Improved bottom topography in the Elbe estuary using wavelet and active contour methods on SAR images

Since launch of the ERS-2 Satellite in 1995, more than 40 SAR images of the intertidal areas of the River Elbe estuary have been received at Neustrelitz antenna station and processed at DFD in near real time. This area is morphodynamical very active, i.e. the bottom topography underlies a continuous change due to sediment transport, with strong impact on the main waterways of the Elbe. To investigate these morphodynamics, the ERS-2 SAR images are very useful, because of their all weather, day and night capability and because the land-water boundary-the shoreline-can be extracted using edge detection methods. The detected shorelines are transformed into isolines of bottom topography by combining them with available tide gauge information. Due to the speckle noise of the SAR image, edge detection from SAR is a demanding task. In this paper a new approach is used combining wavelet methods and active contour models. In a first step this approach takes advantage of Mallats (1992) wavelet algorithm acting as a full resolution noise filter and edge detector. The Mallat algorithm leads to a first guess full resolution land water segmentation. Based on this segmentation an active contour model starting from the wafer side is used as a second step, similar to the approach used by Mason and Davenport (1996). After the active contour model has mapped a connected shoreline, the derived line is combined with height, information from tide gauges and compared to bottom topography provided by the German Federal Waterway Engineering and Research Institute (BAW) to show changes due to morphodynamics.