BRDF MODELS TO PREDICT SPECTRAL REFLECTANCE AND EMISSIVITY IN THE THERMAL INFRARED

This paper presents modifications to the linear kernel bidirectional r eflectance distribution function (BRDF) models from Roujean et CII. an d from Wanner cr nl. that extend the spectral range into the thermal i nfrared (TIR), Our application is to synthesize the TIR optical proper ties of a scene pixel from laboratory component measurements, The angu lar reflectance and emissivity are needed to convert the radiance of a pixel as measured from space to land-surface temperature, The kernel models will he applied to develop a look-up table for the MODIS land-s urface temperature algorithm to estimate the spectral, angular scene e missivity from land cover classification, A shrub scene and a dense ca nopy scene illustrate qualitative differences in angular emissivity th at would not be evident without the kernel model modifications, We con clude that the modified models provide a simple and efficient way to e stimate scene optical properties over a wide spectral range.