非线性回归分析在立地质量评价和树高预测中的应用探讨

在数量化理论Ⅰ方法的基础上，将树龄、立地因子引入非线性公式，建立非线性回归模型。用改进的Ｇａｕｓｓ－Ｎｅｗｔｏｎ迭代法求解模型的待定参数，以此获得的预测模型既能对林地进行立地质量评价，又可准确地预测林分在不同年龄时的树高生长，进而预测林地生产力。用该模型与单纯应用数量化理论Ⅰ模型处理同一原始资料，前者的复相关比为０．９２９８０２７，剩余标准差为１．０９５５４２７，而后者分别为０．７９０４、１．９６３６．预测精度明显提高。

STUDY OF THE APPLICATION OF NONLINEAR REGRESSION ANALYSIS TO EVALUATE SITE QUALITY AND TO PRIDICT TREE HEIGHT

Following the quantification theory I,a new non-linear regression model is established as follows:where Y for tree height;t,tree age and x, site factor. In the model parameters could be calculated with the improved Gauss-Newton iteration method. This statistical model could not only be utilized to evaluate site quality but also precisely to predict height growth and site productivity for the different aged stands. The statistica; accuracy could be obviously increased using this model in multiple correlation ratios were 0.9298 and 0.7904, residuals 1.0955 and 1.9636, respectively.