A merged SMAP – Sentinel-1 soil moisture product using Artificial Neural Networks: a case study in Central Italy

This study aims at merging SMAP and Sentinel-1 (S-1) data for obtaining a surface soil moisture (SMC) product improved in accuracy, spatial and temporal resolution that can be used for hydrological modelling in small basins. A method based on Artificial Neural Networks has been developed and validated in a test area in central Italy. All the S-1 images available on the area between 2014 and 2017 have been considered for the analysis, along with the corresponding SMAP acquisitions. Distributed SMC values, to be used as reference for implementing and validating the algorithm, have been derived from the available in-situ data by using the well-assessed Soil Water Balance hydrological model (SWBM). The research is still ongoing; however, some preliminary results show that the merged ANN SMC product is improved in resolution and accuracy with respect to the SMC obtainable from a single sensor.The ANN SMC was successfully assimilated in the MISC hydrological model for improving the model predictions in small and medium basins.