KAS Symposium
Geospatial Analysis in the Great Plains
Oral Presentation Title/Authors
DEVLOPING A LAND COVER MODELING PROCEDURE FOR THE HIGH PLAINS USING MULTI-DATE
THEMATIC MAPPER IMAGERY. Kevin P. PRICE, Stephen L. EGBERT, Geography
& Kansas Applied Remote Sensing, University of Kansas, Lawrence, KS 66045;
Duane NELLIS, Department of Geography, Kansas State University, Manhattan, KS
66506; and Re-Yang LEE, Kansas Applied Remote Sensing, University of
Kansas, Lawrence, KS 66045.
Abstract
The objective of this study was to develop a repeatable procedure for
modeling land use and land cover within one of the most agro-economically
significant and environmentally sensitive areas of the High Plains
region--Finney County in southwestern Kansas. The method we developed
involves the use of multi-seasonal Landsat Thematic Mapper satellite
remotely sensed images collected in spring, mid-summer, and late summer for
1987, 1989, 1992. TM bands [3 (red), 4 (NIR), and 5 and 7 (MIR)] were used
to produce a 12-band dataset for each year.
The results of the computer classification of these three datasets showed
greater than 95% accuracy in separating cropland from grasslands. By
comparison, conventional single image classification approaches yielded
less than 60% accuracy. We also developed a method for classifying: winter
wheat, grain sorghum (milo), corn, alfalfa, and fallowed lands, and
achieved an overall accuracy of greater than 90%. Our total acreage
estimates by crop type agreed strongly with government reported acreage for
Finney County (r² = 0.93). The ability to accurately classify land
cover at the resolution of Landsat (about 5th acre resolution) is of
considerable importance to hydrologic, agricultural, and environmental
models that require accurate maps of land cover as an input variable. These
land cover maps also provide invaluable information to natural and
agricultural resource managers.
Return to KAS Symposium