建立区域性含立地因子的杉木人工林非线性立地指数混合模型,为预测杉木树高生长提供参考依据。基于4个省份45株优势木307组圆盘解析木数据,选用5种常见树高-年龄生长方程,采用数量化理论I和非线性混合模型等方法,以随机效应构建含初始立地类型的非线性立地指数混合模型,用K-means聚类进一步构建含立地类型组的非线性立地指数混合模型。结果表明,海拔、土壤类型、坡向、坡位为显著影响树高生长的立地因子;Gompertz模型为拟合效果最好的基础模型,R²为0.653;含初始立地类型的非线性立地指数混合模型与基础模型的相比,精度提升了28.02%;含立地类型组的非线性立地指数混合模型与基础模型相比,R²提升了37.21%;模型经交叉检验,基于初始立地类型及聚类后的模型精度较高,证实了模型的实用性。建立非线性立地指数混合模型为区域范围尺度下杉木人工林立地质量评价及立地指数表编制提供了参考。

A regional nonlinear mixed site index model of Cunninghamia lanceolata plantations with site factors was established to provide a reference for predicting the height of Cunninghamia lanceolata in the region.This study is based on data from the analysis of 307 disc groups from 45 dominant trees across 4 provinces,selecting five common tree height-age growth equations.Using quantitative theory I,nonlinear mixed model and other methods,the mixed model of nonlinear site index with initial site type was constructed by using random effects.Finally,the mixed model of nonlinear site index with site type group was further developed by K-means clustering.The results showed that altitude,soil type,aspect,and slope position were the site factors that significantly affected the tree height.The Gompertz model was selected as the basic model,showing the best fitting effect,with an R² of 0.653.Compared with the basic model,the accuracy of the nonlinear site index model with the initial site type is improved by 28.02%.Compared with the basic model,the R² of the nonlinear site index mixed model with site type group was increased by 37.21%.After cross testing of the model,the accuracy of the model was high based on the initial site type and clustering,which confirmed the practicability of the model.The establishment of a nonlinear site index mixed model provides a feasibility for the site quality evaluation and compilation of site index tables for Cunninghamia lanceolata plantations at the regional scale.