限定混合模型模拟不规则和多峰直径结构分布——以长白山云冷杉林为例  

Simulation of multi-peaks and irregular diameter distributions by finite mixture models: a case study on spruce-fir forests in Changbai Mountain

作　　者：杜志[1] 亢新刚[1] 岳刚[1] 徐光[2] 赵东宁[2] 杨艺军 高祥[2] 

机构地区：[1]北京林业大学省部共建森林培育与保护教育部重点实验室,北京100083 [2]汪清林业局,吉林汪清133200

出　　处：《中南林业科技大学学报》2013年第4期43-49,共7页Journal of Central South University of Forestry & Technology

基　　金：国家"十二五"科技支撑项目--阔叶红松林和云冷杉过伐林可持续经营技术与示范(2012BAD22B2);国家林业局948项目(2013-4-66)

摘　　要：以长白山地区6块云冷杉林样地调查数据为基础,采用限定混合Weibull模型方法,并利用传统的负指数模型和Weibull模型作为参照,对直径分布进行了模拟。结果表明:样地直径分布呈多峰、rotated-sigmoid等不规则分布;数学统计表明限定混合Weibull模型的直径拟合效果最好,其次是Weibull模型,最后为负指数模型,尤其在模拟双峰和rotated-sigmoid分布时混合模型优势明显;残差图显示径阶大于28cm后,相对于其它两个模型,限定混合Weibull模型的残差值更接近于0。限定混合模型是一种较好的具有高灵活性和高精度的林分直径模拟方法。Based on the investigation data from six example plots in Changbai Mountain in northeastern China, the finite mixture Weibull model was applied to simulate the tree diameter distribution with. the results of traditional methods of negative exponential model and Weibull model as the comparisons. The results show that the tree diameter distributions of six plots wei＇e presented irregular shapes of multi-peaks, rotated-sigmoid, etc.. The statistics show the fitting effect of finite mixture Weibull model was the best, especially in modelling bimodal and rotated-sigmoid distributions, then was the We ibull model＇s, and negative exponential model＇s was the last. The residual figure displays the residual value of finite mixture model more close to zero than others. The finite mixture model was a better alternative method, with high flexibility and precision to model tree diameter distribution.

关 键 词：云冷杉林 直径分布 限定混合Weibull模型 负指数模型 Weibull模型 

分 类 号：S758.5[农业科学—森林经理学] S791.14[农业科学—林学]