Publication Date

1986

Document Type

Dissertation/Thesis

First Advisor

Luman, Donald E.

Degree Name

M.S. (Master of Science)

Department

Department of Geography

LCSH

Wetlands--Illinois||Artificial satellites in geographical research||Landsat satellites

Abstract

This research analyzed the changes in Landsat Thematic Mapper classification results for USGS Level II land cover categories. Forested Wetlands and Deciduous Forest Lands, when specific elevation zone data are employed as a classification mask. A Landsat-4 Thematic Mapper color transparency, Bands 1, 3, and 4, was the primary data source utilized. Ancillary data consisted of a continuous tone B&W composite of USGS Level II Forested Wetland and Deciduous Forest Land categories, and a B&W composite of five elevation zones that were compiled from the USGS Paducah topographic quadrangle. The cartographically registered composites were converted to 8-bit digital data by an Optronics P-1700 scanning densitometer. The masked multiband Landsat data were classified under four different conditions, i.e., classification for (1) Forested Wetland areas from 300' and Above, (2) Forested Wetland areas from 300' to 400', (3) Deciduous Forest Land from 300' and Above, and (4) Deciduous Forest Land from 400' and Above. The broad and overlapping spectral ranges for the land cover categories in all TM Bands pointed to the variable and correlated nature of the two land cover types. The presence of vigorous vegetation was deduced to be a major influence on the statistical distribution of the classified pixels, while the mixture of healthy vegetation with senesced vegetation was attributed to causing the broad spectral ranges. The overwhelming majority of the Forested Wetland category was found to be in the 300' to 400' elevation range. The spectral similarity of the land cover types led to large errors of commission, and it was found that the use of elevation zone data markedly decreased commission errors. It was concluded that the classification strategy used in this study would be best applied to data which had a high probability of having commission errors, but a low probability of having omission errors. Additional research on this subject should consider Landsat data from earlier in the growing season, and the use of TM Bands 3, 4, and 5.

Comments

Bibliography: pages 98-105.

Extent

ix, 105 pages

Language

eng

Publisher

Northern Illinois University

Rights Statement

In Copyright

Rights Statement 2

NIU theses are protected by copyright. They may be viewed from Huskie Commons for any purpose, but reproduction or distribution in any format is prohibited without the written permission of the authors.

Media Type

Text

Share

COinS