Improvement of TanDEM-X Water Mask by Exploiting the Acquisition Geometry

Abstract

In this paper we propose an enhancement of the TanDEM-X Water Mask Indicator (WAM). The proposed approach takes into account the geometric properties of SAR acquisition scenes, in order to precisely detect image areas affected by severe geometric distortions, namely Shadow and Layover. The presence of such effects can severely reduce the performance of the overall classification. In particular we concentrate upon the developed TanDEM-X WAM classification algorithm which shows misclassified areas in correspondence of high-relief terrain. The proposed approach aims at reducing this kind of misclassification errors, leading to an improvement of the TanDEM-X WAM performance. The impact of the input DEM accuracy is investigated as well, together with the potentials of modifying the detection algorithm itself. The long-term objective is to develop a reliable and efficient algorithm to provide high-resolution large-scale water masks from TanDEM-X interferometric data.