Multiple Landsat images with different inudation levels can be used to increase the accuracy of the SRTM dataset
Correcting SRTM by using lake gradients over time, Cambodia
Supervisors: Ton Hoitink (WUR), Rolf Aalto (University of Exeter).
This research focuses on a new method to increase the accuracy of the Shuttle Radar Topography Mission (SRTM) of a relative flat floodplain which is located in Cambodia along the Mekong river. This floodplain is flooding every year. The amount of water is depending on the discharge of the Mekong River. This results in different inundation levels of the lakes in the floodplain over time. The height around a lake which is not flowing should be more or less the same; it is hard to disagree with the observed water level in the Landsat Images. This together with a relative flat floodplain makes it is possible to obtain new height values for the SRTM. Therefore a model is created in ArcGIS 10.1, which can identify lakes in the floodplain over time. Contour lines are created at the border of these lakes. The contour lines are converted to points,(xy data) and height is extracted from these points. Per image the trend toolset is applied to the height values. These different point clouds are combined into one dataset. Since the floodplain almost floods completely in the most severe flooding there are points distributed over the whole floodplain. There are still some gaps in the data because only 38 different inundation levels are used. These gaps will be filled up using the Inverse Weighting Toolset and the higher places where no water has been will be added. These steps result in a more accurate height model for the SRTM. A reason why it is important to increase the accuracy is that a better estimation of the amount water in the floodplain can be made.