Using the solar elevation angle and radiance conversion to normalize forest spectral signatures

Abstract

A number of spectral signals associated with forest cover were selected in order to test the usefulness of adjustments to Landsat MSS digital count values. The signatures were taken from 26 independent regional land cover inventories in the northeastern US. A regression analysis was conducted which revealed a significant relationship between digital count values and variation in the solar elevation angles in MSS bands four, six, and seven for deciduous forests and in all MSS bands for conifers. When signatures were adjusted for solar elevation angles and MSS sensor calibration differences, the dependence on solar elevation angle was reduced. The adjustments also significantly reduced the variance with Level II forest categories among scenes.