Estimating sub-pixel surface roughness using remotely sensed stereoscopic data

Surface roughness at the scale of 10− 2- 101 m can be estimated using the ratio (RR) between surface-reflected solar radiance measured from two view angles at nearly the same time. As RR is primarily a function of the difference between effective sub-pixel shading observed from the two view-angles, the divergence from unity RR values, which are expected for smooth Lambertian surfaces (they have no shadows), was found to be proportional to roughness at the 10− 2- 101-m scale of geomorphic desert surfaces. Ground-based RR values at ∼1-m resolution, as well as remotely acquired RR values at 4-, 15- and 50-m resolutions, were compared with observed surface roughness in two hyperarid test sites, located in Death Valley, California USA and the southern Negev Desert in Israel. The ASTER (Advanced Spaceborne Thermal Emission and Reflection Radiometer) 15-m stereoscopic capability is identified as an effective resource for obtaining relative sub-pixel surface-roughness estimates that are largely independent of surface composition and relatively insensitive to atmospheric effects.