A Field Evaluation of ERS-2 Synthetic Aperture Radar Data for Soil Moisture Measurement
Jeffrey P Walker, Dinush F Kurera and Paul R Houser
Synthetic Aperture Radar (SAR) backscattering observations, such as the C-band vertically polarized observations from the ERS-2 satellite, have the potential to measure moisture content in a near-surface layer of soil. However, SAR backscattering observations are highly dependent on the topography, soil texture, surface roughness and soil moisture, meaning that soil moisture inversion from a single frequency, single polarization SAR observation is difficult.
The potential for measuring near-surface soil moisture with the ERS-2 satellite has been explored by comparing model estimates of backscattering with the ERS-2 SAR observations. This comparison was made for the 6ha Nerrigundah catchment located in a temperate region of Eastern Australia. ERS-2 data were obtained for the two overpasses coincident with near-surface soil moisture measurements from 15cm TDR probes on a 20m grid. A drop-pin profiler was used to estimate the rms surface roughness and correlation length at 5 locations, and soil texture assessments were made at 19 locations. Topographic influences were accounted for by a DEM. In addition, 1cm soil moisture data was obtained from a soil moisture model that was calibrated to 10 months of measured soil moisture profile data at 13 locations.
Using the theoretical integral equation model of Fung, semi-empirical model of Oh and the empirical model of Oh, it was found that when using the measured data there were rms errors from 3.5 to 8.5dB and r^2 values from 0.00 to 0.25, depending on backscattering model and the degree of filtering. Using the model soil moisture in place of the measured soil moisture reduced the rms errors slightly (0.5 to 2dB) but did little to improve the r^2 values. Likewise, using the first day of ERS-2 observations and measured/model soil moisture to solve for rms surface roughness reduced the rms errors in backscattering coefficient for the second day to between 1 and 3dB, but did not improve the r^2 values.