Retrieval of soil moisture using a decomposition-based model and optical-thermal model with Sentinel-1 and Landsat-8 images

Water content of the soil has a significant role to play in the hydrological cycle and environmental processes. This study includes two phases: (1) approximation of soil moisture at the farm level based on optical-thermal images of Landsat-8 satellite and (2) retrieval of soil moisture by the dual-polarized basis decomposition model with the help of approximate soil moisture that is obtained by the optical-thermal model. In this research, two mechanisms, volume scattering and surface scattering, are considered. Furthermore, in order to model surface scattering, Bragg matrix has been used. The proposed radar model estimates soil moisture without using ground data, although few ground measurements have been used in the optical-thermal model. The Carlson triangular model has been used to approximate soil moisture using optical-thermal images. Three indices, normalized difference vegetation index (NDVI), normalized difference moisture index (NDMI), and moisture stress index (MSI), have been used in the optical-thermal model. Two ground soil moisture datasets are used in this study: (1) Cook Agronomy Farm (CAF) soil moisture data located in the United States and (2) real-time in situ soil monitoring for agriculture (RISMA) soil moisture data located in Canada. The radar model (base decomposition model) achieved a lowest root mean square error (RMSE) of 3.33% and a highest of 11.21%, showing strong accuracy in soil moisture retrieval. The optical-thermal model had a slightly higher minimum RMSE of 4.04% and a maximum of 9.68%. These results suggest that the radar model generally outperforms the optical-thermal model, making it more reliable for accurate soil moisture estimation in agricultural applications, which is crucial for optimizing irrigation and managing resources.