关键词: 土壤养分, 空间变异性, 地统计学

Abstract: In this paper, land use/cover changes of Keerqin Sandy Land in Wengniute Banner, Inner Mongolia, were monitored by using decision tree classification method with threephase (1989, 1999 and 2007) Landsat ETM/TM data. The results showed that based on the decision tree classification method, the overall accuracy of remote sensing image classification was more than 89.4%, and Kappa coefficient reached more than 0.84. In the past 20-years, the main land use types were both grassland and arable land, and these two types of land use accounted for more than 64%. Furthermore, the area of forest land was very small, accounting for 4.8%-5.3%. From 1989 to 2007, the changes of land area were ranked as: arable land > unused land > forest land >habitation > water area > grassland.

Key words: Soil nutrient, Spatial variability, Geostatistics