Comparison of four models to determine surface soil moisture from C-band radar imagery in a sparsely vegetated semiarid landscape

[1] Four approaches for deriving estimates of near-surface soil moisture from radar imagery in a semiarid, sparsely vegetated rangeland were evaluated against in situ measurements of soil moisture. The approaches were based on empirical, physical, semiempirical, and image difference techniques. The empirical approach involved simple linear regression of radar backscatter on soil moisture, while the integral equation method (IEM) model was used in both the physical and semiempirical approaches. The image difference or delta index approach is a new technique presented here for the first time. In all cases, spatial averaging to the watershed scale improved agreement with observed soil moisture. In the empirical approach, variation in radar backscatter explained 85% of the variation in observed soil moisture at the watershed scale. For the physical and best semiempirical adjustment to the physical model, the root-mean-square errors (RMSE) between modeled and observed soil moisture were 0.13 and 0.04, respectively. Practical limitations to obtaining surface roughness measurements limit IEM utility for large areas. The purely image-based delta index has significant operational advantage in soil moisture estimates for broad areas. Additionally, satellite observations of backscatter used in the delta index indicated an approximate 1:1 relationship with soil moisture that explained 91% of the variability, with RMSE = 0.03. Results showed that the delta index is scaled to the range in observed soil moisture and may provide a purely image based model. It should be tested in other watersheds to determine if it implicitly accounts for surface roughness, topography, and vegetation. These are parameters that are difficult to measure over large areas, and may influence the delta index.