|
|
|
Water Resources
|
Detection of Coast Lines of Tonle Sap Lake in Flood Season using
JERS-1 Data for Water Volume Estimation
Detection of Coast Lines in Flood Season
It would be easy to detect water body with optical or thermal bands it its is exposed without any obstacle. However in case of the tonle Sap Lake area, the water boundaries are hidden with tropical forests, which results in difficulty to detect the coast lines with optical bands.
On the other hand, SAR data will have a stronge backscattering when tropical forests are submerged, due to the effect of corner reflection. A simple threshold with the value of 40 in the JERS-1 SAR band was used to automatically detect the coast lines that are submerged in tropical forest after several trials and errors.
JERS-1-SAR data as well as ERS-1 SAR data were used to detect the coastlines partially or completely depending on the data availability at different water levels during the flood season. From the leveling survey done by the first author, a linear interpolation was applied to estimate one meter contour of coast lines as shown in Figure 4.
Figure 4 Estimated Contour Line Map of the Tonle Sap up to the Level of 10 Meters with JERS-1 SAR Data
Estimation of Water Volume
The existing topographic maps of 1:50,000 at one meter interval were produced in 1964 by the Philippines Government. Because of changes in land use and sedimentation as well as of surveying errors, the contour maps would be not fully reliable. In 1964 by the Philippines Government. Because of changes in land use and sedimentation as well as of surveying errors, the contour maps would be not fully reliable. In addition, there was not water depth available under the minimum from topo-maps.
From the ground survey to check the accuracy or justification of contour lines, we found that the contour lines obtained from the remote sensing data were much reliable. Table 1 shows a comparison of the water surface (S) and water volume (V) between the results from satellite data and topographic map.
Table Comparison of Perimeter, Water Surface Area and Water Volume between the Result from Satelite imagery and Existing Topographic Map.
| E1 |
Satellite Data |
Topographic Map |
Difference |
| (m) |
S (Km2) |
V (Km3) |
S (Km2) |
V (Km3) |
ÑS (Km2) |
ÑV (Km3) |
| 1 |
2.504 |
17,624 |
2,504* |
1,624* |
0 |
0 |
| 2 |
3,602 |
4,588 |
3,367 |
4,733 |
-235 |
145 |
| 3 |
4,410 |
8,467 |
4,558 |
8,907 |
148 |
440 |
| 4 |
5.237 |
13,133 |
5,695 |
14,105 |
458 |
975 |
| 5 |
6,532 |
18,724 |
7,078 |
20,639 |
546 |
1,915 |
| 6 |
7,932 |
25,657 |
8,281 |
28,425 |
298 |
2,768 |
| 7 |
9,201 |
34,049 |
9,367 |
37,341 |
166 |
3,292 |
| 8 |
10,276 |
43,477 |
10,458 |
47,357 |
182 |
3,877 |
| 9 |
11,743 |
54,156 |
11,458 |
58,453 |
-153 |
4,297 |
| 10 |
14,573 |
66,740 |
12,859 |
70,753 |
-17,14 |
4,013 |
*Not available in topographic map, but the same amount of satellite data was given.
**The percentage difference of water volume between the maximum water level in 5.6%
Discussion about Accuracy
As the total surface are of the Tonle Sap Lake ranges from 2,504 square kilometers (dry season) to 14,573 square kilometers (maximum flood level), while the perimeter from 465 kilometers (dry season) to 2,104 kilometers (maximum flood level), one or two pixel size error of 18 meter or 36 meters for JERS-1 SAR data that will be expected to give less influence to the accuracy of the water surface area. Suppose that 18 meters or 36 meters multiplied by the perimeter are the error of the surface area, the relative accuracy range from 0.33 percent to 0.66 percent (dry season) and from 0.26 percent to 0.52 percent (flood season) to the total surface area.
Height accuracy of interpolation of DEM will be around one third of contour line, that is about 0.3 meters, which will affect the error of the water volume more than the water surface area. As a whole, the accuracy of the water volume would be about 80 to 90 percent, though we need further study for accuracy check with reference survey.
Conclusion
Detection of coast lines of the Tonle Sap Lake in flood season with JERS-1 SAR data was successfully achieved with the results of water surface area and water volume, that was expected more reliable than the ones from the existing topo-maps.
Acknowledgement
The authors extend sincere thanks to National Space Development Agency of Japan (NASDA) for having provided JERS-1 SAR Data.
|
|
|
|
|
|
|
