文献类型：期刊文献

英文题名：Comparing crown ratio models for spruce-fir broadleaved mixed forests using beta regression and random forest algorithm

作者：Yu, Yang[1] Zhou, Zeyu[1] Sharma, Ram P.[2] Zhang, Lianjin[1] Du, Manyi[1] Zhang, Huiru[1]

第一作者：Yu, Yang

通信作者：Zhang, HR[1]

机构：[1]Chinese Acad Forestry, Expt Ctr Forestry North China, Beijing 102300, Peoples R China;[2]Tribhuwan Univ, Inst Forestry, Kathmandu 44600, Nepal

年份：2024

卷号：225

外文期刊名：COMPUTERS AND ELECTRONICS IN AGRICULTURE

收录：;EI(收录号：20243316863245);Scopus(收录号：2-s2.0-85200818999);WOS:【SCI-EXPANDED(收录号:WOS:001293627800001)】；

基金：Funding for this research was provided by the National Key Research and Development Program (NKRDP) of Ministry of Science and Technology of the People's Republic of China and specially supported by the project "Full-cycle Multifunctional Management Technology for Deciduous Broad-leaved Secondary Forests" (grant number 2022YFD2200503) funded by the Experimental Centre of Forestry in North China, Chinese Academy of Forestry and by the Fundamental Research Funds for the Central Non-profit Research Institution of CAF in China (CAFYBB2023MA030) .

语种：英文

外文关键词：Competition indices; Model system; Stand-level variables; Tree-level variables; Tree species-specific groups; Variables selection

摘要：Tree crown measures, such as crown ratio (CR) can be used as one of the important predictor variables in forest growth and yield models. CR is influenced by the tree and stands factors, site and climate factors, and therefore CR can be modeled as a function of such factors. We developed the CR models by applying the beta regression and random forest (RF) algorithm with the important factors used as predictor variables. We used the long-term experiment plot data from 10,069 trees of the spruce-fir broadleaved mixed forests in Northeast China. We categorized data into four tree species-specific groups (Spruce-Fir, Hard-broadleaved, Soft-broadleaved, and Coniferous groups) based on the similarity of the species. The CR models developed with the RF algorithm outperformed the models developed with the beta regression. Both model types showed the best performance for the Spruce-Fir species-specific group. Among the predictor variables evaluated, diameter at breast height and height to crown base contributed mostly to the CR model, followed by variables describing competition, stand development, and site productivity. Our CR model can be applied with high accuracy to mixed forests and can provide a reliable basis for regulating stand structure and promoting forest quality.