Modeling Biomass Of Pinus Kesiya Var.Langbianensis Natural Forest Under The Background Of Climate Change

Modeling biomass is an important content of research on forest biomass, and both the direct simulation and biomass conversion are the two main aspects of biomas modeling. At present many biomass models had been constructed which include some models using modern mathematical methods. But some problems were found about the research on biomss modeling, such as less models considering the environmental factors,and lack of biomass growth models, and so on. So in the study, the biomass of45plots and128sample trees had been investigated at three typical sites (Tongguan town of Mojiang county, Yunxian town of Simao district, and Nuofu town of Lancang) of Simao pine (Pinus kesiya var. langbianensis) natural forest in Pu’er city, Yunnan province. Then the biomass components mixed models considering fixed effects of environmental factors at both individual tree and stand level, biomass conversion factors (BEFs) of biomass components and root:shoot ratio (R) at the stand level, and aboveground biomass components growth mixed models considering fixed effects of environmental factors at the individual tree level were constructed. Then the best models of the biomass components were selected based on the anlysis of fitting and validation indices.Firstly, based on the power function, it had been constructed the models of the biomass components and aboveground growth of individual tree using the technology of nonlinear mixed effect model considering random effects of the regional effect and fixed effects of environmental factors, such as topographic factors, climate factors, and competition factor. Secondly, at the stand level, considering random effects of the regional effect and fixed effects of environmental factors,such as stand factors, topographic factors and climate factors, it had been constructed the models of the biomass components of tree layer and stand, using the technology of nonlinear mixed effect model. Thirdly, based on the analysis of the BEFs and R, it had been constructed the models including topographic factors, climate factors, and competition factor respectively. Finally, based on the modified Richards function, it had been constructed the growth models of the aboveground components for the individual tree using the technology of nonlinear mixed effect model considering random effects of the regional effect and fixed effects of environmental factors,such as topographic factors, climate factors, and competition factor. The results showed that:(1) Not all mixed effects models considering random effect of regional effect could improve the fitting accuracy of models, some mixed models, such as the bark biomass mixed model of individual tree, and the mixed growth models of single tree branches and leaves of individual tree could not improve the accuracy of model fitting. Meanwhile, the mixed effects models usually had better performance on the model validation. (2) Most mixed models considering fixed effects of environmental factors were better performance of model fitting and validation than the models only considering the random effect of the regional effect. But it was different at the biomass components for the best mixed model considering fixed effects of environmental factors.(3) Most models considering the variance and (or) covariance structures had better performance on model fitting than the ordinary models, but for the different components, they best variance or covariance structure were different. For the models of BEFs and R, it could not improve significantly the fitting accuracy of the models considering the variance structure; Considering the covariance structure of time autocorrelation could not improve the fitting accuracy of the biomass growth models of individual tree; The variance form of most best models were power function, and among the better models which cound imorove the accuracy considering the covariance structure, the covariance structure of the Spherical form was more to the single tree level, but the Gausian form to the stand level.(4) Based on the comprehensive consideration of the indices of model fitting and validation, the better (or) best models for the different components are as follows:Firstly, for the single tree biomass models, the best models for wood, root and total biomass were the mixed effect models including the fixed effects of topographic factors and random effect of regional effect; and the best model for branch biomass was the mixed effect model including the fixed effects of competition factor and random effect of regional effect; and the mixed model including climate factors and random effect of regional effect was the best for the aboveground and leaf biomass; but for the bark biomass, the better models was the model including climate factors.Secondly, for the stand biomass models, the best model for root biomass of stand was the mixed model only considering the random effect of regional effect, but for the other components (including aboveground biomass of tree layer and stand, the total biomass of layer and stand, and the root biomass of tree layer), their best models were the mixed effect models including the fixed effects of topographic factors and random effect of regional effect.Thirdly, for the BEFs and R models of stand, the best BEFs models for branches, leaves, aboveground of tree layer, root, and total biomass were the models including the topographic factors; for the aboveground. root, total biomass of stand and wood biomass, the best BEFs models were the regression models considering the covariance structure. The best biomass R model was also the regression models considering the covariance structure.Finally, for the aboveground biomass growth models of individual tree, the growth models considering the fixed effect of competition factors and random effect of regional effect were best models for wood and aboveground biomass, the best bark biomass growth model was the mixed model considering the fixed effect of topographic factors and random effect of regional effect, and the best growth models of branch and leaf biomass were the regression models considering competitive factor.