Date of Award
Master of Science
John B. Rehder
James R. Carter, Edwin H. Hammond
Landsat satellite imagery of the Great Smoky Mountains in East Tennessee and Western North Carolina exhibits dark tonal reflectances within the Blue Ridge physiographic province unlike any other reflectance patterns found on the remainder of the imagery. Repetitive, seasonal imagery indicate that these unique patterns are dynamic. There also appear to be definable, minute reflectance variations within the patterns themselves, indicating that there are numerous factors accounting for the anomaly. Among the factors discussed are the cover characteristics of the red spruce (picea rubens) and Fraser fir or southern balsam fir (abies fraseri), effects of topography, slope, aspect and ridge orientation, effects of solar angle and azimuth, and shadow zones.
Data were collected for three test sites within the boundary of the Great Smoky Mountains National Park between January 1980 and October 1980. The selected areas were easily accessible and were studied in the field by the author. Landform characteristics were obtained from United States Geological Survey topographic maps of the region. Cover characteristics were obtained from field research and from ancillary data of the National Park Service, the Uplands Research Laboratory, and the University of Tennessee Department of Forestry.
Landsat spectral data were analyzed in a two step method. The first step consisted of visual and photo-mechanical enhancement techniques. This included obtaining Landsat image scenes from the Earth Resources Observations System Data Center (EROS). It was hoped that by utilizing numerous photo-processing techniques previously obscurred data could be enhanced, analyzed, and classified.
The second step consisted of applying a supervised computer classification program to a Landsat digital tape of the study area. The supervised classification program analyzes a Landsat scene on the basis of established training sites that correspond to known locations studied in the field. The computer classification proved more adaptable to enhancing and classifying discrete regions than did photo-mechanical techniques.
The study suggests that more research into the feasibility of utilizing Landsat multispectral data in areas of low accessibility or mountainous terrain needs to be developed. It also suggests that numerous factors influence scene spectral levels and that the best means of examining these factors is through computer classifications based on selected test sites.
Ambrosia, Vincent Gerard, "Terrain Cover and Shadow Discrimination from Landsat Data of the Great Smoky Mountains National Park. " Master's Thesis, University of Tennessee, 1981.