An Improved Method for Water Body Removal in Spaceborne GNSS-R Soil Moisture Retrieval

The global soil moisture (SM) retrievals by the spaceborne Global Navigation Satellite System-Reflectometry (GNSS-R) are significantly influenced by the presence of water bodies. The traditional method is to build a grid based on the location of satellite sampling points and determine the presence or absence of water bodies. In this article, we propose a water body removal method for global spaceborne GNSS-R SM retrievals that combines water bodies and buffers derived from the marginal areas around water bodies as mask data, thus achieving accurate removal of the water body and avoiding margin effects. To verify the effectiveness of the proposed method, the Cyclone GNSS (CYGNSS) data with two different spatial resolutions (36 and 3 km) were used for SM retrieval, and the Soil Moisture Active and Passive (SMAP) Radiometer SM as well as the International Soil Moisture Network (ISMN) were used as references. Results show that the correlation coefficient ( $R$ ) and root-mean-square error (RMSE) of the 36-km grid are 0.50 and 0.057 cm3/cm3, respectively, while the $R$ and RMSE of the 3-km grid are 0.68 and 0.041cm3/cm3, respectively. Such performances are better than the traditional method. Moreover, the method proposed in this article preserves more grids. Take the 3-km spatial resolution, for example, it preserves 2.2-fold grids more than the traditional water body removal method. In the comparison with SMAP SM, the overall improvement of RMSE by using the water body removal method proposed in this article is 16.3% (8.2% for the traditional method). In the in situ validation, the overall improvement of RMSE is 19.4% (-1.2% for the traditional method). Therefore, in the future high spatial resolution SM retrieval, the water body removal method proposed in this article can preserve the maximum area and effectively eliminate the influence of water bodies on SM retrieval.