Remote sensing inversion of soil moisture and analysis of influencing factors in complex cropland environments in karst plateau mountainous areas of Southern China

Accurate soil moisture assessment is crucial for ecological and agricultural purposes. To investigate the soil moisture distribution and its influencing factors under the mountainous areas of the karst plateau, this study took the cultivated plots as a unit, combined with the Sentinel-1 data and the SAR-derived vegetation indices (RVI and CR) calibrated water-cloud model and inverted the soil moisture by PSO-SVM model after removing the vegetation disturbance, and analysed soil moisture distribution in different environments. The results showed that (1) the inversion accuracies of the models using the combination of RVI and VV backscattering coefficients were high; the R2 of the test set as a whole was 0.890, and the RMSE was 2.944%. (2) The trend and spatial distribution of soil moisture were closely correlated with climate, elevation, degree of soil desertification and soil type. (3) Inversion based on plot scale can more accurately reflect the overall moisture status of cropland.