Key points are not available for this paper at this time.
Soil moisture products play a pivotal role in monitoring and predicting droughts that affect crop yield, water supply, and landatmosphere interactions. The availability of various satellite-based soil moisture products allows for a comprehensive investigation of droughts on a large scale. However, limitations in their spatial sampling impact their suitability for regional applications. Accurately inferring sub-pixel heterogeneity is crucial for a representative understanding of regional dynamics and their implications for drought assessment across diverse landscapes. Furthermore, the shallow vertical support of satellite-based soil moisture products hinder the detection and quantification of droughts within the sub-surface. This study leverages multi-scale data fusion, aiming to replicate soil moisture extremes both regionally and in the sub-surface. We integrate ground-based Cosmic-Ray Neutron Sensing (CRNS) data, representing soil moisture within an extensive soil volume, with high-resolution Sentinel-1 data. Employing machine learning models that account for spatiotemporal autocorrelations, our objective is to generate gridded soil moisture data representing regional sub-surface dynamics. Tested in a German catchment, our approach tackles challenges associated with the scarcity of CRNS stations and the complexities of integrating multi-scale data. Our findings establish a foundation for monitoring regional droughts in the sub-surface across extensive areas.
Schmidt et al. (Fri,) studied this question.