Modeling Root Biomass Of Simao Pine (Pinus Kesiya Var.Langbianensis) Natural Forest

The study of forest biomass and carbon stocks has become a globalhotspot with the global warming, the energy conservation and emissions redussionbecomes a highlight in international community. Forest, a main part of biologicalsystems, contains about80%of the aboveground and40%of underground carbonstocks, and plays an important role in global carbon balance. Root biomass is animportant part of forest biomass, accurate estimates of root biomass is the basis ofthe research on forest biomass and the key to accurate estimating forest carbon.49sample trees of Sinao pine (Pinus kesiya var. langbianensis) in Simaodistrict of Yunnan province had been investigated the root biomass, and theindividual biomass models of main root, lateral root and the total root of the Simaopine natural forest had been built by the power function. Firstly, the generalnonlinear model had been built using the basic measuring tree factors. Secondly,topographic, climate and competition factors had been introduced to construct theroot biomass models. Thirdly, the variance and covariance functions of models wereadopted, and the impact of variance and covariance function of the model toaccuracy and heteroscedasticity were analyzed. The results were showed as follows:(1) The biomass models of main root, lateral root and the total root of the Simaopine natural forest were built based on the basic measuring tree factors, the bestfitting models were as follows:BR0.0146D2.0316CW2CL0.41820（R2=0.9505）BRM0.0053D1.78990.85860H（R2=0.9512）BRL0.0048D2.7965CW2CL0.19260（R2=0.8931）(2) After the variance and convariance function were adopted, the accuracy ofmodels were improved and the heteroscedasticity was eliminated. The best model ofthe main root biomass was the model with the variance structure of Fix(D0), and forthe lateral root biomass was the model considering the variance of power faction,but for the total root biomass was the model considering the covariance of Sphericalform. The fitting results as follows: BR0.0119D2.05920.41820HBRM0.005D1.8420.81720HBRL0.0029D2.9326CW2CL0.18730(3) The models including environment factors were better than the general linearmodel for the fitting results, and the models including the topographical factorsconsidering the variance function were the best, the lowest were the modelsincluding the competition factors. After the variance and convariance form wereadopted, the root biomass models were as follows:BR0.0066D2.2943H0.5082(0.05GALT)0BRM0.005D2.0701(0.0229GALT)0.0265GSLO H0.55580BRL0.00003D3.70622(0.01865GASP) CW2CL0.20317(0.04353GALT)0(4) The independent sample data was used to test the basis model and themodel including environmental factors. The test results showed that the generalnonlinear model had the highest accuracy for main root biomass (P=81.9%), Thehightest precision model of lateral root biomass model was the topographic factorswith power form, and the value was72.8%, The precision of the general nonlinearmodel was the best for totlal root biomass, and the value is86.2%.