Abstract: Based on the remote sensing images in the years 1990 and 2000 and the data of digital elevation model,stream systems,railways,highways,precipitation and temperature,the integrated use of Binary Logistic Regression and CA-Markov model was developed to simulate the land use patterns in the loess tableland areas in Xianyang,Shaanxi Province,and the land use types there in the near future were predicted.The results are as follows: ① There was an expansion of construction land and woodland during the period from 2000 to 2010,but the area of arable land,grasslands,waters and unused land was reduced,in which the reduction trend of arable land and the expansion trend of construction land were similar; ② Investigating from the conversion direction of land use types,grassland was mainly changed into woodlands,and then into farmland,construction land and waters; construction land and woodland was mainly from arable land; the change of area of woodland and waters was not significant,unused land was mainly from arable land,and some unused land in small area was changed into arable land; ③ Through inputting the parameters which were combined with the results of Binary Logistic Regression,the application range of CA-Markov model was expanded,and the prediction of land use was improved to some degree; ④ It was predicted that the area of construction land in the study area in 2020 would be expanded dramatically,especially in the surrounding regions of the original construction land,along the railways and highways,and in Sanyuan County of Qindu District,Xianyang City.

Key words: Binary Logistic, CA-Markov, land use/cover change, spatial pattern, loess tableland area, Xianyang