落叶松人工林直径分布动态预估模型  

Predicting Models of Diameter Distribution Dynamic for Larch Plantation

作　　者：赵丹丹[1] 李凤日[1] 董利虎[1] 

机构地区：[1]东北林业大学,哈尔滨150040

出　　处：《东北林业大学学报》2015年第5期42-48,共7页Journal of Northeast Forestry University

基　　金：林业科学技术推广项目[2012]48号;国家林业公益性行业科研专项(201004026)

摘　　要：基于小兴安岭地区和长白山地区102块落叶松人工纯林固定标准地复测数据（10 a间隔期）,采用参数预测模型（PPM）系统,建立了前期Weibull分布参数（b1）与前期林分调查因子模型、前期参数c1与b1之间的回归模型、两期参数b2与b1的回归模型、以及两期参数c2与b2之间的回归模型,并采用似乎不相关回归（SUR）理论,估计了模型的参数;利用＂刀切法＂选择平均相对误差（ME）、平均相对误差绝对值（MAE）、预测精度（P）、相对误差（B）、误差指数（IE）等指标,分别对所建立的参数动态预测方程及直径分布动态预测结果进行了检验。结果表明：所有模型的R2a较好（0.428~0.897）,RMSE均较小（0.37~0.94）,所建立的直径分布动态预测模型具有较好的拟合效果。通过检验,所建立的参数动态模型预估能力较好（-10%95%）,并能较好地预测落叶松人工林未来直径分布（B0=4.38%,B1=12.38%,IE=524）。We used 102 plots with the interval of 10 years for Larch pure plantation in Xiaoxing＇ an Mountains and Changbai Mountains, and followed the Weibull distribution to develop the relationship with the parameter b1 of the initial data and the forest factors of initial data by parameter prediction method （PPM）, and the relationship with two parameters ct and bt, b2 and bl, and c2 and b2. We simultaneously estimated the model coefficients using seemly unrelated regression （SUR）. By the model fitting, all the parameter prediction models were with good model fitting and prediction performance with the model R2a of 0.428-0.897, and small RMsE of 0.37-0.94, and the dynamic diameter distribution models were with a good prediction performance. The relative biases, the error index, the mean relative error, the mean absolute relative error, the precision were used to evaluate the validation statistics of dynamic parameter prediction equations, respectively. The dy- namic diameter distribution models have good precision, and can be used to predict the diameter distribution of Larch plan- tation in future.

关 键 词：落叶松人工林 WEIBULL分布 直径分布动态 参数预估模型 似乎不相关回归 

分 类 号：S718.55[农业科学—林学]