Comparison with RS Imagery
In order to compare the flood maps obtained by processing of the SPOT imagery and the DEM, these two products have been overlayed and the differences and similarities were analyzed.
Figure 1.6 shows, as an example, the comparison of the flooded/non-flooded areas on the image and the DEM for a flood level of 3.00 m.
Figure 1.6 Comparison between flooded areas on image and DEM (flood level : 3.00 m)
Table 1.1 gives the statistics on the differences and similarities. From this table it appears that the percentages of flooded area and non-flooded area obtained by slicing the image and b using the DEM with a water level of 3.00 m are the same (83% and 17% respectively). The flood map derived from the DEM with a water level of 3.50 m gives higher percentage of flooded area. Although the differences were within an acceptable range, there are inconsistencies if one looks at the values of Table 1.2
For 86% (water level: 3.00m) and 88% (water level: 3.50 m) the results of the DEM operations matched the satellite image flood map. However, about 41% of the non-flooded area on the image was shown as flooded on the DEM (water level: 3.00m). For the water level of 3.50 m. value was 56%.
Table 1.1 Comparison of flooded and non-flooded areas on the image and DEM (assumed water levels of 3.00 and 3.50 m).
| |
Image |
Water level: 3.00m |
Water level: 3.50 m |
|
Flooded |
83 |
83 |
90 |
|
Non-flooded |
17 |
17 |
10 |
Table 1.2 Result of overlaying the flood map obtained from the image and the DEM (assumed water levels of 3.00 and 3.50m.)
| |
Water level 3.00 m |
|
Water level 3.50 m |
|
Index |
% |
Index |
% |
|
1 |
10 |
1 |
7 |
|
2 |
7 |
2 |
9 |
|
3 |
7 |
3 |
3 |
|
4 |
76 |
4 |
81 |
key to index:
1 Non-flooded on both image and DFM
2 Flooded on DEM, non-flooded on image
3 Non-flooded on DEM, flooded on image
4 Flooded on both image and DEM
A number of reasons may account for the differences in the results of the different methods. Some soils might have been very moist and had the same reflection as shallow water; the flood level at the time of the acquisition of the image was not known, and/or the assumption that the water surface was level might not have been valid. Probably the most important reason was the quality of the DEM. Firstly the map was 20 years old and as the population in the area had doubled in the meantime, the development of houses along the ridges mould have influenced the topography. Secondly the contours on the map were constructed by interpolating between the points of a grid for which the elevation was measured. The survey was done along lines, and used more or less equal distances between points. The detailed geomorphology was not considered. Consequently the DEM was more inaccurate along the ridges. Thirdly, the addition of 1 m of extra height to the ridges might not have been appropriate in all cases.
Calculation of The Area, Depth and Volume of The Flood Waters
Once the flood maps had been created, the areas could be calculated. The number of flooded raster cells multiplied by the area of 1 cell gave the total flooded area.
For determining the depth of the flood water the DEM was uses. Subtracting the elevation value for each flooded cell from a specified flood level gave the flood depth. The resultant map could be classified to show the deeply, moderately and shallowly flooded lands. Figure 1.7 shows the flood depth map of the flood with a flood level of 3.00 m.
Figure 1.7 Flood depth map (flood level : 3.00 m)
If an area is flooded and the water has to be evacuated, either by pumping or gravity, it is important to know the flood water volume. This volume determines the pump or the spill capacity if it is necessary to drain the water within a limited time period, required to avoid more damage to the standing crop or because a new crop has to be planted for the next season. Having a flood depth map the volume can be calculated by multiplying the area of each cell with the corresponding depth value and then aggregating these volumes per cell for the tow flood levels.
The Effects on Depth, Area and Volume of The Flood Waters Given Different Flooding Scenarios
For the management of flood-prone areas it is important to assess the expected consequences of possible flooding events. The effects of different storm water
Table 1.3 Flood statistics for a flood level of 3.00 m and 3.50 m.
|
Flood level (m) |
Area Flooded (million ha) |
Volume of flood waters (million m3) |
|
3.00 |
2.89 |
1634799 |
|
3.50 |
3.16 |
2443454 |
levels are of interest, as well as the effects of a dike failure at different locations. Does dike failure have the same consequences wherever it my occur or would its occurrence at one point lead to more severe damage than its occurrence at another point? In other words, does it matter where the dike collapses?
For five different locations the effects of flooding in terms of depth, aea and volume were evaluated for different storm water levels. Figure 1.8 shows the assumed locations of the dike failure. Location 1 is the actual location of the dike failure on 10 October 1988.
Figure 1.8 Assumed locations of a dike failure.
Table 1.4 shows the results of the analysis. It reveals that below a storm water level of 3.75 m it does matter where the dike fails. Above this water level all five situations leas to the same effect.
Table 1.4 Volume of flood waters for 5 assumed locations of dike failure
|
Flood level (m2) |
Volume of flood waters (million m3) |
|
Locations |
|
1 |
2 |
3 |
4 |
5 |
|
200 |
0 |
0 |
0 |
0 |
0 |
|
225 |
0.38 |
0 |
0 |
0 |
0 |
|
250 |
5.73 |
0 |
5.73 |
0 |
0 |
|
275 |
11.06 |
0 |
11.06 |
11.06 |
0 |
|
300 |
17.94 |
17.94 |
17.94 |
17.94 |
0 |
|
325 |
25,55 |
25.55 |
25.55 |
25.55 |
0 |
|
350 |
33.39 |
33.39 |
33.39 |
33.39 |
0 |
|
375 |
41.35 |
41.35 |
41.35 |
41.35 |
41.35 |
References
- Gorte, B. et, 1990. Interpolation between insolines based on the Bagefors distance transform. ITC Journal 1990-3, ITC, Enschede, The Netherlands.
- Meijerink, A.M.J., C.R. Valenzuela & A. Stewart (eds.) 1988. ILWIS, the Intergrated Land and Watershed Management Information System. ITC Publication No 7, Enschede, The Netherlands.
- Rahman, A.K.M., 1992. Use of FIS, Remote Sensing and Models for flood studies in Bangladesh. An analytical study in a flood prone polder in Bangladesh. Unpublished MSc Thesis, ITC. 94P.
- Survey of Bangladesh. 1978. Water Development Maps 1:15,840, No 79 (I-11) and 79 (I-11)/7
- Wagner, Th. W., 1989. Preparing for floodplain mapping and flood monitoring with Remote Sensing and GIS. Report of the workshop on remote sensing for floodplain mapping and flood monitoring, Dhka, Bangladesh.
- Worldbank, 1989. Bangladesh Action Plan for Flood Control.
- Wu, Bingfang & Xia, Fuxiang, 1990. Flood damage evaluation system design for a pilot area on Bangladesh floodplain fusing remote sensing and GIS. European Conference and GIS.
