Land Use Analysis Using a Compact Parametrization of Multi-Temporal SAR Data

The present work presents a novel and robust method that exploits multi-temporal Sentinel-1 SAR data for land use classification. The method is built on an interferometric stack of multi-temporal Sentinel-1 SAR data and combines the intensity of backscattering, the short time interferometric coherence and the decorrelation time of scatterers. Promising preliminary results of land use classification and river corridor coverage are presented. The test case is located in the flat area of the Padana plains in west northern part of Italy. The proposed method is a novel, robust and all-conditions tool for improving unsupervised land use classification at large scale. In the future, this technique will be integrated with optical information for testing the potentiality offered to the classification of land-uses and river features.