Machine learning-based global soil moisture estimation using GNSS-R

Retrieval of soil moisture (SM) content is essential for many agricultural and hydrological studies and applications. Remotely sensed SM estimations in high spatial and temporal resolution are a vital requirement in many global studies. Global Navigation Satellite System (GNSS) Reflectometry (GNSS-R), one of the signals of opportunity (SoOp) techniques, has emerged in recent years as a new remote sensing method for SM retrieval in high spatio-temporal resolution. This paper summarizes our studies as a solution to high resolution SM retrieval on a global scale for agroecosystems. We have developed a machine learning (ML) framework to derive a quasi-global soil moisture (SM) product by using spaceborne GNSS-R observations provided by NASA's Cyclone GNSS (CYGNSS) constellation alongside remotely sensed ancillary geophysical data. The learning model is trained using in-situ SM data from the International Soil Moisture Network (ISMN) sites. The produced daily SM retrievals within the CYGNSS spatial coverage are independently compared with the Soil Moisture Active Passive (SMAP) mission's enhanced SM products at a resolution of 9 km x 9 km to evaluate the performance of the model.