BOREAL FOREST ECOSYSTEM CHARACTERIZATION WITH SIR-C/XSAR

This paper discusses early results obtained from Spaceborne Imaging Radar-C (SIR-C) and X-band synthetic aperture radar (XSAR) data over a boreal forest in Saskatchewan, Canda. Multifrequency and multipolarization image data were made available during the SRL-1 (Apr. 10, 1994) and SRL-2 (Oct. 1, 1994) missions. These image data sets were analyzed and maps of forest cover type and above ground woody dry biomass were generated. A portion of the Southren Study Area of the Boreal Ecosystem-Atmosphere Study (BOREAS) was mapped for forest cover type with classification accuracies on the order of 80%, Maps of estimated biomass were also produced that match observed patterns and preliminary ground data, The upper limit of sensitivity of the radar to boreal forest biomass in our study area was about 20 kg/m(2) or 200 tons/ha, The highest average observed biomass in our ground measurements was about 25 kg/m(2). Highest sensitivity of the radar to biomass was attained using April backscatter data and a ratio of L-band HV to C-band HV. Results show that radar estimates of biomass were within +/- 2 kg/m(2) at the 95% confidence level, A comparison of the April and October data sets was conducted to understand the effects of seasons on the analysis. It appears that the frozen trees and wetter background contributes to increased backscattering observed in the April data, These early results indicate that multiple polarization and multiple frequency SAR data can be used to monitor and map northern forest biomes.