 1987,
Monsoon flood boundary delineation and damage assessment using space borne imaging radar
and Landsat data, Photogrammetric Engineering and Remote Sensing 53(4), 405–413.
Islam, M. M. and Sadu, K.: 2000a, Development of flood hazard maps of Bangladesh using NOAA-
AVHRR images with GIS, Hydrological Science Journal 45(3), 337–355.
Islam, M. M., and Sadu, K.: 2000b, Flood hazard assessment in Bangladesh using NOAA-AVHRR
data with geographical information system, Hydrological Processes 14(3), 605–620.
APPLICATION OF REMOTE SENSING IN FLOOD MANAGEMENT 299

Islam, M. M. and Sadu, K.: 2000c, Satellite remote sensing data analysis for flood damaged zoning
with GIS for flood management, Journal of Hydraulic Engineering 44, 301–306.
Islam, M. M. and Sadu, K.: 2001, Flood damage and modelling using satellite remote sensing data
with GIS: Case study of Bangladesh; In: Jerry Ritchie et al. (eds), Remote Sensing and Hydrology
2000, IAHS Publication, Oxford, pp. 455–458.
Islam, M. M. and Sadu, K.: 2002, Development of priority map remote sensing data for flood counter
measures by geographical information system, Journal of Hydrological Engineering 7(5), 346–
355.
Jensen, J. R., Rutchey, K., Koch, M., and S. Narumalani: 1995, Inland Wetland Change Detection
in the Everglades Water Conservation Area 2A Using A Time Series of Normalized Remotely
Sensed Data, Photogrammetric Engineering & Remote Sensing 61(2), 199–209.
Jin, Y. Q.: 1999, A flooding index and its regional threshold values for monitoring in china from
SSM/I data, International Journal of Remote Sensing 20(5), 1025–1030.
Jones, K. H.: 1998, A comparison of algorithms used to compute hill slope as a property of the DEM,
Comput. Geosci. 24(4), 315–323.
Kundus, P., Karszenbaum, H., Pultz, T., Paramuchi, G., and Bava, J.: 2001, Influence of flood condi-
tions and vegetation status on the radar back scatter of wetland ecosystem, Canadian Journal of
Remote Sensing 27(6), 651–662.
Lee, J., Snyder, P. K., and Fisher, P. F.: 1992, Modeling the effect of data errors on feature extraction
from digital elevation models, Photogrammetric Engineering & Remote Sensing 58(10), 1461–
1467.
Liu, R. and Liu, N.: 2002, Flood area and damage estimation in Zhejiang, China, Journal of
Environmental Management 66, 1–8.
Liu, Z. L., Huang, F., Li, L.Y., and Wan, E. P.: 1999, Dynamic monitoring and damage evaluation of
flood in northwest Jilin with remote sensing, In: Proc. 20th Asian Conference of Remote Sensing,
Hong Kong, 22–25 November.
Long, N. T. and Trong, B. D.: 2001, Flood monitoring of Mekong River Delta, Vietnam using ERS
SAR Data, In: Proc. 22nd Asian Conference of Remote Sensing, Singapore, 5–9 November.
Lowry, R. T., Langham, E. J., and Murdy, N.: 1981, A preliminary analysis of SAR mapping of
Manitoba flood, May 1979, In: Proc. Satellite Hydrology, Fifth Anniversary. William T. Pecora
Memorial Symposium on Remote Sensing 1979, American Water Resource Association Technical
Publication No. TPS81-1, pp. 316–323.
Lyon, J. D., Yuan, G., Lunetta, R. S., and Elvidge, C. D.: 1998, A change detection experiment using
vegetation indices, Photogrammetric Engineering and Remote Sensing, 64, 143–150.
McGinnis, D. F. and Rango, A.: 1975, Earth Resource Satellite System for flood monitoring,
Geophys. Res. Lett. 2(4), 132–135.
Melack, J. M., Hess, L. L., and Sippel, S.: 1994, Remote sensing of lakes and floodplains in the
Amazon Basin, Remote Sensing Review 10, 127–142.
Michener, W. K. and Houhoulis, P. F.: 1997, Detection of vegetation changes associated with exten-
sive flooding in a forested ecosystem, Photogrammetric Engineering & Remote Sensing 63(12),
1363–1374.
Miller, L. D., Yang, Y., Matthews, M., and Irons, R. T.: 1983, Correlations of rice yields to radiomet-
ric estimates of canopy biomass as a junction of growth stage, In: Proc. 4th Asian Conference of
Remote Sensing, 10–15 November, Colombo, Srilanka, A.6.1–A 6.21.
Moore, I. D., Grayson, R. B., and Ladson, A. R.: 1991, Digital terrain modelling: A review of
hydrological, geomorphological, and biological applications, Hydrological Processes 5, 3–30.
Morrison, R. B. and Cooley, M. E.: 1973, Assessment of flood damage in Arizona by means of
ERTS-1 imagery, In: Proc. Symposium on Significant Result Obtained from the Earth Resource
Satellite 1, Vol. 1, New Carrollton, Maryland, pp. 755–760.
Morrison, R. B. and White, P. G.: 1976, Monitoring flood inundation, USGS Professional Papers No.
929, ERTS-1, A New Window on Our Planet, pp. 196–208.
300 JOY SANYAL AND X. X. LU

Nico, G., Pappalepore, M., Pasquariello, G., Refice, A., and Samarelli, S.: 2000, Comparison of
SAR amplitude vs. coherence flood detection methods- a GIS application, International Journal
of Remote Sensing 21(8), 1619–1631.
Oberstadler, R., Honsch, H., and Huth, D.: 1997, Assessment of the mapping capabilities of ERS-1
SAR data for flood mapping: A case study of Germany, Hydrological Processes 10(10), 1415–
1425.
Okamoto, K. and Fukuhara, M.: 1996, Estimation of paddy field area using the area ratio of categories
in each pixel of Landsat TM, International Journal of Remote Sensing 17, 1735–1749.
Okamoto, K., Yamakawa, S., and Kawashima, H.: 1998, Estimation of flood damage of North Korea
in 1995, International Journal of Remote Sensing 19(2), 365–371.
Patel, N. K., Singh, T. P., Sahai, B., and Patel, M.: 1985, Spectral response of rice crops and its
relation to yield and yield attributes, International Journal of Remote Sensing 6, 657–664.
Pope, K. O., Rejmankova, E., Paris, J.F. et al.: 1997, Detecting seasonal flooding cycles in marshes
of the Yucatan Peninsula with SIR-C polarimetric radar imagery, Remote Sensing of Environment
59(2), 157–166.
Profeiti, G. and MacIntosh, H.: 1997, Flood management through Landsat TM and ERS SAR data:
a case study, Hydrological Processes 11(10), 1397–1408.
Quan, W., Watanabe, M., Hayashi, S., and Murakami, S.: 2003, Using NOAA AVHRR data to assess
flood damage in China, Environmental Monotoring and Assessment 82, 119–148.
Rango, A. and Anderson, A. T.: 1974, Flood hazard studies in the Mississippi River Basin using
remote sensing, Water Resource Bulletin 10(5), 1060–1081.
Rango, A. and Solomonson, V. V.: 1974, Regional flood mapping from space, Water Resource
Research 10(3), 473–484.
Rango, A. and Solomonson, V. V.: 1977, The utility of short wavelength (<1 mm) remote sensing
techniques for the monitoring and assessment of hydrological parameters, In: Proc. 11th Int.
Symp. on Remote Sensing of Environ., Ann Arbor, MI, 25–29 April 1977, pp. 55–64.
Rashid, H. and Pramanik, M. A. H.: 1993, Areal extent of the 1988 flood in Bangladesh: How much
did the satellite imagery show? Natural Hazards 8, 189–200.
Reed, S., Johnson, D., and Sweeney, T.: 2002, Application of national geographic information system
data base to support two-year flood and the threshold run-off estimate, Journal of Hydrological
Engineering 7(3), 209–219.
Rejesk, D.: 1993, GIS and risk: A three cultural problem, In: M. F. Goodchild, B. O. Parks, and L.
T. Steyaerts (eds), Environmental Modeling With GIS, Oxford University Press, New York, pp.
49–267.
Rocha, J. S., Marques, Z., Ramos, I., and Ameida, R.: 1994, Simulation of risk flood areas on GIS, In:
Tsakiris, G. and Santos, M. A. (eds), Advances in Water Resource Technology and Management,
Balkema, Rotterdam, pp. 375–383.
Rosenqvist, A., Shimada, M., Chapman, B., Freeman, A., Grande, G. De, and Saatchi, S.: 2000,
The global rainforest mapping – a review, International journal of Remote Sensing 21(6 & 7),
1375–1387.
Rosenqvist, A., Forsberg, B. R., Pimentel, T., Rauste, Y. A., and Riche, J. E.: 2002, The use of space
borne radar data to model inundation patterns and trace gas emission in the central Amazon flood
plain, International Journal of Remote Sensing 23(7), 1303–1328.
Ruangsiri, P., Sripumin, R., Polongam, S., Kanjanasuntorn, P., and Wongparn, S.: 1984, State of
flooding in the Mun-Chi River Basin area, N. E. Thailand by digital Landsat data analysis,
Report: Remote Sensing Division, National Resource Council of Thailand, Bangkok, Thailand.
Sado, K., and Islam, M. M.: 1997, Satellite remote sensing data analysis for flooded area and weather
study: Case study of Dhaka city, Bangladesh, Journal of Hydraulic Engineering 41, 945–950.
Saikh, M., Green, D., and Cross, H.: 2001, A remote sensing approach to determine environmental
flows for wet lands of the Lower Darling River, New South Wales, Australia, International
Journal of Remote Sensing 22(9), 1737–1751.
APPLICATION OF REMOTE SENSING IN FLOOD MANAGEMENT 301

Sharma, P.: 1999, Flood risk mapping of Dikrong subbasin of Assam, Proc. Map Asia 1999. Source
: http://www.gisdevelopment.net/proceedings/mapindia/1999/index.htm
Shibayama, M. and Akiyama, T.: 1989, Seasonal visible, near-infrared and multi infrared spectra
of rice canopies in relation to LAI and above- ground dry phytomass, Remote Sensing of the
Environment 27, 119–127.
Singh, A.: 1989, Digital change detection techniques using remotely-sensed data, International
Journal of Remote Sensing 10, 989–1003.
Smith, L. C.: 1997, Satellite remote sensing of river inundation area, stage and discharge: A review,
Hydrological Processes 11, 1427–1439.
Smith, C. and Price, J.: 2002, A new generation of flood management, Arc User 5(4), 20–22.
Tappen, G. D., Tyler, D., Wehde, M., and Moor, D.: 1992, Monitoring rangeland dynamics in Senegal
with AVHRR data, Geocarto International 1, 87–98.
Tholey, N., Clandillon, S., and De Fraipont, P.: 1997, The contribution of space borne SAR and
optical data in monitoring flood events; examples in Northern and Southern France, Hydrological
Processes 11(10), 1427–1439.
Toutin, T.: 2002, DEM from stereo Landsat 7 ETM+ data over high relief areas, International Journal
of Remote Sensing 23(10), 2133–2139.
Townsend, P. A. and Walsh, S. J.: 1998, Modelling flood plain inundation using integrated GIS with
radar and optical remote sensing, Geomorphology 21(98), 295–312.
Wadge, G., Wislocki, A. P., Pearson, J., and Whittow, J. B.: 1993, Mapping natural hazards with
spatial modeling System, In: P. M. Mather (eds), Geographic Information Handling Research
and Applications, John Wiley and Sons Inc., New York.
Wang, W., Hess, L. L., Filoso, S., and Melack, T. M.: 1995, Understanding the radar back scattering
from flooded and non-flooded Amazonian Forests: result from canopy back scatter modelling,
Remote Sensing of Environment 54(3), 324–332.
Wang, Y., Colby, J. D., and Mulcahy, K. A.: 2002, An efficient method for mapping flood extent in
a coastal flood plain using Landsat TM and DEM data, International Journal of Remote Sensing
23(18), 3681–3696.
Wiesnet, D. R., McGinnis, D. F., and Pritchard, J. A.: 1994, Mapping of the 1973 Mississippi River
floods by the NOAA-2 satellite, Water Resource Bulletin 10(5), 1040–1049.
Wolock, D. M.: 1995, Effects of subbasin size on topographic characteristics and simulated flow
paths in sleepers river watershed, Vermont, Water Resour. Res. 31(8), 1989–1997.
Yamagata, Y. and Akiyama, T.: 1988, Flood damage analysis using multi temporal Landsat TM data,
International Journal of Remote Sensing 9(3), 503–514.
Yang, C., Zhou, C., and Wan, Q.: 1999, Deciding the flood extent with Radarsat SAR data and Image
Fusion, In: Proc. 20th Asian Conference of Remote Sensing, Hong Kong, 22–25 November.