基于不同预测变量的橡胶树地上生物量模型

Establishment of Aboveground Biomass Models Based on Different Predictors for Hevea brasiliensis

在海南岛范围内,橡胶林种植面积很大,但对于这种经济树种大多评估它的生产效益和经济效益,很少建立生物量估计模型。基于海南岛54株实测生物量数据,建立橡胶树不同组合自变量非线性回归模型和非线性度量误差模型。由于生物量与自变量之间存在非线性关系,在拟合模型时采用最小二乘建模技术。模型的选择基于解释系数(Pseudo-R2)、均方根误差(RMSE)和平均预测误差(MPE)。结果表明,通用模型的解释系数(Pseudo-R2)和平均预测误差(MPE)分别为0.70~0.94、-0.8%~1.1%,选择基于胸径和木材基本密度的二元模型为最优通用模型。以最优通用模型地上生物量估计值为总量,构建各分项生物量相容性模型,其中树干生物量模型的Rseudo-R2、RMSE值分别为0.92、28.4 kg,拟合效果优于树枝和树叶生物量模型。总体来看,添加树高、树冠和木材基本密度因子进行橡胶树地上生物量模型的构建十分必要。

Tropical rubber(Hevea brasiliensiss)plantations are widely distributed in Hainan island,it is of importance to establish models for the cultivation of rubber trees.However,most models are focused on the production and economic benefits,few are on the estimation of biomass.This study based on the measured biomass data of 54 plants in Hainan island,nonlinear regression models and nonlinear measurement error models of different combinations of independent variables of H.brasiliensiss were established.Since biomass revealed nonlinear relationships to the independent tree variables,we used a nonlinear least square modeling technique when fitting models.Selection of models was based on Pseudo-R2,root mean square errors(RMSE)and mean prediction errors(MPE)from a leave-one-out-cross-validation procedure.The results showed that,Pseudo-R2 and relative MPE of the models varied from 0.70 to 0.94 and-0.8%to 1.1%.The optimal universal model(OUM)that was based on binary model of DBH and basic wood density was selected.Taking the amount of above-ground biomass that was estimated from OUM as total biomass,compatibility models of biomass components were constructed,from which Pseudo-R2 and RMSE of the trunk model were 0.92 kg and 28.4 kg,respectively,model fitting results were better than branch and leaf biomass models.Overall,it is necessary to add factors such as tree height,canopy and wood basic density to construct the aboveground biomass model of H.brasiliensis.