摘要
本文从改进传统CA模型在土地利用变化时空模拟方面存在缺陷和不足的角度出发,结合Markov模型的总量预测优势以及CA模型的空间演化优势,提出基于Logistic回归的元胞适应度计算方法等,以此构建开发出扩展CA模型,并基于2000年、2008年ETM影像以及相关辅助数据对深圳市2000年-2015年土地利用变化进行了模拟研究。结果表明:①耕地和建设用地是深圳市变化最为显著的土地利用类型,建设用地增加的主要来源是耕地和园地,其次是未利用地;②深圳市未来土地利用方式仍将以建设用地快速扩张为主,若继续目前的发展趋势,未来耕地保护压力巨大;③经检验,扩展CA模型模拟精度达到83.74%,能够满足反映区域土地利用变化趋势的需要,可以为地方政府在制定土地政策时提供有价值的决策参考依据。
Urbanization has been accelerated after China’s reform and opening up over the past 30 years. Land use problems have increasingly emerged in the context of rapid economic development. For a long period of time, local governments have stick to ideas of ＂earning money based on lands＂ or ＂pursuing GDP by land resources＂, which are characterized by rapid expansion of land for construction, declining quality of cultivated land, wasting and lying land resources, and illegal land use. Many cities are currently faced with the plight of land resources bottleneck. As such, there emerges a need to renew our thinking of land resources exploration and to strengthen studies on land use risk assessment, prediction, and management. In the meanwhile, advancing methods of land use simulation is particularly important for land use management. Prediction models are the core of examining the land use change system because they allow analysis, forecasting, and early alarming. According to the connotation of spatial expansion of the land use system, a cellular automation （CA） model was adopted to examine the land use change. There are, however, deficiencies in the standard CA model, such as iteration time or making conversion rule. In order to build a extended CA model for simulating land use change, the authors combined advantages of the total prediction by the Markov model with advantages of spatial evolution by the CA model, put forward a method of cellular fitness calculation based on Logistic regression, and then built and developed the extended CA model. A C＋＋ language-based software system on VC.NET environment was developed, which integrates Markov, Logistic, CA, and GIS technology. It was applied to simulate the land use change of Shenzhen City during the period 2000-2008 using Landsat ETM＋ images acquired in 2000 and 2008. Results show： 1） arable land and construction were the primary land use types experiencing the most significant changes in Shenzhen City, suggesting the main sources of the increased land for construction being arable land, garden land, and unused land; 2） continuing expansion of land for construction was the primary trends in land use change in Shenzhen City in the future, suggesting great pressures of arable land protection if we expect to continue the current development trend; and 3） the accuracy of the extended CA model was roughly 83.74%, which can satisfy the need for revealing trends in regional land use change, and provide valuable reference for policy-making for local governments.
出处
《资源科学》
CSSCI
CSCD
北大核心
2011年第1期127-133,共7页
Resources Science
基金
国土资源部"百名优秀青年科技人才计划"科技项目:"城市土地利用预警体系及虚拟政策实验室建设"
国家自然科学基金项目:"基于扩展CA的城市社会水循环时空演化研究"(编号:50709042)
关键词
扩展元胞自动机
土地利用
时空模拟
深圳市
Extended cellular automata
Land use
Spatio-temporal simulation
Shenzhen City
