文献类型：期刊文献

中文题名：基于随机森林算法的桉树人工林单木生物量预估模型

英文题名：Single TreeBiomass Prediction Model of EucalyptusPlantationsBased on Random Forest Algorithm

作者：宋杰[1] 赵俊[1] 何普林[1] 成雅君[2] 黄润霞[2] 竹万宽[2]

第一作者：宋杰

机构：[1]中林集团雷州林业局有限公司,广东湛江524043;[2]中国林业科学研究院速生树木研究所/广东湛江桉树林生态系统国家定位观测研究站,广东湛江524022

年份：2024

卷号：41

期号：2

起止页码：11-16

中文期刊名：桉树科技

外文期刊名：Eucalypt Science & Technology

收录：CSTPCD

基金：广东省林业科技创新项目(2022KJCX020);广西科技计划项目(桂科AB23026010);广东湛江桉树林生态系统国家定位观测研究站运行补助项目(KS2024160017);林业生态监测网络平台运行项目数据采集(2024CG232)。

语种：中文

中文关键词：桉树;单木生物量;岭回归模型;异速生长模型;随机森林算法

外文关键词：Eucalyptus;single tree biomass;ridge regression model;allometric growth model;random forest algorithm

分类号：S758.1

摘要：单木生物量模型是估测森林生物量的基础。通过标准木法实测雷州半岛地区90株桉树单株生物量数据,随机划分60个样本数据作为训练集,30个样本数据作为验证集。以林龄、树高和胸径为自变量,单木生物量为因变量,使用岭回归模型、异速生长模型和随机森林算法构建模型,采用决定系数(R^(2))、均方根误差(RMSE)和平均绝对误差(MAE)对模型进行评价。结果表明:随机森林模型的R~2、RMSE和MAE无论在训练集还是验证集均高于岭回归模型和异速生长模型。由随机森林模型的因子重要值可知,胸径是影响单木生物量的主要因子。引入林龄因子后的随机森林模型可以提高单木生物量的预测精度,为碳汇计量提供基础数据和模型支撑。
Single-tree biomass models serve as foundations for estimating forest biomass.In order to develop such a model,biomass data from 90 Eucalyptus trees in the Leizhou Peninsula area were measured using the standard wood method.Date from sixty of these sample trees were randomly selected as the training set,and the other 30 sample trees were used for model validation.The model used ridge regression,heteroscedastic growth,and random forest models with stand age,tree height,and diameter at breast height as independent variables and single tree biomass as the dependent variable.The model's performance was evaluated using the coefficient of determination(R^(2)),root mean square error(RMSE),and mean absolute error(MAE).The study found that the random forest model outperformed the ridge regression model and the heteroskedastic growth model in both the training and validation sets,as evidenced by higher values of the R2,RMSE,and MAE.The factor importance values of the random forest model indicated that diameter at breast height was the most significant factor affecting single tree biomass.In summary,the introduction of the forest age factor can improve the prediction accuracy of single-tree biomass in the random forest model.This provides basic data and model support for carbon sink measurement.