Research On Basal Area Growth Model For Oak Natural Forest In Hunan Province

The stand growth and harvest simulation of natural mixed forest is still in the early stage,and it is not yet unified modeling method.With the increasing importance of natural mixed forest in national economy and ecological protection,it is urgent to study suitable models for the growth prediction and management of mixed forest.The stand structure of natural mixed forest determines that the expression of stand age,tree species composition,site index and competition effect must be solved when the mixed forest growth model is built.In view of this,the study will consider influence of forest type and site type effects,and take basal area growth model as the core to construct oak forest stand growth model by using nonlinear mixed effect model method.It's aim to built a new set of basal area growth model constructing method,and provide reference and theoretical basis for the stand growth harvest and management of natural mixed forest.This paper will take the oak natural forest in Hunan province as the research object,and use nonlinear mixed effect model method to construct site index model of oak natural forest with random effect.By considering the differences of natural mixed forest tree species and site types,the study takes the site index as driving variables to built basal area growth model of oak natural forest in Hunan province from three levels of the whole stand,single tree and stand storey.The main research contents and conclusions are as follows:(1)The index model of natural forest was constructed.Three climate factors including mean annual temperature(MAT),mean annual precipitation(MAP)and mean temperature difference(TD)of influencing the growth of forest average dominate height were screened out by using multivariate stepwise regression analysis.The climate-sensitive site index model is constructed by model re-parameterization.On this basis,this study built site index model with random effects by considering the influence of forest type and site type effects,and calculated the site index of different sample plot in standard age of 20 years.Compared with the climate-sensitive ground index model,the determination coefficient(R2)was increased by 106.72%,the mean absolute error(MAE)was decreased by 68.47%,and the root mean square error(RMSE)was decreased by 58.66%.(2)The Richards and Schumacher models were used to simulate the growth of stand basal area.The results showed that the Schumacher model with stand age,average dominate height and stand density index had the best simulation effect.On this basis,this study built nonlinear mixed effect model with forest type and site type effects.According to the results of random effect parameter construction and precision evaluation,the optimal form parameter construction of three kinds of random effect was determined,and the model fitting effect of different random effects was compared.Compared with the basic model,the determination coefficient(R2)was increased by 6.04%,the mean absolute error(MAE)was decreased by 49.11%.and the root mean square error(RMSE)was decreased by 64.24%.(3)It proposed a method of basal area growth model for oak natural secondary forest in Hunan province based on storey identification,and built storey basal area growth model.After the storey identification of using whole tree height clustering,IUFRO standard and TRSRAT,the model of Schumacher was used to fit the basal area of whole stand,upper layer and lower layer.It's the coefficient of determination(R2)were all above 0.92.Compared with the whole stand basal area growth model without storey identification,the coefficient of determination(R2)of different forest layer with storey identification was increased from 0.9259 to 0.9455?0.9846.The accuracy of the basal area growth model based on IUFRO standard was the highest(R'=0.9704,MAE=1.4588,RMSE=2.1786).Compared with the whole stand basal area growth model without storey identification,the determination coefficient(R2)was increased by 5.0%,the mean absolute error(MAE)and the root mean square error(RMSE)was respectively decreased by 39.8%and 26.4%.The compatible storey basal area growth model was built by adjustment in proportion,compared with the independent prediction model of each forest layer,the mean absolute error(MAE)and the root mean square error(RMSE)of upper layer was respectively decreased by 75.36%?68.01%,the mean absolute error(MAE)and the root mean square error(RMSE)of lower layer was respectively decreased by 79.42%.?78.04%?(4)The results of factor screening by multivariate stepwise regression analysis showed that single tree height(H),stand average breast diameter(Dg),annual heat:moisture index(AHM),site index(SI)and richness index(Gleason)had a significant effect on single tree basal area.By using model re-parameterization method,this study based on the maximum parameter A of tree growth,and built the re-parameterization basal area growth model of single tree.On the basic,the random effects of stand type effect,stand type effect and tree hierarchy effect were taken as random effects respectively,and the basal area growth model of single tree with them were built.The results showed that the basal area growth model of single tree with the clustering of site types had optimal simulation effects.Compared with the basic model,the determination coefficient(R2)was increased by about 26.63%,the mean absolute error(MAE)was decreased by 31.40%,the root mean square error(RMSE)was decreased by 54.23%.(5)It is proved that the random effect is scientific and reasonable to construct the basal area growth model of natural mixed forest.The simulated results can be used as the basis for the stand growth harvest and management of natural mixed forest.