基于修正函数的杉木人工林单木冠幅预测模型1）

在福建省顺昌县，选择50块不同林龄、密度和立地条件的杉木人工纯林为研究对象，采用修正方程法构建杉木单木冠幅模型。在冠幅－树龄潜在生长函数的基础上，以地位指数、林分密度、竞争指数和胸径作为修正变量，根据指数函数和幂函数的组合形式构建误差函数，基础函数与误差函数的乘积即为冠幅修正模型。结果表明：在16种组合的修正方程中，4项幂函数乘积组合的模型拟合效果最佳（ R2＝0．876）；使用未参与建模的数据对模型进行检验，检验结果显著；将修正模型与一元线性回归和多元线性回归模型进行对比分析，发现无论是决定系数还是残差值，修正模型的拟合效果最优，因此，修正模型可以更好地预测杉木冠幅生长。

An Individual Crown-Width Model for Chinese fir Plantation Based on Modified Function

An individual crown-width model was developed by modified equations based on a data from 50 plots of Cunninghamia lanceolata plantations located in Shunchang County , Fujian Province with various stand density , ages and site quality .In the model building , the potential increment of crown-width model was developed , the error function can be assembled with the combination of exponential function and power function , taking stand index , stand density , competition index and di-ameter at breast height ( DBH) as modified variables .The tree crown-width model was characterized by the product of a-bove two functions .The optimal model was in the form of the product of power functions with the determination coefficient of 0.876.With the data not involved in modeling , the model was verified and the result was significant .In addition, either fitting effects or residuals , the modified model was significantly accurate than that of unary linear regressive model and mul -tiple linear regression models .Therefore , the modified model can better predict the crown growth of Chinese fir .