Proceedings Article | 27 September 2006
KEYWORDS: Landsat, Remote sensing, Image classification, Ecosystems, Climatology, Satellites, Analytical research, Geographic information systems, Image enhancement, Earth observing sensors
Multitemporal remotely sensed data provide an accurate, economical means to analyze the changes in land cover over time. Land cover change in the region of Baiyang Lake that is the biggest freshwater lake in North China effects local eco-environment intensely. Based on the Landsat (TM) data for 1987, 1991, 1996, and 2002, and employing the maximum-likelihood method, the land cover was classified into seven types, farmland, forest land, urban land, village, water body, wetland and bare land. The overall classification accuracies averaged 86% and the Kappa coefficient is 0.75. Then the transition matrix of The LCC was obtained by overlaying land post-classification map. Between 1987 and 2002 the amount of farmland decreased from 63.9% to 58% of the total land area, wetland decreased from 4.5% to 3.3%, while forest land increased from 2.6% to 3.3%, urban land increased from 1.2% to 2.6%, village increased from 26.1% to 29.1%, water body increased from 1.3% to 3.3%, the amount of bare land was unchanged. Land cover change can not take place independently but has certain linkages with the socioeconomic factors and mutations in natural conditions. Precipitation controlled the area of water and wetland, and human practice process restricted conversions of farmland, urban land, village and forest land.
