SAR terrain correction for improved soil moisture estimation in a mountain watershed

The spatial and temporal distribution of soil moisture can be viewed as a key descriptor of hydrologic processes in a watershed. If SAR is to be a useful tool for studying and identifying these hydrologic processes, then the effects of topography must be separable from the response due to soil moisture within any microwave data set. This is especially true in mountainous terrain where much stronger reflections are observed from radar-facing slopes than from slopes facing away from the radar. To investigate the impact of topography on the accuracy of SAR-derived soil moisture estimates, AIRSAR data were analyzed from a field experiment conducted in the mountainous semiarid Reynolds Creek watershed near Boise, Idaho during 1991. Reynolds Creek watershed is almost entirely covered with sparse sagebrush vegetation which is essentially transparent at L-band wavelengths. In an attempt to correct for topography, a USGS 30 m DEM of Reynolds Creek was registered to the AIRSAR data, and the changes in radar cross-section area and antenna pattern for each pixel due to slopes in the range and azimuth directions were computed using JPL´s POLCAL 4.0 software. Initial results indicate that the 30 m resolution of the USGS DEM may be too coarse for terrain correction of the AIRSAR data for this area. Use of higher resolution DEM for sub-watersheds within Reynolds Creek produced improved calibration of the terrain effect, and increased the correlation between radar backscatter and soil moisture