中文摘要：

以云南省香格里拉县的高山松林为研究对象,选取遥感因子与地形因子作为模型中的固定效应,并以幂函数模型为基础进行林分生物量基本模型的构建;采用混合效应模型技术,根据海拔高低把样地划分为6个区域,将区域效应作为随机效应,并在基础混合效应模型基础上考虑方差和协方差结构,构建林分生物量混合效应模型,以此估测高山松林分生物量。利用AIC、BIC和Log Lik 3个拟合指标评价模型的拟合效果,利用SRE、MRE和AMRE进行最终林分生物量混合效应模型的独立性检验。结果表明：从模型拟合结果看,考虑区域效应的混合效应模型的拟合效果明显高于基础模型,其AIC和BIC值最低,Log Lik达到最大;从模型独立性检验看,考虑区域效应的混合效应模型的绝对平均误差最小（AMRE=31.52%）,精度达到77.83%。综合分析,混合效应模型可有效提高高山松林分生物量估测精度。

英文摘要：

n this paper,the Pinus densata forest of Shangri-La City of Yunnan Province was taken as the research object,and the remote sensing factors and terrain factors were selected as fixed effects,based on the best power function model,the mixed-effect model of forest stand biomass was constructed by using the technology of mixed-effect models,and the sample plots were divided into 6 regions according to the altitude taking the regional effect as a random effect,and considering the variance and covariance structure on the basis of the mixed-effect model,so as to estimate the biomass of Pinus densata forest.The fitting effect of the model was evaluated by using AIC,BIC and Log Lik fitting parameters,and the SRE,MRE and AMRE were used to test the independence of the final mixed-effect model of stand biomass mixing.IThe results showed： in view of the fitting effect,the mixed model,which took the regional effect into consideration,was better than the basic model,its AIC and BIC values were the lowest,the Log Lik reached the maximum; in view of the independence test,the AMRE of the mixed model,which took the regional effect into consideration,was the lowest（ AMRE = 31.52%）,its prediction accuracy was 77.83%. By analysis,the mixed-effect model can effectively improve the accuracy of the estimation of biomass of Pinus densata forest.