Soil moisture is a key variable that affects ecosystem carbon and water cycles and that can directly affect climate change. Remote sensing is the best way to obtain global soil moisture data. Currently, soil moisture remote sensing products have coarse spatial resolution, which limits their application in agriculture, the ecological environment, and urban planning. Soil moisture downscaling methods rely mainly on optical data. Affected by weather, the spatial discontinuity of optical data has a greater impact on the downscaling results. The synthetic aperture radar (SAR) backscatter coefficient is strongly correlated with soil moisture. This study was based on the Google Earth Engine (GEE) platform, which integrated Moderate-Resolution Imaging Spectroradiometer (MODIS) optical and SAR backscattering coefficients and used machine learning methods to downscale the soil moisture product, reducing the original soil moisture with a resolution of 10 km to 1 km and 100 m. The downscaling results were verified using in situ observation data from the Shandian River and Wudaoliang. The results show that in the two study areas, the downscaling results after adding SAR backscattering coefficients are better than before. In the Shandian River, the R increases from 0.28 to 0.42. In Wudaoliang, the R value increases from 0.54 to 0.70. The RMSE value is 0.03 (cm³/cm³). The downscaled soil moisture products play an important role in water resource management, natural disaster monitoring, ecological and environmental protection, and other fields. In the monitoring and management of natural disasters, such as droughts and floods, it can provide key information support for decision-makers and help formulate more effective emergency response plans. During droughts, affected areas can be identified in a timely manner, and the allocation and scheduling of water resources can be optimized, thereby reducing agricultural losses.

Soil moisture is a key variable that affects ecosystem carbon and water cycles and that can directly affect climate change. Remote sensing is the best way to obtain global soil moisture data. Currently, soil moisture remote sensing products have coarse spatial resolution, which limits their application in agriculture, the ecological environment, and urban planning. Soil moisture downscaling methods rely mainly on optical data. Affected by weather, the spatial discontinuity of optical data has a greater impact on the downscaling results. The synthetic aperture radar (SAR) backscatter coefficient is strongly correlated with soil moisture. This study was based on the Google Earth Engine (GEE) platform, integrates Moderate-Resolution Imaging Spectroradiometer (MODIS) optical and SAR backscattering coefficients, uses machine learning methods to downscale the soil moisture product, reducing the original soil moisture with a resolution of 10 km to 1 km and 100 m. The downscaling results were verified using in situ soil moisture observation data from the Shandian River and Wudaoliang. The results show that in the two study areas, the downscaling results after adding SAR backscattering coefficients are better than before. In Shandian River, the R increases from 0.28 to 0.42. In Wudaoliang, the R increases from 0.54 to 0.70. The RMSEs are 0.03(cm3/cm3). The downscaled soil moisture products play an important role in water resource management, natural disaster monitoring, ecological and environmental protection, and other fields. In the monitoring and management of natural disasters such as droughts and floods, it can provide key information support for decision makers and help formulate more effective emergency response plans. During droughts, affected areas can be identified in a timely manner, and the allocation and scheduling of water resources can be optimized, thereby reducing agricultural losses.