Abstract
The aim of this work is to extend radarclinometry technique to regions where the two current assumptions, (i) Lambertian backscattering and (ii) homogeneous areas, are not possible. For this, we replace the traditional Lambertian model by a backscattering diagram provided by the integral equation model (IEM). Then, to take into account the roughness heterogeneity of the region, we introduce a classification process. In the global algorithm for DEM (digital elevation model), the first iteration classification is determined according to minimum distance, and then the derived altitude is also used to constrain classification. The process is iterative. Validation is performed on four sites in Israel and Tunisia, which represent different cases of arid or semi‐arid regions. Both qualitative and quantitative results are satisfactory. Limits of the method are stressed by the sand dune case.
Acknowledgments
The data were acquired in the framework of the FLAUBERT (FLood in Arid Unit By Earth Remote Technics) project, supported by the European Community DGXII. We also thank Eilon Adar (Ben Gurion University) for ground data and the referees for their useful comments.