Parametrization of a Dielectric Mixture Model to Retrieve Soil Moisture at Field Scale Using Sentinel-1 Data and in Situ Soil Moisture Measurements

In the framework of the H2020 APOLLO project a significant focus has been put on the use of high resolution satellite based products to serve small farmers in their agricultural practices. This paper presents first results achieved by using Sentinel-1 Synthetic Aperture Radar (SAR) data as input to a semi-empirical soil moisture (SM) retrieval model based on an algorithm originally developed for ERS SAR data and successively modified to handle ENVISAT ASAR acquisitions. In this work the model has been adapted to Sentinel-1 images and calibrated by using ground measurements taken in two test sites characterized by bare soil and cotton cultivation, aiming at testing its capability to represent the SM behavior at different stages of the vegetation cycle. The model performance has been assessed through the correlation R and root mean squared error (RMSE) between in situ and satellite retrieved SM data. Very good results have been achieved for bare soil (R>0.8, RMSE<;0.04 m
³
m
^-3
); however, the model performed worse in the cotton fields (R<;0.6, RMSE>0.08 m
³
m
^-3
).