Submitted to: Remote Sensing of Environment
Publication Type: Peer Reviewed Journal
Publication Acceptance Date: 7/21/1998
Publication Date: N/A
Citation: N/A Interpretive Summary: Synthetic aperture radar (SAR) has potential use as a means of measuring soil moisture from space or high altitude airborne platforms. However, the signal from other landscape features, such as topography, vegetation and surface roughness, may overwhelm and confound the soil moisture signal. Previous researchers tried to correct for these effects using incident angle as the most significant landscape variable. Their corrections were not effective. In this paper, we determined the effects of vegetation, surface roughness, and topography on SAR response. In addition, we used statistics to determine the optimal combination of instrument variables, wavelength and polarization, for utilizing SAR for soil-moisture measurement. The instrument variables L band (describes the wavelength) and HH polarization were found to be an optimal combination. We also found that a sparse vegetation cover of sagebrush did not affect the SAR signal. Finally, we were able to significantly improve the corrections of effects of landscape variables on SAR response by incorporating surface roughness effects. This should improve our ability to monitor soil moisture from space.
Technical Abstract: Synthetic aperture radar (SAR) may potentially measure soil- water content from space. However, the signal from landscape features such as topography, vegetation and surface roughness may overwhelm and confound the soil water signal. Previous work using incident angle alone for corrections have not been effective. In this paper we determined the effects of vegetation, surface roughness, and topography on SAR backscatter. In addition, multivariate statistics were used to determine the optimal sensor wavelength and polarization. L band and HH polarization were found to be an optimal combination of sensor variables for soil moisture determination. By comparing NDVI and the SAR vegetation index, we found that the sparse vegetation cover of sagebrush in the study area did not effect L band SAR backscatter. The root mean square surface height, determined using an empirical algorithm was used to stratify the study area into low and high surface roughness areas. The high surface roughness areas were found to match very and extremely rocky areas, classified using a soil map. Correction of SAR data was significantly improved with the incorporation of surface roughness.