景观动态的Markov模型研究——以长白山自然保护区为例  

The Markov model analysis of landscape dynamic: A case researches in Changbai Mountain Natural Reserve

作　　者：韩文权[1] 常禹[1] 

机构地区：[1]中国科学院沈阳应用生态研究所

出　　处：《生态学报》2004年第9期1958-1965,共8页Acta Ecologica Sinica

基　　金：国家自然科学基金资助项目 ( 3 0 0 0 0 0 2 5 ;3 0 3 70 2 66;40 3 3 10 0 8)~~

摘　　要：在 Markov模型假设的基础上 ,利用长白山自然保护区 1975年 MSS、1985年和 1997年 TM卫星遥感数据 ,在遥感图象处理软件和 GIS软件协助下 ,对遥感影像的计算机监督分类结果 (共分为 13类 )进行处理 ,对 Markov模型的可利用性进行分析与检验 ,得出长白山自然保护区景观变化无后效性 ,符合 Markov模型条件。根据 1985～ 1997年转移概率计算步长 10 a(1985～ 1995年 )的转移概率矩阵 ,从 1975年计算 1985年各景观类型的面积与 1985年各景观类型的实际面积值对比 ,计算得 χ2 >χ20 .0 5(12 ) ;再分别用 1975～ 1985年和 1985～ 1997年的转移矩阵计算 1995年和 2 0 4 7年各景观类型的面积 ,分析得χ2 >χ20 .0 5(12 ) ;对两阶段的转移概率矩阵分析得到 χ2 >χ20 .0 5(14 4 ) ;说明两阶段的 Markov转移过程不具同一性 ,属于两个不同的 Markov过程。不同景观类型转移方式对χ2 值的贡献率可以说明其对景观动态的重要性 ,分析结果表明有重要贡献的类型分别为 :阔叶红松林 5 2 .0 0 % >山杨白桦林 2 4 .6 6 % >云冷杉林 11.4 2 % >落叶松林 2 .4 3% ,说明这 4种景观类型的转移方式对长白山自然保护区的景观动态起重要作用 ,尤其以阔叶松林的作用最大 ;同时对Markov models provide excellent tools in detecting and predicting landscape changes. In order to determine whether the Markov models can be used to predict the future landscape pattern in Changbai Mountain Natural Reserve (CMNR), the reserve was classified into 13 landscape types based on remote sensing (RS) imagery, MSS for 1975 and TM for 1985 and 1997, by using maximum likelihood classification algorithm. Therefore, landscape maps for three different decades were produced and exported to GIS for further processing. The study of the consistency of landscape change from 1975 to 1997 showed that the landscape change in CMNR is a stochastic process maintaining the property of independence that corresponds to a Markov process. The transitional probabilities in the period of 1975～1985 and 1985～1997 were obtained by dividing the area of change from one landscape type to another by the area of the landscape type, respectively. The expected area of each landscape type in 1985 was calculated by multiplying the area vector of 1975 by the transitional probability matrix of a 10-year period (1985～1995), which was deduced from the transitional probability matrix of a 12-year period (1985～1997). The expected area of each landscape type in 1995 can be calculated by multiplying the area vector of 1975 by the transitional probability matrix of a 10-year period (1975～1985). The area of each landscape type in 2047 multiplied by the transitional probability matrix from 1985 to 1997 was also predicted. A Chi-square test showed that the differences between the actual area vector and the expected area vector in 1985 and 1995, respectively, are significant at a confidence level of 95%. When combining the transition matrix from 1975 to 1985 with that from 1985 to 1997 the difference is significant at the confidence level of 95%, which indicates that the landscape changes of the two periods correspond to two different Markov processes. The contribution ratio of each landscape type to χ~2 suggests its relevance to the landscap

关 键 词：长白山自然保护区 MARKOV模型 景观动态 

分 类 号：S759.9[农业科学—森林经理学]