Segmentation of coherence maps for flood damage assessment

Multitemporal coherence map has been successfully used for various types of surface characterized by the same level of interferometric correlation. This paper examines the possible benefits of this source of information to detect the damages caused by the flooding of rivers and lakes. As test case the Yangtze river flooding of summer 1998 was chosen. The interferometric coherence map was obtained using ERS tandem mission images during the flood and a pair dating December 1995. The low level of water surface coherence allows to segment the flooded areas in an easy way with respect to the method which exploits the backscatter intensity. The approach proposed is based on fuzzy connectivity concepts of the flooded areas with reference to the original river bed.