基于树干不同高度直径的落叶松立木材积方程

摘要

One-variable and two-variable volume equations were established based on 15 sets of diameters at different relative height and diameter at breast height (1.3m) for Larix gmelinii in Daxinganling.All models were fitted using GNLS in S-PLUS.Variance functions (exponential function, power function and constant plus power function) were incorporated into generalized models to reduce heteroscedasticity.Coefficient determination (R2), root mean square error (ERMS), mean absolute bias (BMA), and mean percentage of bias (BMP) were employed to evaluate the precision of different individual volume models.The best one variable and two variable volume equations were found based on the diameter of thirty percent of relative height.By model validation, one variable model based on diameter at 30% relative height reduced ERMS, BMA and BMP by 25.6%, 24.7%, and 24.7%, respectively, in comparison with traditional one-variable model.Compared with traditional two-variable model, the model based on diameter at 30% relative height reduced ERMS、BMA and BMP by 55.6%, 41.2%, and 41.2%, respectively.Prediction precision of two-variable model is better than that of one-variable model.%以大兴安岭兴安落叶松为研究对象,基于树干15个不同相对高度处直径和1.3m处胸径分别建立一元和二元材积方程.利用S-PLUS软件的广义非线性GNLS模块对各模型进行拟合.采用指数函数、幂函数和常数加幂函数对立木材积模型拟合过程中产生的异方差现象进行校正.采用确定系数(R2)、均方根误差(ERMS)、平均误差绝对值(BMA)和相对误差绝对值(BMP)4个指标对模型进行评价.结果表明,基于树干相对高度30%处直径的一元和二元模型拟合效果最好.模型检验结果表明:相对于传统的一元模型,基于相对树高30%处直径的一元模型的ERMS、BMA和BMP分别降低了25.6%、24.7%、24.7%;相对于传统的二元模型,基于相对树高30%处直径的二元模型的ERMS、BMA和BMP分别降低了55.6%、41.2%、41.2%.二元模型的检验精度明显优于相应的一元模型.