The utility of calibration - Landsat 4 TM simulator data, Patrick Draw, Wyoming
A novel calibration approach was used for Landsat 4 TM simulator data acquired at the Patrick Draw test site, Wyoming, where the 'true' reflectance spectrum for any pixel area cannot be practically determined because of mixing of spectral components. An empirically derived function was first applied to the scanner data to remove the brightness gradient due to directional scattering and surface geometry. Based on the observation that the distribution of scanner brightness values (DN) were Gaussian, the minimum, mean, and maximum DN for each band were extracted and plotted versus the similar percent reflectance values determined from a large representative collection of field reflectance data for the area. These plots provided linear transformations relating DN value to reflectance in each band.