A fused soil moisture product integrating linear fusion method and error characteristics approach

Soil moisture is critical for understanding the hydrological cycle and managing climatic extremes such as floods and droughts. This study evaluates the accuracy of satellite-derived soil moisture estimates from remote sensing satellite products Soil Moisture Active Passive (SMAP), Soil Moisture and Ocean Salinity (SMOS), and Advanced Microwave Scanning Radiometer-2 (AMSR-2) in the absence of ground observations. A selective fusion algorithm is introduced that enhances accuracy by fusing data only in regions where it improves the precision of soil moisture estimates, rather than applying a uniform fusion approach across all regions. Using Extended Triple Collocation (ETC) analysis, the error characteristics of individual satellite products are assessed across spatial, temporal, and meteorological aspects. Additionally, Mutual Information (MI) analysis with precipitation data quantifies the information content of each product to validate their effectiveness. Our results demonstrate that SMAP consistently exhibits the lowest error characteristics and highest information content. Notably, AMSR-2 shows superior performance in hot desert and semi-arid regions, highlighting the need for a spatially adaptive fused soil moisture product. An integrated fusion algorithm, guided by a decision map based on these analyses, optimizes the selection of soil moisture products across India’s diverse regions. This decision map revealed the highest spatial coverage by SMAP (62.5%), followed by AMSR-2 (25.7%) and SMOS (5.6%), with the remainder attributed to various combinations of products that provide superior soil moisture estimates in specific contexts. This modified fusion approach is implemented using a Linear Weight Fusion (LWF) technique, significantly enhancing the reliability of the resultant soil moisture product, particularly evident from its validation against data from three in-situ stations. The integrated approach not only refines the accuracy of satellite-derived soil moisture data but also provides a robust framework for enhancing hydrological model predictions, especially in drought monitoring.