Estimation of Surface Soil Moisture Based on Improved Multi-index Models and Surface Energy Balance System

Soil moisture is a vital resource that plays a critical role in arid and semi-arid areas. In the present study, a new approach was adopted to estimate surface soil moisture based on multi-index models using reflective and thermal indices as well as surface energy balance system–Iran (SEBS–Iran) in pastures and farmlands in Qom province, Iran in 2016–2017. To select the best model based on remote sensing (RS) indices, 12 models were designed and after analysis, the best ones were selected. Afterward, the results of the SEBS–Iran algorithm and the improved multi-index model [normalized multi-band drought index (NMDI), normalized difference vegetation index (NDVI), land surface temperature (LST) and the temperature vegetation dryness index (TVDI)] were calibrated with field data in the two studied fields (pastures and farmlands). The findings indicated that the multi-index model NMDI–TDVI–LST–NDVI (R = 0.95) and SEBS–Iran (R = 0.93) both had significant correlations with measured soil moisture. Regarding both models in farmlands and pastures, the SEBS–Iran regression model was closer to the line of fit, and R2 in the two fields was 0.95 and 0.96, respectively. Compared to SEBS–Iran, the multi-index model showed lower coefficient of determination in pastures (0.71) due to the higher accuracy of SEBS–Iran in areas with lower vegetation density. Generally, both methods were found to be suitable for soil moisture estimation. The multi-index model can be used to estimate soil moisture in densely vegetated areas on a large scale due to its simplicity and good accuracy. Moreover, the highly accurate SEBS–Iran model can be used even in sparsely vegetated areas.