COMPARISON OF TWO METHODS FOR SOIL MOISTURE MAPPING AT 1KM RESOLUTION FROM SENTINEL-1 AND MODIS SYNERGY

This paper presents two methodologies retrieving soil moisture from SAR remote sensing data. The study is based on Sentinel-1 data in the VV polarization, over a site in Urgell, Catalunya (Spain). By modeling the backscatter difference w.r.t. NDVI, the soil moisture corresponding to a specific NDVI value can be retrieved. The first algorithm is already developed in West Africa[1] from ERS scatterometer data to estimate soil water status. In this study, it is adapted to Sentinel-1 data and take into account the high repetitiveness of data in optimizing the inversion approach. Another new method is developed based on the backscatter difference between two adjacent days of Sentinel-1 data w.r.t. NDVI, with smaller vegetation change, the backscatter difference is more sensitive to soil moisture. The validation of the two methods is done with field data acquired in study site, with an rms error about 0.08 m(3)/m(3) for method 1 and 0.07 m(3)/m(3) for method 2 in volumetric moisture.