High resolution soil moisture estimation and evaluation from Earth observation

Soil moisture is an important component of the Earth system and plays an important role in land-atmosphere interactions. Remote sensing of soil moisture is of great scientific interest and the scientific community has made significant progress in soil moisture estimation using Earth observations. Currently, several operational coarse spatial resolution soil moisture datasets have been produced and widely used for various applications such as climate, hydrology, ecosystem and agriculture. Due to the strong demand for high spatial resolution soil moisture in regional applications, much effort has been recently devoted to the generation of high spatial resolution soil moisture from either Sentinel-1 observations or downscaling of existing coarse resolution soil moisture datasets. The aim of this study is to evaluate high spatial resolution soil moisture products derived from multisource satellite observations. First, the COSMOS-UK measured soil moisture was used to validate existing satellite-based soil moisture datasets including SMAP_9km, SMOS_1km, Sentinel-1, and Sentinel-1/SMAP combined products. Second, an approach based on triple collocation was applied to inter compare these satellite products in the absence of a reference dataset. Third, two merging schemes including a simple average and a triple collocation method were used to develop a combined satellite soil moisture product based on existing satellite soil moisture datasets. From the above analysis, it is found that merging all the above soil moisture data provides a better estimate of soil moisture than any of them alone. Therefore, we conclude that combining existing satellite-based soil moisture products has the potential to provide the best estimate of high spatial resolution soil moisture in the UK.