Applications of digital terrain data to enhance the utility of remotely sensed multispectral image data

Abstract

This research demonstrates two different applications of digital terrain data for improving multispectral classification of forest and range cover types. One method involves a stratification process to determine probabilities of occurrence and to develop an ecological distribution model for the six major vegetation types within the eastern San Francisco Volcanic Field in Arizona. The second method employs reflectance modeling techniques to reduce brightness variations resulting from topography and illumination in Landsat image data of the area in order to examine the intrinsic qualities of the natural surface cover. Image processing techniques are used to simulate the topographic effect on surface radiance and for quantifying Landsat scene modulation due solely to topography and sun angle. These two pre-classification modeling techniques lead to higher accuracies of cover type classifications. Further investigations and model integration are recommended to gain a more thorough understanding of vegetation cover characteristics in mountainous terrain.