四川省黄葛树古树单木冠幅预测模型OA
Predictive Modelling of Single-tree Crown Width of Ancient Ficus virens Aiton Trees in Sichuan Province
利用混合模型理论构建黄葛树古树冠幅模型,为黄葛树古树冠幅预测提供理论依据和实际指导.以四川省7 652株黄葛树古树为研究对象,从8个冠幅-胸径基础模型拟合结果确定最佳基础模型,在最优基础模型中引入单木和土壤因子,构建广义模型,并分析各因子的相对重要性以及对冠幅生长的影响.基于构建的广义模型,构建区域水平混合模型,并通过留一交叉验证方法对模型进行评估,比较各模型预测精度.结果表明,黄葛树古树最佳基础模型为逻辑斯蒂模型;分别引入树龄(AGE)、土壤有机碳含量(SOC)、土壤pH,能够更好地解释四川省黄葛树古树生长差异;引入区域水平随机效应能够显著提高模型预测精度,采用留一交叉验证法对其检验,平均绝对误差(MAE)为2.825 3 m;采用经验最佳线性无偏预测(EBLUP)进行模型校准,随着实测样本量的增加,模型预测性能得到改善.
A mixed model theory was applied to construct a crown width model for the ancient trees Ficus virens trees,with a view to providing theoretical basis and practical guidance for the prediction of the ancient Ficus virens trees.The best base model was determined from the results of fitting eight crown width models using 7 652 ancient Ficus virens trees in Sichuan Province as the study object.Single-tree and soil factors were introduced into the optimal base model to construct a generalised model and analyse the relative importance of the factors and their effects on crown width growth.Based on the constructed gener-alised model,the mixed-effects model theory was used to construct the crown width model,and the con-structed model was tested using the leave-one-out cross-validation method to compare the prediction accu-racy of each model.The results indicate that the best base model for the ancient Ficus virens is the logistic model;tree age(AGE),soil organic carbon content(SOC),and soil pH were introduced separately to bet-ter explain the growth differences of ancient Ficus virens trees in Sichuan Province;The introduction of re-gional level random effects can significantly improve the model prediction accuracy,and the mean absolute error(MAE)of its test using leave-one-out cross-validation is 2.825 3 m;Empirical best linear unbiased prediction(EBLUP)was used for model calibration,and the predictive performance of the model improved with the increase of measured sample size.
林富成;苏月秀;马伟虎;刘凯;李守剑;罗华军;李江江;马文俊
四川省林业和草原调查规划院,四川 成都 610081||四川省林业草原古树名木保护工程技术中心,四川 成都 610081四川省林业和草原调查规划院,四川 成都 610081||四川省林业草原古树名木保护工程技术中心,四川 成都 610081四川省林业科学研究院 森林和湿地生态恢复与保育四川省重点实验室,四川 成都 610081四川省林业和草原调查规划院,四川 成都 610081||四川省林业草原古树名木保护工程技术中心,四川 成都 610081四川省林业和草原调查规划院,四川 成都 610081||四川省林业草原古树名木保护工程技术中心,四川 成都 610081四川省林业和草原调查规划院,四川 成都 610081||四川省林业草原古树名木保护工程技术中心,四川 成都 610081四川省林业和草原调查规划院,四川 成都 610081||四川省林业草原古树名木保护工程技术中心,四川 成都 610081四川省林业和草原调查规划院,四川 成都 610081||四川省林业草原古树名木保护工程技术中心,四川 成都 610081
农业科技
黄葛树古树冠幅模型逻辑斯蒂模型混合效应模型
Ficus virensancient treescrown width modellogistic modelmixed effects model
《西北林学院学报》 2026 (2)
40-48,9
四川省林业和草原调查规划院科研基金项目"四川古树资源精准评价与健康维护关键技术研究"(LGKT2025103)、"四川省古树名木资源价值存续与价值维持(挖掘)关键技术研究及应用"(LGKT202103).
评论