| 基于分层贝叶斯方法的高山松单木生物量模型 |
| |
| 引用本文： | 黄金君,舒清态,王柯人,席磊,孙杨,罗浩.基于分层贝叶斯方法的高山松单木生物量模型[J].西北林学院学报,2022,37(3):126-132. |
| |
| 作者姓名： | 黄金君 舒清态 王柯人 席磊 孙杨 罗浩 |
| |
| 作者单位： | (西南林业大学 林学院，云南 昆明 650224) |
| |
| 摘 要： | 以香格里拉市典型森林生态系统高山松林为对象，在前期进行Ⅰ区和Ⅱ区共115株高山松单木地上生物量实测基础上，以异速生长方程为单木生物量基础模型，并采用分层贝叶斯方法、非线性混合模型法、贝叶斯方法和非线性最小二乘法进行异速生长参数拟合，运用决定系数(R2)、估测精度(E)、均方根误差(RMSE)等指标对模型参数拟合效果进行评价。结果表明:1)从拟合精度看，4种方法的模型拟合效果均较好，R2均达到了0.98以上。但分层贝叶斯方法估计结果更优，其R2=0.985 6，E=84.76%和RMSE=39.75 kg；2)通过对比不同方法的差异发现，加入了区域随机效应的分层贝叶斯方法和非线性混合模型法的拟合效果均优于未加入区域随机效应的贝叶斯方法和非线性最小二乘法。分层贝叶斯方法在拟合高山松单木生物量模型中具有更大优势，模型拟合效果最好。加入了随机效应的分层贝叶斯方法和非线性混合模型法可以提高单木生物量模型的估计精度，采用分层贝叶斯方法进行高山松单木生物量模型参数估测，为大尺度样本数据模型参数估测方法提供新思路。
|
| 关 键 词： | 分层贝叶斯方法 非线性混合模型法 贝叶斯方法 非线性最小二乘法 单木生物量模型 |
Biomass Model of Pinus densata Individual Tree Based on Hierarchical Bayesian Method |
| |
| HUANG Jin-jun,SHU Qing-tai,WANG Ke-ren,XI Lei,SUN Yang,LUO Hao. Biomass Model of Pinus densata Individual Tree Based on Hierarchical Bayesian Method[J].Journal of Northwest Forestry University,2022,37(3):126-132. |
| |
| Authors: | HUANG Jin-jun SHU Qing-tai WANG Ke-ren XI Lei SUN Yang LUO Hao |
| |
| Institution: | (College of Forestry,Southwest Forestry University,Kunming 650224,Yunnan,China) |
| |
| Abstract: | The hierarchical Bayesian method was adopted to estimate the parameters of the Pinus densata individual tree biomass model to provide a new idea for the model parameter estimation of large-scale sample data.P.densata forest,a typical forest ecosystem in Shangri-La City,was used as the research object.Based on the actual measurement of the above-ground biomass of a total of 115 P.densata trees in Zone Ⅰ and Ⅱ,the allometric growth equation was used as the basic model to estimate the individual tree biomass.Furthermore,hierarchical Bayesian method,nonlinear mixed model method,Bayesian method,and nonlinear least square method were adopted to fit allometric growth parameters.The coefficient of determination (R2),estimation accuracy (E),root mean square error (RMSE) and other indicators were applied to evaluate the fitting effect of model parameters.The research results showed that 1) from the perspective of fitting accuracy,the model fitting effects of the four methods were all good,and R2 was above 0.98.However,the hierarchical Bayesian method had better estimation results,with R2=0.985 6,E=84.76% and RMSE=39.75 kg.2) By comparing the differences of different methods,it was found that the hierarchical Bayesian method in which the regional random effects were added and the fitting effect of the nonlinear mixed model method were better than the Bayesian method and the nonlinear least square method without regional random effects.The hierarchical Bayesian method had greater advantages in fitting the P.densata individual tree biomass model,and the model fitting effect was the best.The hierarchical Bayesian method and the nonlinear mixed model method with random effects could improve the estimation accuracy of the single tree biomass model. |
| |
| Keywords: | Hierarchical Bayes method nonlinear mixed model method Bayesian method nonlinear least square individual tree biomass model |
|
| 点击此处可从《西北林学院学报》浏览原始摘要信息 |
| 点击此处可从《西北林学院学报》下载免费的PDF全文 |
|
