基于平均优势木胸径的湖南杉木人工林立地质量评价模型

Site quality evaluation model of Chinese fir plantation in Hunan based on the DBH of average dominant trees

【目的】研究立地因子对湖南杉木平均优势木胸径生长的影响,构建含立地随机效应的立地质量评价模型,以解决杉木立地质量评价的问题。【方法】基于湖南杉木人工林509块样地实测数据,在统计之林软件中使用数量化方法Ⅰ筛选出对杉木平均优势木胸径生长影响显著的立地因子,将立地因子按照标准分级、组合,构建初始立地类型;通过R语言拟合杉木人工林平均优势木胸径与年龄的相关关系,从4种基础生长模型中筛选出最优基础模型;应用K-means聚类将参数值相近的立地类型聚类成立地类型组,并将聚类后形成的立地类型组作为随机效应加入最优基础模型中,构建混合效应模型;通过导算,得到立地质量评价模型。【结果】1)对杉木平均优势木胸径产生显著影响的因子包括海拔、坡位、土壤类型和土壤厚度,影响程度从高到低的顺序为:坡位>土壤类型>土壤厚度>海拔。2)4个候选基础生长模型的拟合精度均较低(R2=0.6415~0.6420),选取单分子式(Mitscherlich)(R2=0.6420)作为构建立地质量评价的基础模型。3)考虑立地效应对模型的影响,将不同立地因子及其组合作随机效应进行非线性混合效应模拟,确定系数R2从0.6415~0.6420提高到0.6642~0.8258,拟合精度的高低与主导立地因子的显著性紧密相关,其中含立地类型的混合模型模拟精度最高(R2=0.8258)。4)以确定系数≥0.95为聚类精度标准将立地类型划分为10个立地类型组,含立地类型组的混合模型在便于应用的同时提高了建模的精度(R2=0.8409)。【结论】立地因子对杉木平均优势木胸径生长具有显著影响。同时,将立地因子作为随机效应加入混合模型中能够提高预测的准确性,更适合于多种立地类型下的立地质量评价。

【Objective】This study aims to investigate the impact of site factors on the diameter at average breast height dominant Chinese fir trees in Hunan province and develop a site quality evaluation model with site random effects to address the issue of site quality assessment.【Method】Based on data collected from 509 plots of Chinese fir plantations in Hunan province,significant site factors influencing the DBH growth of average dominant fir trees were identified using quantitative method I in statistical forest software.Initial site types were then established by classifying and combining these influential factors according to standard criteria.The correlation between DBH and age of average dominant trees was fitted using R language,and the optimal basic growth model was selected from four candidate models.K-means clustering was employed to group similar parameter values into distinct site type groups,which were subsequently added as additional random effects into the optimal basic model,resulting in a mixed-effects evaluation model for assessing site quality.【Result】1)Elevation,slope position,soil type,and soil thickness were found to influence the DBH of average dominant Chinese fir trees,with slope position exerting the greatest impact followed by soil type,soil thickness,and elevation.2)Amongst the four candidate basic growth models tested,their fitting accuracy was relatively low(R2=0.6415-0.6420).In this study,we selected Mitscherlich molecular formula(R2=0.6420)as the basic model for evaluating site quality.3)The influence of site effects on our model predictions,nonlinear mixed-effect simulations were conducted incorporating different combinations of individual site factors and their effects;this resulted in an improved determination coefficient ranging from 0.6642-0.8258 compared to previous values ranging from 0.6415-0.6420.The fitting accuracy was closely associated with the significance of the dominant site factor.The simulation accuracy of the mixed model incorporating site type exhibited the highest value(R2=0.8258).4)Site types were categorized into 10 groups based on a clustering accuracy standard with a determination coefficient≥0.95.The mixed model utilizing site type groups proved to be convenient for practical application and enhanced modeling accuracy(R2=0.8409).【Conclusion】Site factors exerted significant effects on the diameter at breast height growth of average dominant Chinese fir trees.Additionally,incorporating site factors as random effects into the hybrid model improved prediction precision,rendering it more suitable for evaluating site quality across diverse site types.