Fusion of SAR and TM data for quantitative estimation of an extended range of validity

Near-simultaneous AIRSAR and Landsat TM data are used, along with radar scattering models and optical regression-based models, to estimate vegetation variables for a mixed old-growth dense forest in the US Pacific Northwest. The main variable of interest is the foliage biomass, which is related to fundamental scattering properties of the forest stands used to model SAR data, through the allometric relations for the species under study. Polarimetric and interferometric SAR data and the Landsat TM data are first coregistered, taking care of the effects of surface topography. Relationships between foliage mass and TM data channels are derived through a regression analysis applied to ground measurements of the variable. Analytical scattering models are used to derive parallel relationships between SAR data and foliage mass. A nonlinear estimation algorithm applied simultaneously to both data types results in estimates of foliage mass which are expected to be more accurate and have a larger range of validity than what could be obtained from either data type alone