Research And Construction Of Stem Taper Equation Of Chinese Fir Plantation Based On Climatic Factors

The taper is an important index to describe the trunk shape,and the taper equation has become the preferred method for compiling the timber yield table.At present,the research on the traditional taper equation has become mature,but today’s climate is changing,and climatic factors are one of the important factors affecting the growth of trees.However,there are few studies on the influence of climatic factors on trunk taper in China.Therefore,it is of great significance to construct a trunk taper equation including climatic factors for both theoretical research and production practice.In this paper,the trunk taper of Chinese fir plantation in Hunan Province was taken as the research object.Based on the measured data of 75 Chinese fir plantation plots in Yongzhou City,considering the influence of climatic factors,stand density factors and forest land factors on the trunk taper of Chinese fir,the nonlinear mixed effect model method was used to construct the trunk taper equation of Chinese fir plantation with single factor,nested two levels and nested multi levels.At the same time,four machine learning methods(random forest,support vector machine,K-nearest neighbor method,multi-layer perceptron)were used to construct the stem taper equation of Chinese fir plantation,and the feasibility of using machine learning methods to construct the stem taper equation was discussed.The mixed effect model and machine learning model of stem taper of Chinese fir plantation were compared.The main results are as follows:(1)The quantitative method I in Forstat was used to screen the climatic factors and forest land factors that had significant effects on the trunk taper of Chinese fir.Among them,the climatic factors that had a significant effect on the taper of Chinese fir were the growth accumulated temperature(DD5)and the average maximum temperature in summer(Txmax).The forest land factor is elevation(HB).(2)The nonlinear mixed effect model method was used to construct the stem taper equation of Chinese fir plantation with single factor,nested two levels and nested multi levels.Among the four alternative basic models,Kozak(2004)is the optimal basic model,and its adjusted coefficient of determination Ra2 is 0.9595.In the mixed effect model of stem taper of Chinese fir plantation,the best fitting effect was the nested multi-level stem taper equation of Chinese fir plantation,with Ra2 of 0.9863,which was about 2.8%higher than the basic model.RMSE and MAE were 0.5335 and 0.3527,respectively,which were about 40.4%and 46.5%lower than the basic model.In the stem taper equation of Chinese fir plantation with single factor,the model with stand density factor increased the most,and its Ra2 was 0.9710.In the stem taper equation of Chinese fir plantation with nested two levels,the model with climate factor and stand density factor has the largest increase,with Ra2 of 0.9816.(3)Four machine learning methods(random forest,support vector machine,K-nearest neighbor method,multi-layer perceptron)were used to construct the stem taper equation of Chinese fir plantation.Among the machine learning models of stem taper of four Chinese fir plantations,the multi-layer perceptron model had the best fitting effect,with Ra2 of 0.9599.RMSE and MAE were 0.8872 and 0.6661,respectively.(4)The mixed effect model and machine learning model of stem taper of Chinese fir plantation were compared.Through comparison,it is concluded that the Ra2 of the mixed effect model(Ra2=0.9863)and the machine learning model(Ra2=0.9599)with the best fitting effect are better than the basic model(Ra2=0.9595).In these two types of models,the mixed effect model has higher fitting accuracy,but it has the disadvantages of complex model structure and difficult model parameter estimation.In contrast,the machine learning model is relatively simple in structure,does not need to estimate the parameters,can quickly process and analyze a large amount of data,and the model accuracy is relatively high.Therefore,in the actual construction of the taper equation,it is recommended to use the mixed effect model method to construct the taper equation when the model accuracy is high.When a large amount of data needs to be processed,it is recommended to use machine learning methods to construct the taper equation.