首页>>学术检索

哑变量在燕山地区华北落叶松人工林生长模型中的应用1)

标题: 哑变量在燕山地区华北落叶松人工林生长模型中的应用1)
英文标题: Application of Dummy Variable in the Growth Model of Larix principis-rupprechtii Plantations in Yanshan Re-gion
作者: 王勇,王鑫梅,牟洪香,孙圣,杨明明
英文作者: Wang Yong,Wang Xinmei,Mou Hongxiang,Sun Sheng,Yang Mingming
出版时间: 2014-01-01
机构: 河北农业大学
关键词: 华北落叶松人工林,生长模型,非线性回归,哑变量
英文关键词: Larix principis-rupprce htii plantation,Growth model,Nonlinear regression,Dummy variable
刊名: 东北林业大学学报
英文刊名: Journal of Northeast Forestry University
卷号: 10
基金: 1)河北省林业科学技术研究项目
页码: 7
摘要: "以燕山地区华北落叶松人工林为研究对象,在收集大量临时样地和解析木资料的基础上,筛选出合适的数据,根据承德市隆化县林管局、木兰林管局和塞罕坝机械林场3个地方的地区特征引入哑变量的概念,利用非线性回归的方法对华北落叶松人工林胸径、树高、材积生长过程进行拟合,结果显示3个地区华北落叶松人工林的胸径、树高、材积生长量没有显著差异。基于哑变量建立了3个地区华北落叶松人工林的胸径、树高、材积生长模型。结果表明:含有哑变量的Schumacher模型对华北落叶松人工林胸径、树高生长过程的拟合效果最好,决定系数R2分别达到了0.9941和0.9929,残差平方和分别为4.7339和6.8849;含有哑变量的二次函数对华北落叶松人工林的材积生长过程拟合最好,决定系数R2为0.9928,残差平方和为0.0023。通过对所建立生长模型的适应性检验,结果表明残差分布比较均匀,预估精度都达到95%以上,说明模型的预估效果比较好。"
英文摘要: "Larix principis-rupprechtii plantations were selected as study example in Yanshan region .On the basis of mass data about temporary sample plots and analytical trees , screened appropriate data , the concept of dummy variable was introduced ac-cording to the different areas character of Longhua County , Weichang County and Saihanba Machine Forest Farm .Nonlin-ear regression was used to simulate the models of DBH , tree height and volume of L.principis-rupprechtii plantations .The increments of DBH , tree height and volume with dummy variable of Larix principis-rpu prechtii plantations had no significant difference in the three regions .The growth models of DBH ,tree height and volume were established based on dummy varia -ble.The dummy variable Schumacher equation was the best model of the DBH and tree height , the determination coeffi-cient R2 , respectively, achieved 0.994 1 and 0.992 9, the residual sum of squares achieved 4.733 9 and 6.884 9, the dum-my variable quadratic function was the best model of the volume with the determination coefficient R2 of 0.992 8, and the residual sum of squares of 0.002 323.With the independent test of established models , the residuals distributed equality and the accuracy was above 95%."