摘要
建立了模拟山桉人工林直径生长的动态马尔科夫随机模型。估计模型参数用到了山桉密度试验的直径观测数据。采用多项分对数（ｍｕｌｔｉｎｏｍｉａｌｌｏｇｉｔ）描述直径转移，模型的建立应用了两阶段的广义最小二乘法：第一阶段模拟每一个直径级的直径转移与密度和年龄之间的关系，第二阶段处理第一阶段模型参数随直径级的变动，得到以密度、年龄和径级为自变量预测直径转移概率的紧凑表达式。统计检验表明，应用该模型预测山桉林分直径分布的有效性优于采用正态分布、韦布尔分布或Ｓｂ分布的传统方法并可实际应用于立地指数范围在２２—４４米的山桉人工林分。
A time-dependent Markov model was developed to predict diameter growth of mountain ash(Eucalyptus regnans F.Muell) plantations. Themodel was established using data from a Nelder espacement trial. Thecomponents of diameter growth was based on multinomial logit regression using two-stage generalized least square procedures: The first stage comprising regressions of diameter transition probabilities against logarithm on stocking and age for each diameter class and the second relating the first stagecoeficients to diameter. Comparisons between this model and conventionalmethods (variously Normal, Weibull and Sb distributions) were made and it was showed that the time-dependent Markov model was markedly acclurater than the others. This time-dependent Markov model could be used to predict diameter growth for plantations with site index range 28-44m.
出处
《林业科学》
CAS
CSCD
北大核心
1994年第4期338-345,共8页
Scientia Silvae Sinicae
关键词
马尔科夫
随机过程
直径生长
林木
Markov process, Diameter growth, Paramrter estimation,Multinomial logit, Generalized least squares
