兴安落叶松材积模型中的异方差研究

Study on heteroscedasticity of Larix gmelini(Rupr.)in volume model

为进一步提高兴安落叶松材积模型估计精度,文章选择V=aDbHc为材积模型形式,对模型的异方差性进行了研究。文章分别使用了图示法及戈德菲尔特-夸检验方法证实模型中存在较强的异方差性,并分别以因变量,自变量及模型本身构造权函数,以加权回归估计和普通非线性回归估计方法结果进行对比分析。研究结果表明:加权回归估计优于普通非线性回归估计;在构造的众多权函数中,以权函数1/D2H为最优;并进一步证实不同的模型有不同的最优权函数形式。

This paper study heteroscedasticity of I.arix gmelini（Rupr. ）volume model with for improving estimate precision on the model. The heteroscedasticity was affirmed in volume model using cartography and Goldfeld--Quandt test, and weigh ting function was made from independent variable, attributive variable and model, and weighting regression was a contrast to u sual nonlinear regression. Study results show that weighting regression was better with usual nonlinear regression; 1/ D2H was had best to so; and it was proved that various model had different optimal weighting function form.