Effects of within-season dielectric variations on terrain classification using SIR-C/X-SAR

The authors investigate the sensitivity of the SIR-C/X-SAR instrument to within-season moisture perturbations for the Michigan Forests Test Site (MFTS). Precipitation events occurred at the conclusion of the October mission and thus analysis of the last four data-takes can demonstrate the relationships between SAR backscatter (σ°) and precipitation at each polarization and frequency. It is expected that backscatter is affected by incident precipitation and that the presence/absence of intercepted precipitation will be found to have the greatest impact at the shorter wavelengths and for vegetation classes characterized by low relative biomass. Four scenes for SRL-2 were acquired at a common viewing geometry (41° angle of incidence) and 24-hour revisit interval. The means and standard deviations are used to measure the sensitivity of feature vectors to intercepted precipitation and soil moisture content conditions. Differences are generated using the October 7 scene (taken before any precipitation events) as the baseline. T-tests are used to determine significance of the differences. Results show (1) that changes in σ° of agricultural fields are correlated with near-surface soil moisture and (2) changes in σ° of forested areas corresponds to senescence of deciduous foliage and the competing effects of intercepted precipitation on crown-layer attenuation. Typically, intercepted precipitation changes scattering mechanisms at shorter wavelengths (C-band) and mainly enhances attenuation at L-band. T-tests show these daily differences are commonly significant. Hence, land-cover classifications and biophysical retrieval algorithms must consider these sources of scene variability in σ°