Datasets

v1.0a

The machine learning model is trained over ISMN sites between in-situ soil moisture measurements and CYGNSS observables in contiguous U.S. and then applied for global soil moisture estimates.

v1.0b

The model is constructed between collocated SMAP soil moisture retrievals and CYGNSS observables and then applied for all available CYGNSS data.

Both v1.0a and v1.0b use the Random Forest machine learning algorithm and are generated from the CYGNSS Level-1 data (version 2.1). The products are available in netCDF-4 format with each file containing the daily soil moisture estimates at 9 km. Quality flags are provided separately for each product and users should use caution when interpreting the data.

v1.0c

The model is constructed between collocated SMAP soil moisture retrievals and CYGNSS DDMs and then applied for all available CYGNSS data.