Method

The MSU-GRI CYGNSS soil moisture products are generated from the CYGNSS land observables with the inclusion of multiple remote sensing land geophysical (e.g., topography, land cover, and soil texture) data via the machine learning algorithm. The proof-of-concept work is first conducted over 18 International Soil Moisture Network (ISMN) sites in Southern U.S. ( Eroglu_RS19) and then expanded to available ISMN sites over the contiguous U.S (Senyurek_RS20a) The quasi-global CYGNSS soil moisture products are derived by applying the machine learning models for all valid CYGNSS land observables. For global products, two different models are utilized. The first model is constructed between collocated ISMN measurements and CYGNSS reflectivity and applied for global estimates (Senyurek_RS20b). The second model uses collocated SMAP soil moisture retrievals and CYGNSS observables (Lei_2021). Both approaches can deliver daily 9 km soil moisture products.