STATISTICAL CHARACTERISTICS OF SIMULATED RADAR IMAGERY FROM BARE SOILSURFACES - EFFECTS OF SURFACE-ROUGHNESS AND SOIL-MOISTURE VARIABILITY

The potential of high-resolution radar imagery to estimate various hyd rological parameters, such as soil moisture, has long been recognized. Image simulation is one approach to study the interrelationships betw een the radar response and the underlying ground parameters. In order to perform realistic simulations, we incorporated the effects of natur ally occurring spatial variability and spatial correlations of those g round parameters that affect the radar response, primarily surface rou ghness and soil moisture. Surface roughness and soil moisture images w ere generated for a hypothetical 100 x 100 m bare soil surface area at 1 m resolution using valid probability distributions and correlation lengths. These values were then used to obtain copolarized radar scatt ering coefficients at 2 GHz (L band) and 10 GHz (X band) frequencies u sing appropriate backscatter models, which were then converted to a di gital number within 0-255 gray scale in order to generate radar images . The effect of surface roughness variability causes variability in th e radar image, which is more apparent under smooth soil conditions. On the other hand, the inherent spatial pattern in soil moisture tends t o cause similar patterns in the radar image under rougher soil conditi ons. The maximum difference between contrast-enhanced mean values of t he radar image digital number due to moisture variations occurs at sur face roughness values in the 1.5-2.0 cm range.