基于结构方程模型的森林健康评价

Evaluation of forest health based on structural equation model

指标权重是影响森林健康评价结果的关键因素。探索更科学的指标权重确定方法,为森林健康评价提供全新的思路,也为提升森林评价结果的公认性提供理论依据和方法基础。将林分作为研究尺度,以研究区杉木纯林、杉木混交林和天然次生林3种典型森林为研究案例,依据复杂系统思想构建森林健康评价指标体系,将森林组织结构、生产力、土壤状况和稳定性4个准则层因子作为一阶因子,森林健康作为4个要素的二阶因子,依据传导机制建立森林健康评价结构方程模型进行指标权重确定,并对其健康进行评价。结果表明:1)构建的森林健康评价结构方程模型平均绝对适配度指数χ2/df为1.793,介于1~3,RMSEA为0.079,增值适配度指数CFI、NFI和TLI的数值分别为0.943、0.908和0.929,假设模型与观测数据适配,综合来看模型整体拟合效果良好,可满足研究需要。2)森林健康评价体系中4个准则层因子——生产力、组织结构、稳定性和土壤状况的权重分别为0.2515、0.2263、0.2277和0.2944,说明土壤状况是决定森林健康的关键要素。3)研究区3种森林健康评价的综合分值处于2.1772~4.1708,健康等级集中在健康、亚健康和中健康3种状态,总体上,3种森林的健康等级为天然次生林>杉木混交林>杉木纯林。结构方程模型具有充分提取原始数据信息和克服多重共线性影响等方面的优势,采用结构方程模型进行森林健康评价指标的赋值思路是科学合理的,可为森林健康评价提供一个全新的思路,研究区3种森林的健康评价结果较客观地反映了3种森林的健康现状。

Index weight is a key factor in the evaluation of forest health. Exploring a scientific method for determining the index weight provides a new approach to evaluate forest health and a theoretical basis to improve the recognition of the evaluation of forest health. Here,we developed a structural equation model for assessing forest health of three typical forest types(Chinese fir pure forest,Chinese fir mixed forest,and natural secondary forest),in which forest structure,forest productivity,soil condition,and stand stability were set up as the first-order factors and forest health as the second-order factor. The index weight in the structural equation model was determined based on the conduction mechanism. The structural equation model was used to assess the health status of the three forest types. The results showed that:(1) For the structural equation model,the chi square degree of freedom(χ2/df) was 1.793,ranging 1-3;the root mean square error of approximation(RMSEA) was 0.079;the comparative fit index(CFI),normed fit index(NFI) and Tacker-Lewis index(TLI) were 0.943,0.908 and 0. 929,respectively. The modelling results were compatible with the observed data. The structural equation model ran well and satisfied research requirements.(2) The index weights in the forest health evaluation system were 0.2515,0.2263,0.2277,and 0.2944 for forest productivity,forest structure,stand stability,and soil condition,respectively,indicating that soil condition was the key factor driving forest health.(3) The comprehensive scores of forest health for the examined three forest types ranged2.1772-4. 1708,indicating that the health status of the three forests were in the health,subhealth and medium-health status. In general,the health degree of the three forests ranked as natural secondary forest > Chinese fir mixed forest > Chinese fir pure forest. Our findings suggest that the structure equation model had advantages of extracting all information from original data and overcoming the multicollinearity uncertainties. The structural equation model provided a scientific basis and reasonable approach for the assessment of forest health. Results of health assessment showed that the three forest health evaluation indices could objectively reflect forest health status.