Applying a machine learning method to obtain long time and spatio-temporal continuous soil moisture over the Tibetan Plateau

Abstract

Soil moisture is a key variable in the exchange of water and energy between the land surface and the atmosphere. Long time series of and spatio-temporal continuous soil moisture is of great importance to meteorological and hydrological applications, such as weather forecasting, global change and drought monitoring. In this study, the Essential Climate Variable (ECV) soil moisture product of the Tibetan Plateau (TP) from 2002 to 2015 was reconstructed using the General Regression Neural Network (GRNN) based on reconstructed MODIS products, i.e., LST, NDVI, and Albedo. Results show that the ECV soil moisture could be well reconstructed with R2 higher than 0.71, RMSE less than 0.05 cm3 cm-3 and absolute Bias less than 0.03 cm3 cm-3 for both grids of 0.25°×0.25° and 1°×1°, compared with the in-situ measurements in 2012 over the TP. The reconstructed long time series of and spatio-temporal continuous soil moisture could be valuable in hydrometeorological studies of the TP.