Acta Scientiarum Naturalium Universitatis Pekinensis

Previous Articles     Next Articles

An ANN-CA Modeling Method for Land Cover Change in the Karst Area of China: A Case Study of Maotiao River Basin

WANG Lei, WANG Yang, CAI Yunlong   

  1. Center for Land Study, College of Urban and Environmental Sciences, Peking University; Laboratory for Earth Surface Processes MOE, Beijing 100871;
  • Received:2011-02-25 Online:2012-01-20 Published:2012-01-20



  1. 北京大学城市与环境学院土地科学中心, 地表过程分析与模拟教育部重点实验室, 北京100871;

Abstract: The land use of Karst areas in Southwest China changed drastically due to its fragile eco-environment properties and the disturbance from human activities. Modeling and analyzing these changes is impeded by its complex surface features and the intensive human-environment interaction. In order to solve this problem, a combined ANN-CA modeling method is used on Maotiao River Basin in Guizhou Province by the modeling results compared with the actual land use map. The results show 87.62% accuracy with confusion matrix, 57.36% with figure of merit, and all of the selected landscape indexes are similar between them, proving this method a good option for the land use change analysis in Karst and other similar areas.

Key words: land use/cover change, artificial neural network, cellular automata, ANN-CA model, Maotiao River Basin in Guizhou Province

摘要: 为了准确把握西南喀斯特地区土地利用格局的变化规律, 以贵州省猫跳河流域为例, 采用人工神经网络与元胞自动机的耦合模型对喀斯特地区1990?2002年间的土地利用格局变化进行了模拟。将模拟结果与实际土地利用图进行对比发现: 在数量变化方面, 两者的混淆矩阵和性能指数的总精度分别达到了87.62%和57.36%; 在空间格局方面, 模拟结果的景观格局指数均接近真实值。研究结果表明, 该模型模拟精度较高且可操作性强, 能够作为西南喀斯特地区小尺度范围土地利用变化研究的有效工具。

关键词: 土地利用/覆被变化, 人工神经网络, 元胞自动机, ANN-CA耦合模型, 贵州省猫跳河流域

CLC Number: