随着人口和经济的不断增长，人地之间的矛盾不断加剧。城市的扩展，是经济发展的必然趋势，一方面国民经济的发展必然促进城市化的发展，另一方面不断侵占耕地和绿地，加剧水资源的紧张状况，同时这种城市化又常常发展不平衡，导致城中农村或村中城市的出现，因此急需对此进行动态监测和合理决策。利用遥感监测技术提供实时监测信息，对于城市发展进行现状分析和宏观调控，从而促进国民经济的持续稳定发展起到推动作用。许多研究致力于分析及建立模型模拟城市的发展，探讨这一综合体的发展机制，并已取得不少成果。随着我国城市化进程的加速，这些研究越来越显示出其重要作用，特别是利用遥感技术探讨将时间系列与空间分析相结合的模型，对于研究城市发展的潜在影响是十分有意义的。本文以珠海为研究区，基于2001年、2002年、2003年、和2004年的中巴影像，利用遥感技术，将这几年的遥感图像进行分类，在分类遥感图像的基础上，建立了马尔可夫单独模型，马尔可夫分别与直接邻域以及距离加权法结合的模型，并在这两种模型的基础上考虑了每一个像素与道路、河流的最近距离从而建立直接邻域综合法以及距离加权综合法模型。并分别以8，24，48，80，120，168，224不同的邻域在2003年分类图的基础上对2004年进行预测分析，并用2004年的分类图进行结果评估，结果表明采用80或120邻域的距离加权综合法效果最佳。说明了马尔可夫链方法与邻域信息结合的模型是可行的，通过数学模型的运用，可以帮助我们深入研究时空变化，从而认识客观世界，预测未来和再现过去。

With the growth of the population and the economy, the conflict between human and land aggravates ceaselessly. Two results are leaded by the city expansion, which is the necessarily trend of the economy development. At one hand, the city developmental rate will be necessarily accelerated by the growth of the national economy; at another hand, arable land and forest is invaded and the problem of lacking water resource is also aggravated. So the land use should be dynamically supervised and correctly decided. The persistent growth of country’s economy can be impelled by the analysis of the city developmental status quo with the use of remote sensing technology which provides the indispensable information for dynamically supervision.A lot of researchers dedicate themselves to analyzing and modeling the city development and improving the syntheses’ developmental mechanism, as a result, they have achieved great success. These researches especially on the models which combine the time series and the spatial analysis by remote sensing technology are tremendously important to study city developmental latent effect.ZhuHai city is the experimental area in this paper and the 4 different China-Brazil Earth Resource Satellite images which are taken in 2001,2002,2003,2004 respectively are classified with the technology of remote sensing. Based on the classification, five models are built: one model containing Markov chain alone, two models combining Markov chain with direct neighbors and weighted distance respectively, and two integrated models containing the nearest distance from every pixel to road and river based on former two models. With the five models, we simulated land use transition of ZhuHai city in 2004 using 8, 24, 48, 80, 120,168 and 224 neighbors respectively based on the land use map in 2003. The result shows that the weighted distance integrated method with 80 neighbors or 120 neighbors is the best model by evaluating the simulative result of 2004 land use map with the truly 2004 land use map. At the same time, the result indicates the feasibility of the method containing Markov chain and spatial information. The use of mathematic models could help us to research the space-time change, finally help us understand the impersonal world by foreseeing the future and emerging the elapsed things.