中文摘要：

基于2000年及2009年广州市TM影像解译得到的土地利用类型图，生成城市用地类型数据，选取道路、水系、城镇政府机构以及城市用地现状等影响因子，借助元胞自动机技术，构建城市用地变化的逻辑回归元胞自动机模型，模拟2009年广州市的城市用地变化，并验证预测结果；再基于2009年城市用地类型图，预测2018年广州城市用地情形．结果显示元胞自动机模型在城市用地变化的预测方面是可行的；若按照前一时段的发展趋势，至2018年时，建设用地面积将达25．215％，超过最新一轮城市规划拟定的指标，表明未来城市发展过程中，城市节约集约利用土地的必要性、迫切性；研究还表明，元胞状态发生转变的概率阈值的进一步研究是十分有价值的．

英文摘要：

According to land-use graphs interpreted from TM images, urban land-use classifications of Guang- zhou were generated. Road, urban rail transit, water bodies and government agencies and its situation of urban land were selected to construct coupling models of cellular automata and logistic regression, which aims to fore- cast urban land-use change. Based on urban land classification in 2000, urban land use in 2009 was first simu- lated. While being verified and revised parameters using urban land classification in 2009, the coupling models were finally used to predict urban land-use change in 2018. Results showed cellular automata model has a cer- tain feasibility in predicting urban land-use change. In accordance with the trend of the previous period, to 2018, the construction area will reach 25. 215% of the total area, which is beyond the planning figure in the latest round urban master plan. This disclosed that intensive land would be improved in future urban develop- ment process. In addition, further study on probability threshold of the models is required and is also very valu- able work.