本文利用元胞自动机模型和多智能体模型相结合的方法，在GIS技术手段的支持下构建了一个能够兼顾两种模型优点的城市土地利用动态模型，并以上海市实证研究对象，模拟了上海市2005年的城市土地利用状态，并分别预测了2010年和2020年上海市土地利用的动态演化结果。根据1990年、1995年、2000年和2005年遥感数据解译的上海市城市土地利用数据，首先分析了上海市土地利用的基本特征和景观格局的动态变化，通过六个景观格局指数客观反映了上海市土地利用现状，结果表明：城镇用地的扩张方式仍以空间集聚为主，并在15年来一直保持在相对稳定的水平，景观的空间复杂度和边缘密度等变化不大。上述结论为后面以上海市为例进行元胞自动机和多智能体相结合的城市土地利用动态模型的实证研究提供了可行的前提。模型构建的数据来源主要有上海市数字城市数据（包括学校、医院、银行、居民点、商业、工业、公园、绿地等专题信息）、上海市数字高程图、上海市人口经济数据等。模型分为两大部分，元胞自动机模型和多智能体模型，两者在GIS环境下集成，利用多智能体模型弥补元胞自动机模型的缺陷，使模型更多地综合城市系统演化的社会、经济、文化和政策等影响因素。在元胞自动机模型中定义城市系统中的各种自然、社会和交通等要素，通过“自下而上”的自发转换获得一种城镇用地转换概率。在多智能体模型中，把城市中的智能体分为政府、居民两种类型，根据其自身特点设定其在土地利用转化过程中的作用。不同类型的智能体根据自身需求具有不同的行为规则，每种类型的智能体根据自己的偏好选择城镇用地转化位置，最终通过协商达成一致。城市土地利用的变化和空间的扩张由智能体的行为，智能体之间和智能体与环境之间的相互影响决定。多智能体模型中也产生一种城镇用地转换概率。最后将两个模型的概率结合起来，同时考虑到随机因素的影响得到最终的城镇用地转化概率。根据模型模拟出的上海市2005年的城市土地利用状态和实际土地利用状态的Kappa系数的平均值达到0.75以上，说明模型具有较高的可信度。最后预测出2010年和2020年上海市土地利用状态，并运用景观格局指数法进行了分析，为上海市城市发展和城市土地利用规划提供决策支持。

Using cellular automata model and multi-agent model combining method with the support of GIS technology a model to combine the advantages of two dynamic models of urban land use was built and put into empirical study in Shanghai, as simulated the Shanghai urban land use in 2005, the state, and were predicted in Shanghai in 2010 and 2020, the dynamic evolution of land use results. According to the 1990, 1995, 2000 and 2005 remote sensing data interpretation of the Shanghai Urban land use data, first analyzes the basic characteristics of land use in Shanghai and the dynamic changes of landscape pattern, landscape pattern index by six objective reflects land use in Shanghai, the results show that: the expansion of urban land is still room for gathering the main way in 15 years has remained relatively stable level, the landscape of space complexity and the edge density changed little. The conclusion is followed in Shanghai for cellular automata and multi-agent combination of the dynamic model of urban land use empirical research to provide a viable premise. Model data are mainly from the city of Shanghai Digital data (including schools, hospitals, banks, residential, commercial, industrial, parks, green spaces and other special information), digital elevation map of Shanghai, the Shanghai Population and Economic data. Model includes two parts, cellular automata model and multi-agent model, both are integrated in the GIS environment, using multi-agent model make up the deficiencies of cellular automata models , to make the model more comprehensive with social, economic, cultural and policy influencing factors. In the cellular automata model of urban systems in the definition of natural, social and transport factors, through "bottom-up" to obtain a spontaneous conversion of urban land conversion probability. In the multi-agent model, the agents in the city are divided into government, the residents of two types, according to the characteristics of their own set of land use transformation in the role. Different types of agents have different needs according to their own rules of conduct, each type of agent selected according to their preferences location of urban land conversion, the final agreement through consultation. Urban land use change and the expansion of space by the behavior of an agent, agents and between agent and environment, the interaction between the decision. Multi-agent model of urban land conversion also produces a probability. Finally, the probability of the two models together, taking into account the impact of random factors in the final of the urban land conversion probability. According to the model to simulate the Shanghai urban land use in 2005, the state and the actual land use status of the Kappa coefficient of above average to 0.75, indicating the model has high credibility. Finally the land use conditions of Shanghai in 2010 and 2020 were predicted, and the landscape pattern index was analyzed for the Shanghai urban development and the urban land use results were able to provide support for planning decision making.