尾巨桉林分生长对间伐和修枝响应的分析与预测  

Analysis and Prediction of Eucalyptus urophylla×E.grandis in Response to Thinning and Pruning

作　　者：张士韬 欧阳林男 陈少雄[1] 杨嘉麒 ZHANG Shi-tao;OUYANG Lin-nan;CHEN Shao-xiong;YANG Jia-qi(Research Institute of Fast-growing Trees,Chinese Academy of Forestry,Zhanjiang 524022,Guangdong,China;College of Forestry,Nanjing Forestry University,Nanjing 210037,Jiangsu,China)

机构地区：[1]中国林业科学研究院速生树木研究所,广东湛江524022 [2]南京林业大学林学院,江苏南京210037

出　　处：《林业科学研究》2024年第2期48-59,共12页Forest Research

基　　金：“十四五”国家重点研发计划项目(2023YFD2201001)。

摘　　要：[目的]探究不同强度的间伐与修枝对尾巨桉林分生长的影响,建立BP神经网络模型并验证模型对间伐和修枝处理下尾巨桉林分生长的预测作用,为尾巨桉的高效经营技术提供理论指导。[方法]以南方国家级种苗示范基地20%、40%和60%间伐与38.18%、42.39%和48.18%修枝强度的尾巨桉为对象,在处理后连续7 a调查林分生长指标,计算累积增量,分析不同间伐和修枝处理对林分生长增量的影响,并运用BP神经网络建立了5种林分生长累积增量对间伐和修枝响应的预测模型,以均方根误差、Kappa值和Pearson相关系数对模型预测效果进行对比,判断最优模型。[结果]间伐促使尾巨桉林胸径、冠幅和单株材积的增长,但不促进树高增长和出材量增加,60%间伐的样地尾巨桉胸径和单株材积增量最高,20%间伐的冠幅增量最大,不间伐的树高增量最高。修枝促进尾巨桉胸径增长,对树高和蓄积量增长无影响,38.18%修枝的胸径增量最高,60%间伐+48.18%修枝是处理中有助于尾巨桉林分生长的组合。间伐和修枝均能促进尾巨桉林分径阶分布右偏,但修枝的效果不如间伐显著。综合来看,隐含层节点数为4的尾巨桉BP神经网络模型预测结果的均方根误差最低,Kappa系数和r值最高,可预测7 a内的间伐和修枝效果。[结论]间伐和修枝均显著促进尾巨桉林分生长和径阶分布右偏。高强度间伐和修枝相结合更有助于尾巨桉人工林生长和大径材培育。合理的BP神经网络模型能准确地预测间伐和修枝对林分生长的促进效果,是林分生长预测的优异模型。[Objective]To explore the effects of thinning and pruning with different intensities on the growth of Eucalyptus urophylla×E.grandis,the BP neural network model was developed to predict the growth of E.urophylla×E.grandis under thinning and pruning treatment,for providing theoretical guidance for efficient management technology of E.urophylla×E.grandis.[Method]Taking 20%,40% and 60% thinning and 38.18%,42.39% and 48.18% Based on the E.urophylla×E.grandis in the Southern National Forest Seedling Demonstration Base with treatments of 20%,40% and 60% thinning and 38.18%,42.39% and 48.18% pruning,the stand growth indexes were investigated for 7 years after treatment,and the effects of different thinning and pruning treatments on the growth increment of the stands were analyzed.Furthermore,BP neural network was used to predict the response of cumulative increment to thinning and pruning.Root-mean-square error,Kappa and Pearson correlation coefficient were used to compare the prediction effect of models,and the optimal model was determined.[Results]Thinning treatment promoted the growth of DBH,crown width and tree volume,but did not promote the growth of height and wood production.The increment of DBH and tree volume was the highest in the stands with 60% thinning intensity,the increment of crown width was the highest in the stands with 20% thinning intensity,and the increment of height was the highest in the control stands.Pruning promoted the growth of DBH,but did not promote the growth of height and wood production.DBH increase was the highest in the stands with 38.18% pruning intensity.The treatment with 60% thinning and 48.18% pruning was conducive to the growth of E.urophylla×E.grandis.Both thinning and pruning could promote the right-sided distribution of diameter class,but the effect of pruning was not significant.In summary,the BP neural network model with 4 nodes in the hidden layer had the lowest root-mean-square error and the highest Kappa value and r value.[Conclusion]Thinning and pruning can signific

关 键 词：间伐 修枝 尾巨桉 BP神经网络模型 林分生长 

分 类 号：S792.39[农业科学—林木遗传育种]