Return to HydroWeb Homepage
Electronic journal of the International Association
for Environmental Hydrology
On the World Wide Web at http://www.hydroweb.com
JEH Volume 10 (2002), Paper 6, September 2002
Posted September 21, 2002
DERIVING LAND COVER OF A LARGE AGRICULTURAL WATERSHED FROM MULTI-TEMPORAL
LANDSAT SCENES
Samar J. Bhuyan
Arizona Dept. of Environmental Quality, Phoenix, Arizona, USA
Kyle R. Mankin
J. M. Shawn Hutchinson
Douglas G. Goodin
James K. Koelliker
Kansas State University, Manhattan, Kansas, USA
ABSTRACT
Determining land-cover characteristics of large watersheds for use
in hydrologic models has been enhanced by application of remotely sensed
data and the technologies used to interpret them. The objective of this
paper was to describe and evaluate a process for obtaining land-use information
for a large watershed using multi-temporal Landsat-5 Thematic Mapper (TM)
images. The Kanopolis Lake watershed, which covers 6316 km 2 in central
Kansas, USA, was evaluated. Land-cover data for 1992 was derived for use
in the Agricultural NonPoint Source Pollution (AGNPS) hydrologic model.
Due to shape, large size, and geographical location of this watershed,
two early-summer Landsat TM images were required to cover the entire watershed.
Both scenes were classified separately and then combined together to estimate
land-cover information. ISODATA (Iterative Self-Organizing Data Analysis
Technique) algorithm of unsupervised classification followed by supervised
classification was performed. Initially, a Level-1 classification scheme
was used, which differentiated cover classes among water, agricultural,
rangeland, forest, residential, and barren areas. The agricultural areas
were reclassified into winter wheat and summer crop and rangeland classified
into low, medium, and high cover. Good agreement was found with other published
land-cover spatial data, with consistent results across both Landsat scenes.
Reliability, use of readily available data, and reasonable ease of use
make these methods appropriate for hydrologic modeling of small to large
watersheds.
Reference: Bhuyan S. J., K. R. Mankin, J. M. S. Hutchinson,
D. G. Goodin, J. K. Koelliker; Deriving Land Cover of a Large Agricultural
Watershed from Multi-temporal Landsat Scenes, Journal of Environmental
Hydrology, Vol. 10, Paper 6, September 2002.
CONTACT:
Prof. Kyle R. Mankin, Ph.D.
Biological and Agricultural Engineering
Kansas State University
147 Seaton Hall
Manhattan, KS 66506
E-mail: kmankin@ksu.edu