短叶松同龄纯林林分平均直径的时序分析

摘要

根据BoX-Jenkins建模原理,将ARIMA模型拟合时间序列的方法用于同龄纯林林分平均直径生长的拟合及预测研究.利用EvieWs软件分别对短叶松同龄纯林样地1号、2号、3号的林分平均直径序列建模和预测,选出最优模型形式.然后用样地4的林分平均直径序列对所得到的模型形式进行适应性验证.通过用这4块样地数据对最优模型形式的拟合和预测,结果显示,最优模型形式剔除AR(2)项的AR(3)模型能够较好的拟合平均直径序列,决定系数(R2)分别为0.907 5、0.909 7、0.910 3、0.949 8,并且预测相对误差(Bias)很小(＜1%).在直径生长序列的近期预测中应用时间序列的ARIMA模型方法较为合适.%According to Box-Jenkins principle and ARIMA model, the time series of the average diameter of Jack pine even-aged pure stand was fitted and predicted.A model was constructed and predicted for the data of sample plot 1,2,3 by Eviews software to select out the form of the optimal model.The data from plot 4 were used for model feasibility verifying.After fitting and prediction for four sample plots, the results showed that the AR(3) model rejecting AR(2) term better fit the series of the average diameter, R12 = 0.907 5 R22 =0.909 7 R32 =0.910 3 R42 =0.949 8, and bias was very small(＜1％).ARIMA model was suitable for short-term prediction of the series of diameter.