Biophysical interpretation of a spectral mixture model based on a radiative transfer model and observational data

Linear spectral mixture models were used to invert mixed spectral responses of a target into proportions of plant and soil components at the Earth´s surface. However, utilization of the mixture models has been limited by a lack of biophysical interpretation of the results. The objective of this study was to relate the decomposed components of a mixture model with known biophysical properties of the surface. A radiative transfer model was used to generate “mixtures” reflectances using measured optical properties of “pure components”. The radiative transfer model was used to create canopy reflectances for varying leaf area index (LAI), and intercepted photosynthetically active radiation (IPAR) over a range of mixed targets. The mixture model was applied to the simulated data and validated with observational ground data (LAI, IPAR and reflectances) measured at the HAPEX-Sahel (1992) bush/grassland and millet sites. The mixture model used a combination of “pure component” spectral signatures (full green canopy and soils) measured at the site. The method was applied to subsets of a nadir image of the Advanced Solid State Array Spectroradiometer (ASAS) with and without atmospheric contamination, which resulted in quantified estimates of the abundance of vegetation and soils. Site LAI and IPAR could also be estimated from the ASAS imagery