Improving Soil Moisture Spatio-Temporal Resolution Using Machine Learning Method

Abstract

Surface soil moisture (SM) plays an essential role in the water and energy balance between the land surface and the atmosphere. The published soil moisture products, having a low spatial resolution of 25–40 km, and low temporal resolution of 2–3 days, limits their applications at regional scale. In this study, the spatio-temporal resolution of Fengyun (FY) SM products was improved using a machine-learning model named the General Regression Neural Network (GRNN), with the help of selected six high spatial resolution parameters, including Land Surface Temperature (LST), Normalized Difference Vegetation Index (NDVI), albedo, Digital Elevation Model (DEM), Longitude (Lon) and Latitude (Lat) after gap-filled as input variables. An implements tested over the Tibetan Plateau (TP) showed that the spatio-temporal resolution of FY-3B SM was improved from 0.25° and 2–3 days to 0.05° and 1-day. The high spatio-temporal resolution SM can enhance our understanding of water-energy cycle under climate change.