基于MCR-ANN-CA模型的包头市生态用地演变模拟  

Simulation of Ecological Land Transition in Baotou City Based on MCR-ANN-CA Model

作　　者：刘建华 张启斌[1] YANG Di 岳德鹏[1] 于强[1] 杨斓 LIU Jianhua;ZHANG Qibin;YUE Depeng;YU Qiang;YANG Lan(Beijing Key Laboratory of Precision Forestry,Beijing Forestry University,Beijing 100083,China;Department of Geography,University of Florida,Gainesville FL 32611,USA)

机构地区：[1]北京林业大学精准林业北京市重点实验室,北京100083 [2]弗罗里达大学地理系,盖恩斯维尔fl32611

出　　处：《农业机械学报》2019年第2期187-194,共8页Transactions of the Chinese Society for Agricultural Machinery

基　　金：国家自然科学基金项目(41371189);"十二五"国家科技支撑计划项目(2012BAD16B00)

摘　　要：以内蒙古自治区包头市为研究区,耦合最小累积阻力(MCR)模型、人工神经网络(ANN)和元胞自动机(CA)构建MCR-ANN-CA模型。利用MCR模型量化包头市各用地类型演变为生态用地时的阻力,构建CA适宜性规则;利用ANN模型提取CA邻域转换规则,基于包头市2006、2011年土地利用数据及归一化植被指数(NDVI)、高程、坡度、水体距离、人口密度多项数据,对2016年生态用地演变情景进行模拟,以2016年实际生态用地分布为参照,将该模型模拟结果与CA-Markov模型的模拟结果进行对比(以2016年实际景观分布为参照),结果显示,两种模型模拟结果的卡帕一致性指数(Kappa index of agreement,KIA)分别为0. 89和0. 87,相对误差分别为3. 10%和5. 31%,MCR-ANN-CA模型显示了较高的模拟精度。To simulate the ecological land transition of Baotou City,Inner Mongolia,the MCR-ANN-CA model was built.This model was composed of minimum cumulative resistance model(MCR),artificial neural network(ANN)and cellular automata model(CA).The MCR model was used to simulate the resistance faced by ecological land during its transition.To measure the resistance,the data of normalized difference vegetation index(NDVI),digital elevation model(DEM),slope,distance from water,population density and industrial park distribution were normalized and superimposed to produce the resistance surface of MCR model.Then a cumulative consumption resistance surface was generated by using the cost distance tool of ArcGIS software.The generated surface was taken as the suitability map of CA model.The ANN model was used to extract CA neighborhood transfer rules and consider the neighborhood land use structure.The CA model combined the cumulative consumption resistance surface and the neighborhood transfer rules extracted by using the CNN model.And the MCR CNN CA model was finally formed.Based on the land use data of 2006 and 2011,the transition of construction land of Baotou City in 2016 was simulated by using the MCR-ANN-CA model.The simulation result of the model was compared with the CA Markov model.The Kappa index of agreement(KIA)of the two models were 0.89 and 0.87,and the relative errors were 3.10%and 5.31%,respectively.The MCR-ANN-CA model showed high simulation precision.

关 键 词：生态用地演变 元胞自动机 人工神经网络 最小累积阻力模型 包头市 

分 类 号：K903[历史地理—人文地理学]