Asian Journal of Geoinformatics, Vol.6, No.

2 (2006)

Asian Journal of Geoinformatics, Vol.6, No.2 (2006)

Flood Inundation Mapping and Damage Assessment

Using Multi-temporal RADARSAT and IRS 1C
LISS III Image
Md. Rejaur Rahman

Assistant Professor, Dept. of Geography & Environmental Studies

University of Rajshahi, Rajshahi-6205, Bangladesh.

Abstract

The availability of SAR data from the RADARSAT satellite offers an opportunity for continuous observation of
flood events since it has penetration capacity through clouds. This makes it possible to monitor the progress of flood and to
produce accurate, rapid and cost effective flooding maps. A study using combined RADARSAT SAR and IRS LISS III 1C
images in flood detection, the analysis of spatial extend and temporal pattern of flood inundation and damage assessment
was described in this paper. The main objective of this study was to monitor flood by using multi temporal RADARSAT
data (04, 11, 13 & 20 September 2003). The SAR images were first co-registered, filtered and calibrated. Then the
inundated areas were extracted from the SAR images using threshold method. Flood damage assessment of the study area
was another objective of the study. For this, a pre-flood land use/land cover map was prepared using IRS 1C LISS III
image and supervised technique was used for classification of image. By overlaying flood-inundated map with land use/
land cover map, flood damage was assessed. The percentage of inundation of each of the land cover classes and each village
at the time of flood was analyzed. The study reveals that the combination of microwave and optical satellite data can be
utilized advantageously for inundation mapping and damage assessment.

1. Introduction 1997, Delmeire 1997, Profeti and Macintosh 1997, Tholey

et al. 1997). The analysis of spatial extent and temporal
Unusual or above normal surface water flow that pattern of flood inundation from remotely sensed imagery
inundate otherwise high ground is called a flood. Flood is of critical importance to flood mitigation. The flood
hazard has long been recognized as one of the most inundation map helps the decision makers to make a scien-
recurring, widespread and disastrous natural hazards in the tific assessment and for better management of relief acti-
densely populated regions of South Asia. In many parts vities. On the other hand, based on satellite data acquired
of Indian subcontinent, flooding reaches catastrophic during flood, pre-flood and post flood along with the
proportions during the summer season. Flooding is not ground information, flood damages can be estimated.
just confined to monsoon Asia but is a globally pervasive Based on the duration of flooding, magnitude of the flood,
hazard. Therefore, flood monitoring for damage and relief area affected, types of land use features etc., flood damage
management is a prerequisite. In recent years, satellite map can be prepared.
technology has become extremely important to provide
cost-effective, reliable and critical mechanism for preven- 2. Objectives of the Study
tion, preparedness and relief management of flood
disaster. With the availability of multiple satellite data, it The study was initiated with the following objectives:
is now possible to monitor flood situation in the particular 1) Flood inundation mapping using multi temporal
area. It can also provide approximate duration of flood- RADARSAT image.
ing. The cloud penetrating capability of radar has long been 2) Study of flood wave propagation using multi-date
an attractive feature to flood monitoring. It is inspiring to flood maps.
see that the use of space-borne radar imagery for flood 3) Flood damage assessment and create a GIS data-
monitoring has been increasing (Imhoff et al. 1987, Blyth base.

© 2005 AARS, All rights reserved.

Corresponding : [email protected]
Ph: (Resi): +88 0721 761140

AJG_3042_Rahman 1 19/04/2006, 9:46 AM

 Flood Inundation Mapping and Damage Assessment Using Multi-temporal RADARSAT and IRS 1C LISS III Image

3. Location and Extent RADARSAT data was acquired for the date of 04, 11, 13
and 20 September 2003 and LISS III data was acquired
Puri district of Orissa State of India was selected for for the date of 12 July 2003. The spatial resolution for
this study. It is situated in the eastern part of Orissa and is RADARSAT and LISS III is 100m and 23.5m respectively.
one of the four holy dhams of Hinduism. Puri is also Some field data such as cross check of flooded area and
called “Sri Purusottama Dham” or “Martya Vaikuntha”, extent; damages information from the villagers etc. and
the abode of Lord Vishnu on earth. The study area lies data obtained by GPS (Global Positioning System) were
between 19°25' to 20°30' N latitude and 85°00' to 86° used in this study. ILWIS (Integrated Land and Water
30' E longitude. The location of the study area is shown Information System), Erdas Imagine and ArcView image
in Fig. 2. The District is in deltaic zone of Mahanadi processing and GIS software was used for image process-
system with major rivers like Kushabhadra, Bhargabi, ing and analysis.
Daya and Devi with tributaries and distributaries of small The overall methodology of this study is shown in
rivers, such as Luna, Makara, Rajua, Prachi, Dhanua, Fig. 1. However, the method followed is discussed in the
Ratnachira. It was reported that these rivers and their following sections.
congestion pattern along with high rainfall intensity are
the main causes of flooding in the study area. The network 4.1 Image processing-Geometric Correction
of rivers spread through the district with geographical Raw digital image usually contain geometric distortions
disadvantage makes the district flood-prone and thus the and can be corrected by analyzing well-distributed ground
district experiences flood at regular intervals. control points (GCPs) occurring in the image. For this study
two types of geometric correction procedure were used
4. Data Used and Methodology: i.e. image-to-map registration and image-to-image
registration. The LISS III scene was first spatially geo-
For this study RADARSAT-1 (ScanSAR wide) and referenced to a Universal Transverse Mercator (UTM) map
IRS 1C LISS III (multi-spectral) digital data was used. projection using the first order affine transformation and

Fig.-1: Flow Diagram Showing Methodology of the Study

AJG_3042_Rahman 2 19/04/2006, 9:46 AM

 Asian Journal of Geoinformatics, Vol.6, No.2 (2006)

Fig.-2: Location of the Study Area

resample with a nearest neighbour algorithm. Ground filtering techniques can be used. Here filtering technique
control points were identified from topographic maps and was used and the median filter was applied for this pur-
total of 26 GCPs were collected and well defined on the pose.
image. A root mean square error of less than quarter pixel
(0.25) was accepted for the correction process. The image 4.3 Preparation of Thematic Maps
was then resampled by converting 23.5m pixel size. The All the thematic maps (boundary, villages, blocks,
geometrically corrected LISS III image was cross checked drainages, road, etc.) were scanned and transferred for on-
with the topographic map to optimize accuracy. After that screen digitization to generate various vector maps. The
in the image-to-image registration procedure, multi tem- scanned maps were first geo-referenced, resample and
poral RADARSAT images were geometrically corrected then polygon and raster maps were created separately by
with validated geocoded LISS III image, used as a referenced polygonization and rasterization in ILWIS (GIS) envi-
for the registration of RADARSAT images. This method ronment. A pre-flood river course and water bodies were
consists of collecting ground control point pairs from a digitized form the pre-flood LISS III 1C image.
split-screen display of both the uncorrected and master
image. For RADARSAT images, root mean square was 4.4 Creation of a Flood Map using RADARSAT multi-
less then quarter pixel (0.25) in the geometric correction. temporal Imagery
These images were then resampled using nearest neigh- In RADARSAT image inundated areas appear in dark
bour algorithm with affine transformation and 100m pixel tone, which can be clearly delineated from the non-inun-
size. dated areas due to the surface roughness of water and land.
The inundated area was extracted from the SAR images
4.2 Reduction of speckle in RADARSAT SAR imagery by using threshold method. The threshold method is as
Speckle is the result of the interaction between the radar simple as (Gong and Xiao, 2001):
pulse and the different scatterers of a distributed target. It
has a great significance in SAR image interpretation, since Water, if DN < T0 and
it limits the radiometric resolution and therefore the sub- Land, if DN > T0
sequent ability discriminate between different intensity
levels. To remove the speckle noise, which is system phe- Where, DN is the Digital Number value of RADARSAT,
nomenon and is not result of spatial variation of average and T0 is the threshold.
reflectivity of the radar illuminated surface, two methods The DN values within field information areas were
(FAO/ESA, 1989): SAR image multi-look processing and extracted from the images of all four dates to determine

AJG_3042_Rahman 3 19/04/2006, 9:46 AM

 Flood Inundation Mapping and Damage Assessment Using Multi-temporal RADARSAT and IRS 1C LISS III Image

image classification threshold (T0). After that using the land use/land cover map, “cross” operation of ILWIS was
simple technique of density slicing open water and non-water used. The Cross operation performs an overlay of two
(land) areas were differentiated from the image in ILWIS raster maps. Pixels on the same positions in both maps are
(GIS) environment. To delineate the actual flooded area, compared; the occurring combinations of class names,
permanent water bodies (e.g. river, lake, ponds etc.) were identifiers or values of pixels in the first input map and
subtracted from the open water. Same techniques were those of pixels in the second input map are stored. These
applied to all RADARSAT images to get the actual flooded combinations give an output cross map and a cross table.
area. With all flooded maps, the nature of spatial extent of The cross table includes the combinations of input values,
flood was analyzed and prepared a super-imposed map to classes or IDs, the number of pixels that occur for each
show how flooding evolved through time. combination and the area for each combination. From the
cross table, area of the combination of flooded area and
4.5 Land use / Land cover Map Generation each land use/land cover classes were calculated for assess-
To generate a land use/land cover supervised classi- ing the damaged/affected area of each class.
fication technique was used for image classification over
FCC (False Color Composite), which was generated using 5. Results and Discussion
band 3, 2 and 1 of IRS LISS III 1C. In supervised classi-
fication, first training samples were given for each of the 5.1 Flood Inundation Mapping
land use/land cover classes. According to Level-1 classi- Earlier it was stated that four RADARSAT data were
fication of land use (Lillesand and Keifer, 2002) training used for flood inundation mapping for Puri district. Form
samples were recognized in six classes viz. crops land, all RADARSAT data sets, flood inundation layer were
fallow land, forest/plantation area, sandy area, water bodies derived and area of inundation were calculated. Fig. 3
and urban areas. Maximum likelihood classifier was used shows the inundation area and spatial extent of flood over
for land use/land cover classification. time period. These inundation layers along with village
layer were crossed to get the village wise flood inundation
4.6 Flood Damage Assessment area. Table 1 depicted that out of 1710 villages, 902 villages
Flood inundation layer extracted from RADARSAT were affected due to floods in 04 September and the area
image, was intersected with land use/land cover map for of inundation was 89918 hectares or 25.87% of the total
assessment of damage. The percentage of inundation of area of Puri district. In 11 September, the flood situation
each of the land cover classes and each village at the time was further worsened and number of affected villages
of flood was analyzed. For overlaying the flood map and increased to 1000 and inundated area was 102963 hectare,

Fig.-3: Computed Flood Map (obtained from RADARSAT Imagery)

AJG_3042_Rahman 4 19/04/2006, 9:46 AM

 Asian Journal of Geoinformatics, Vol.6, No.2 (2006)

which was 29.63% of the total area. The Table also area. So, it may be said that the situation of flood in 11
highlighted that the worst condition of flood was reduced September was catastrophic in nature. Moreover, Table 2
in 13 September and the number of affected villages were highlighted the details of flood-affected villages and
counted 779. Furthermore, flood inundation shows that affected areas according to the percentage of flooded area
in 20 September, the numbers of affected villages were of the village area. From this table flood situation of the
come down to 466 and the inundated area was 44063 study area can be predicted.
hectare, which was 12.68% of the total area (Table 1). So Block wise analysis of flood affected area shows that
it may said that from 13 September it was receding period out of 11 blocks, 2 blocks namely Kanas and Bramhagiri
of flood. were affected severely where flooded area was about 74%
With the four flood maps, the nature of spatial extent of and 58% of the total block area respectively during peak
flood was analyzed from flood distributions in both space flood time (Table 3). This situation was followed by another
and time (Fig. 4). The study shows how the area was 2 blocks namely Puri Sadar and Delang where the flood
inundated over the time periods. It also shows the rate of affected area was about 43% and 40% of the total block
recession over the time. It may be noted from this figure area.
that the peak flood was occurred on 11 September. Floods become havoc when the flood stays for a long
Furthermore, statistical analysis was done using inun- duration. Using the four flood maps, which were prepared
dated maps and village map for showing the flood-affected from RADARSAT images of 04, 11, 13 and 20 September,
villages as percentage of the flooded area of village area. a flood duration map was prepared to know the flood
Table 2 depicted that on 11 September, the number of duration of the study area (Fig. 5). Table 4, which was
affected villages were 293 where the flooded area was based on Fig. 5, depicted that the 36626 hectares area,
more than 80% of the village area and flood affected area which was 35.57% of the highest flooded area, was inun-
was 39446 hectare, which was 38.31% of total affected dated for 16 days. The table also highlighted that about

Table-1: Flood Affected Information’ 2003- Puri District

Date No. of Affected Villages Area Affected (in hectare) % of total Area
September’ 04' 2003 902 89918 25.87
September’ 11' 2003 1000 102963 29.63
September’ 13' 2003 779 79161 22.78
September’ 20' 2003 466 44063 12.68

Fig.-4: Nature of Spatial Extent of Flood

(Super-imposed Flood Maps of 04, 11, 13, & 20 September 2003)

AJG_3042_Rahman 5 19/04/2006, 9:46 AM

 Flood Inundation Mapping and Damage Assessment Using Multi-temporal RADARSAT and IRS 1C LISS III Image

Table-2 Distribution of Flood Affected Villages (As % of Flooded Area of the Village Area)

Category September-04' 2003 September-11' 2003 September-13' 2003 September-20' 2003

No. of Flooded % of the No. of Flooded % of the No. of Flooded % of the No. of Flooded % of the
Villages Area Total Villages Area Total Villages Area Total Villages Area Total
(in Hec.) Flooded (in Hec.) Flooded (in Hec.) Flooded (in Hec.) Flooded
Area Area Area Area
<20% 121 3644 4.05 127 4163 4.04 126 3725 4.71 111 3429 7.78
20-40% 198 11453 12.74 181 11219 10.90 154 9998 12.63 110 6808 15.45
41-60% 170 15703 17.46 166 14583 14.16 149 13114 16.57 102 8205 18.62
61-80% 197 26770 29.77 233 33552 32.59 169 24701 31.20 80 13894 31.53
>80% 216 32348 35.97 293 39446 38.31 181 27623 34.89 63 11727 26.61
Total 902 89918 100.00 1000 102963 100.00 779 79161 100.00 466 44063 100.00

Table-3: Block Wise Flood Affected Area

Block Name Total Area September’ 04 September’ 11 September’ 13 September’’ 20

in Hectare Flood % of Flood % of Flood % of Flood % of
Affected Total Affected Total Affected Total Area Affected Total Area
Area Area Area in Area Area in Area in Hec
in Hec Hec Hec
Astarang 16444 3938 23.95 5409 32.89 4143 25.19 949 5.77
Bramhagiri 32768 14407 43.97 19157 58.46 13916 42.47 11042 33.70
Delang 20907 9003 43.06 8384 40.10 5584 26.71 2122 10.15
Gop 37975 7656 20.16 9839 25.91 7900 20.80 1476 3.89
Kakatpur 15806 798 5.05 3416 21.61 1951 12.34 233 1.47
Kanas 22552 15899 70.50 16618 73.69 15524 68.84 10995 48.75
Krishnaprasad 105450 14911 14.14 14123 13.39 13798 13.08 12629 11.98
Nimapara 30176 7796 25.84 7982 26.45 4378 14.51 365 1.21
Pipli 20878 1681 8.05 1566 7.50 1058 5.07 379 1.82
Puri Sadar 27586 9677 35.08 11730 42.52 7710 27.95 2833 10.27
Satyabadi 17038 4725 27.73 5282 31.00 3849 22.59 1460 8.57
Total 347580 89918 25.87 102963 29.62 79161 22.77 44063 12.68

29223 hectares or 28.38% and 37114 hectares or 36.05% was only 3.36% of the total area. Other than these, water
of the flood-affected area was inundated for 9 days and bodies occupied an area of 84292.73 hectares, which was
7 days respectively. So it may be said that the flood-2003 24.25% of the study area while sandy and urban area
become catastrophic due to long duration of flood in the occupied only 1.30% and 0.23% of the total area respec-
study area. tively.

5.2 Land Use/ Land Cover Mapping 5.3 Flood Damage Assessment
For predicting the flood damage assessment, a pre- It was mentioned earlier that flood inundation layers
flood land use/land cover map was prepared using IRS 1C which were extracted from RADARSAT data, intersected/
LISS III data. The scenario of land use/land cover of the crossed with land use/land cover map for damage assess-
study area is shown in Fig. 6 and Table 5. The table depicted ment. The derived result is shown in Table 6. The table
that nearly 51% of the study area was under crop cultiva- shows that an area of 59941.60 hectares of crops land
tion and fallow land occupied another 20.36% of the area. was affected on 4 September while it was increased to
Forest/plantation area occupied 11690.47 hectares, which 65991.39 hectares on 11 September, which was 37.60% of

AJG_3042_Rahman 6 19/04/2006, 9:46 AM

 Asian Journal of Geoinformatics, Vol.6, No.2 (2006)

Fig.-5: Map Showing Flood Duration (04 September to 20 September 2003)

Table-4: Flood Duration of the Study Area

Flood Duration Area in Hectares % of the Highest Flood Area

7 Days 37114 36.05
9 Days 29223 28.38
16 Days (severe Flood) 36626 35.57
Total 102963 100.00

Fig.-6: Spatial Extent of Land use/land cover

(Derived by supervised classification using IRS 1C LISS III band 3, 2,1)

AJG_3042_Rahman 7 19/04/2006, 9:46 AM

 Flood Inundation Mapping and Damage Assessment Using Multi-temporal RADARSAT and IRS 1C LISS III Image

Table-5: Spatial Extent of Different Landuse/ Land cover Bramhagiri and 8537.34 hectares or 48.12% of the total crops
area in Puri Sadar block on 11 September.
Land use Area in Hectare % of the total Area
In this study an attempt was also made to calculate the
Crops 175503.89 50.50 population density in the flood affected area. For this reason
Fallow Land 70746.48 20.36 the population density map was prepared using the village
Forest/Plantation 11690.47 3.36 population data collected form field authorities. This popu-
lation density map was crossed with flood maps of different
Sandy 4525.14 1.30
dates. Table 9 depicted that that about 40% of the total
Urban 785.13 0.23 flood affected area was under low population density
Water Bodies 84292.73 24.25 while about 35% and 15% of the total flood affected area
Total 347580.84 100 was under medium and high population density respectively
during the peak flood time. About 9% of the flood affected
area was in very high population density during peak flood
the total crops area. Fig. 7 shows the crop affected/damaged time. This analysis noticed that more than 55% of the
areas of the study area during different flood events. Other flood affected area was under medium to very high popu-
flood-affected areas were fallow land, sandy area and lation habited area. Due to high population habitation, in
forest/plantation. The affected area of these types of land these areas the damage caused by flood will be severe.
use/land cover was given in Table 6. Further Table 9 shows that the number of flood-affected
Moreover, severely flood affected area, which was the population was 500297 on 11 September.
common flooded area in all the four flood maps was Beside these, to identify the affected road due to flood
extracted. This severely flood affected area was crossed in Puri district, a road map was intersected with flood
with the land use/land cover map to know the affected/ maps of different time periods. For this, ArcView 3.2 was
damaged area by land use/land cover. Table7 shows that used and clipping operation was applied using Geoprocess-
464 villages were under severe flood condition and the ing Wizard. Table 4.10 depicted that on 4 September the
affected area was 36626 hectares including 26325.37 total length of the flood affected road was 860.28km,
hectares of cropped land. Due to the long duration (16 days) which was 30.53% of the total road length in the study
of flood, the crops of this area were affected severely. area. The length of the flood affected road further increased
Block-wise crops damaged area was also calculated on 11 September to 964.27km, which was 34.22% of the
for this study, which is shown in Table 8. From this table it total road length. From the 13 September flood situation
may be said that in Kanas and Bramhagiri block of Puri was improved and flood affected area was reducing so the
district, crops area was affected/damaged severely. It length of affected road was also reducing from 13 September.
noticed that the crop-damaged area was 6640.86 hectares Further Table 4.10 shows that Road with Tar (semi-metal)
or 83.26% in Kanas, 17137.48 hectares or 69.70% in and Kutcha (un-metal) road was affected more due to

Fig.-7: Crop Affected Areas (During Flood 04-20 September 2003)

AJG_3042_Rahman 8 19/04/2006, 9:46 AM

 Asian Journal of Geoinformatics, Vol.6, No.2 (2006)

Table-6: Flood Affected/Damage Area by Land use/Land cover

Land Use/ Sep’04 Sep’11 Sep’13 Sep’20

cover Area % of the Area % of the Area % of the Area % of the
(Hectare) total Crops (Hectare) total Crops (Hectare) total Crops (Hectare) total Crops
area area area area
Crop Land 59941.60 34.15 65991.39 37.60 52123.67 29.70 32710.26 18.64
Fallow Land 29976.74 - 36038.68 - 27205.44 - 11122.48 -
Sandy 486.04 - 1117.48 - 416.4 - 597.09 -
Forest/
Plantation 72.34 - 373.04 - 48.98 - 62.13 -

Table-7: Severe Flood Condition (16 Days Duration)

No. of Villages Area Affected Landuse/Landcover Affected/Damaged Area

In Hectare (in Hectare)
464 36626 Crop Land 26325.37 (Fully Damaged)
Fallow Land 10040.62
Forest/Plantation 23.75
Sandy Area 214.27

Table-8: Block Wise Crops Damage-September’ 11(Area in Hectare)

Land Use Block Name Total Crops Area Damaged Crop Area % of Total Crop Area
Crop Land Astarang 5388.63 1916.14 35.56
Crop Land Bramhagiri 24587.44 17137.48 69.70
Crop Land Delang 13596.51 4804.58 35.34
Crop Land Gop 18489.99 4339.53 23.47
Crop Land Kakatpur 8064.4 2527.65 31.34
Crop Land Kanas 7976.09 6640.86 83.26
Crop Land Krishnaprasad 20838.49 8880.68 42.62
Crop Land Nimapara 26509.44 6433.66 24.27
Crop Land Pipli 17175.2 552.69 3.22
Crop Land Puri Sadar 17743.07 8537.34 48.12
Crop Land Satyabadi 15038.6 4220.79 28.07

Table-9: Population Density in the Flood Affected Area and Affected Population

Population September’ 04 September’ 11 September’ 13 September’ 20

Density
Affected % of the Population Affected % of the Population Affected % of the Population Affected % of the Population
Area in Total Affected Area in Total Affected Area in Total Affected Area in Total Affected
Hectare Affected Hectare Affected Hectare Affected Hectare Affected
Area Area Area Area
Low 38945.06 43.11 58417 41779.42 40.40 62669 35755.32 44.89 53633 24292.81 54.76 36439
Medium 31227.25 34.57 140523 36322.48 35.12 163451 27015.79 33.92 121571 12714.12 28.66 57214
High 13108.76 14.51 98316 16009.23 15.48 120069 10956.86 13.76 82176 4738.58 10.68 35539
Very High 7048.26 7.80 136334 9299.89 8.99 154108 5923.27 7.44 126181 2620.65 5.91 87603
Total 89918.33 100 433590 102963.02 100 500297 79161.24 100 383561 44063.16 100 216795

AJG_3042_Rahman 9 19/04/2006, 9:46 AM

 Flood Inundation Mapping and Damage Assessment Using Multi-temporal RADARSAT and IRS 1C LISS III Image

Table-10: Flood Affect on Transport Network

Road Types Total September’ 04 September’ 11 September’ 13 20-Sep

Length
(In Km) Affected % of the Affected % of the Affected % of the Affected % of the
Length Total Length Total Length Total Length Total
Length Length Length Length
State Highway 205.79 44.71 21.72 56.22 27.32 33.48 16.27 6.87 3.34
Major District Road 152.27 50.63 33.25 52.74 34.63 40.49 26.59 22.63 14.86
Road with Tar 1629.06 447.59 27.47 527.69 32.39 359.34 22.06 159.07 9.76
Kutcha Road 771.50 317.36 41.13 327.62 42.46 291.82 37.82 176.41 22.86
Grand Total 2817.97 860.28 30.53 964.27 34.22 727.57 25.82 364.98 12.95

flood in the study area. It was depicted that more than 30% under medium to very high population habited area
and 40% of Road with Tar and Kutcha road was affected during the peak flood time. This analysis also shows
respectively due to flood on peak flood time (11 September). that the total number of flood affected population were
500297, which was 33.38% of the total population.
6. Conclusion
During the peak flood time the total length of the flood
affected road was about 964km, which was 34.22%
The following conclusions are drawn from the study. of the total road length of the study area. It was
observed that Road with Tar and Kutcha road was

Flood maps shows that 29.63% of the total geogra- affected more due to flood in the study area.
phical area was flooded during the flood-2003 and out On the basis of above findings, it is clear that the study
of 1710 villages of Puri district, 1000 villages were shows a simple and effective way to use of remote sensing
flood affected. The study also depicted that out of 1000 and geographical information system for creating flood
flood affected villages, in 293 villages flood-affected inundation map, time series maps of flood extent, monitor-
area was more than 80% of the village area. Further- ing areal changes of inundation and damage assessment.
more, out of 1000 flood affected villages, 464 villages RADARSAT images have potential and capable to moni-
were under severe flood condition due to long duration toring the flood event and easy to identify accurately the
of flood (16 days). flooded area. Duration of flooding can also be estimated

Block wise analysis of flood affected area shows that from the multiple coverage satellite data. Therefore, the
out of 11 blocks, 2 blocks namely Kanas and Bramhagiri satellite remote sensing technology is extremely useful in
was severely affected where flooded area was about monitoring the spatial extent and spreads of water during
74% and 58% of the total block area respectively the floods. The combination of microwave and optical
during the peak flood time. satellite data can be utilizes advantageously for flood

Flood duration analysis shows that 35.57% of the total inundation mapping and damage assessment.
flood affected area was inundated for more than two
weeks (16 days) while 28.38% and 36.05% of the flood Acknowledgements:
affected area was inundated for 9 days and 7 days
respectively. The author is grateful to Prof. Karl Harmsen, Director,

Flood damage analysis shows that about 37.60% of Center for Space Science and Technology Education in
the total cropped area (the main cropping season of Asia and the Pacific (CSSTEAP)-affiliated to the United
the area) was affected due to flood. Block wise crop Nations and Dr. P.S. Roy, Dean, Indian Institute of Remote
damage analysis indicated that in 3 blocks namely Sensing (IIRS) for supporting this work and providing the
Kanas, Bramhagiri and Puri Sadar crops were affected data for Puri District. He also thankful to Dr. K.H.V. Durga
severely. It was found that about 83% of the total Rao, Scientist, National Remote Sensing Agency (NRSA),
cropped area of Kanas block was affected which was Hyderabad, India for his guidance and suggestion. The author
followed by Bramhagiri (69.70% of the crop land) and also thankful to Mr. C. Jeganathan, Scientist, Geoinformatics
Puri Sadar (48.12% of the cropped area) Block. Division of IIRS for his kind support and help. Further the

Population density in the flood affected area noticed author thankfully acknowledges the contribution of Dr. S.K.
that more than 55% of the flood affected area was Saha, Coordinator of PGD Course, CSSTEAP, India.

10

AJG_3042_Rahman 10 19/04/2006, 9:46 AM

 Asian Journal of Geoinformatics, Vol.6, No.2 (2006)

References: space borne imaging radar and Landsat data. Photo-

grammetric Engineering & Remote Sensing, 53, 405-
Adam, S., Wiebe, J., Collins, M., and Pietroniro, A., 1998, 413.
Radarsat flood mapping in the Peace-Athabasca Delta, Lillesand, T.M. and Kiefer, R.W., 2002, Remote Sensing
Canada, Canadian Journal of Remote Sensing 24(1), and Image Interpretation, (New York: John Wiley &
69-79. Sons, Inc.)
Blyth, K., 1997, Floodnet: a telenetwork for acquisition, Profeti, G., and Macintosh, H., 1997, Flood management
processing and dissemination of Earth observation through Landsat TM and ERS SAR data: a case study.
data for monitoring and emergency management of Hydrological Processes, 11, 1397-1408.
floods. Hydrological Processes, 11, 1359-1375. Srivastava, Y.K., Doley, B., Pal, D.K., Das, R.K., Sudhakar,
Delmeire, S., 1997, Use of ERS-1 data for the extraction of S., Adiga, S., Venkatachary, K.V., and Srivastava, S.K.,
flooded areas. Hydrological Processes, 11, 1393-1396. 2000, High resolution remote rensing data & GIS
FAO/ESA, 1989, Principle of Radar Imagery, Fao Remote techniques in updation of infrastructure details for
Sensing Series No.46. flood damage assessment - a case study. Proceedings of
Gong, P.Y. Sheng, and Xiao, Q., 2001, Quantitative dyna- GIS Development. 2000.
mic flood monitoring with NOAA AVHRR, Interna- Tholey, N., Clandillon, S., and Fraipont, P. D., 1997, The
tional Journal of Remote Sensing. 22, No.-9, 1709- contribution of space borne SAR and optical data in
1724. monitoring flood events: examples in northern and
Imhoff, M. L., Vermillion, C., Story, M. H., Choudhury, A. southern France. Hydrological Processes, 11, 1409-
M., Gafoor, A., and Polcyn, F., 1987, Monsoon Flood 1413.
boundary delineation and damage assessment using

11

AJG_3042_Rahman 11 19/04/2006, 9:46 AM