Estimation Of Forest Ecosystem Carbon Budgets Based On Non-linear Regression And Parameter Estimation On Beichanshan Island

Regression analysis is one of the important contents in mathematical statistics. The research on its theory and methods have attracted lots of attention, because it has an extensive application. In recent years, the study of nonlinear regression analysis has become more and more important. On the one hand, the study of linear regression analysis was perfect. on the other hand, the model of used or experimental were almost nonlinear in the actual application. However, the model of nonlinear regression was more complex and difficult to get the parameter estimation. This paper was to estimate the main carbon budget process of pinus thunbergii forest in beichangshang island, and build carbon storage and Soil respiration factor model, which based on the nonlinear regression and parameter estimation. The main contents and results were as follows:First of all, the parameter of pinus thunbergii carbon storage logistic equation was estimated by robust regression methods. The results was that the average carbon storage and the average annual amount of carbon sequestration were 56.84 t/hm2,3.13 t/hm2, respectively. In order to improve the accuracy and applicability of pinus thunbergii carbon storage equation, we built the soil environment factor model to improve the logistic equation, which based on principal component analysis method. It could evaluate quantitatively the effects of soil environment factor on carbon sequestration rate by the improved equation.Then the soil carbon flux dynamic characteristics were analyzed, it was concluded that forest land and growth period also have significant influence on soil carbon flux through variance analysis. The estimation of soil carbon flux annual emissions was 1.13 t/hm2. The relationship between soil carbon flux and soil temperature, soil moisture was fitted, and the nonlinear model parameter was estimated by the gauss-Newton iterative method.