Soil moisture evaluation using multi-temporal synthetic aperture radar (SAR) in semiarid rangeland

There have been several efforts to utilize satellite-based synthetic apertu re radar (SAR) measurements to determine surface soil moisture (to 5 cm) co nditions of rangeland regions. The results have been mixed since the relati on between the SAR signal and surface soil moisture is confounded by variat ions in topographic features, surface roughness and vegetation density. We designed an experiment to investigate the sensitivity of C-band SAR backsca tter (sigma (0)) to surface soil moisture (theta (s)) in a semiarid rangela nd and to test a data-fusion approach based on both optical (Landsat TM) an d radar (ERS-2 SAR) measurements to improve regional estimates of surface s oil moisture content. The data-fusion approach [Sano, E.E. 1997. Sensitivit y analysis of C- and Ku-band synthetic aperture radar data to soil moisture content in a semiarid regions. Ph.D. Dissertation. University of Arizona, AZ] utilized the difference between dry- and wet-season SAR sigma (0) to no rmalize roughness effects, and utilized surface reflectance in optical wave lengths to account for differences in vegetation density. We focused the st udy on three flat, uniformly vegetated sites of known surface roughness, mo nitored variations in surface soil moisture, vegetation density and SAR sig nal over time, and obtained eight optical/SAR image pairs throughout the dr y and wet seasons. For these sparsely vegetated sites during this dry year (1997), we found that the SAR signal was not significantly attenuated by sp arse green vegetation cover (green leaf area index ( 0.35) and dense standi ng brown vegetation cover (brown leaf area index up to 1.5). Consequently, the optical data was not required for this application, and the approach co uld be implemented by simply taking the difference between the dry- and wet -season SAR sigma (0) (sigma (0) - sigma (0)(dry)). For a data set of eight dates at three study sites, we confirmed that the relation between ERS-2 C -band SAR sigma (0) and theta (s) was weak (r(2) = 0.27); yet for the same data set, that the relation between sigma (0) - sigma (0)(dry) and theta (s ) was strong and significant (r(2) = 0.93). This study also raised two conc erns: (1) the overall sensitivity of SAR sigma (0) to theta (s) was relativ ely low, and (2) the approach required a high level of accuracy in the esti mate of green leaf area level that may not be obtainable with standard opti cal remote sensing algorithms. In any case, the positive results from this study should encourage the use of a multi-temporal SAR and optical/SAR fusi on for monitoring semiarid range conditions, and improving management of sc arce resources. Published by Elsevier Science B.V.