A novel approach to improve spatial detail in modeled soil moisture through the integration of remote sensing data

In this work the possibilities of combining modelled (GEOtop, Hydrological model) and remotely sensed (ENVISAT ASAR WS) soil moisture content (SMC) values were investigated introducing a novel approach for data fusion on a product level. Data fusion was performed through the definition of a correction term for the modelled SMC dataset. For the determination of this term machine learning (Support Vector Regression) was used. As a reference dataset in-situ SMC measurements were considered. The benefit of the proposed method was successfully shown as R² between modelled and measured SMC values was improved from 0.11 to 0.61.