林分生长数据的时序分析探讨

Time Series Analyses of Stand Growth Data

将时间序列分析的一些理论及方法用于林分生长的建模和收获预估。利用Box-Jenkins方法对9个短叶松同龄林分的直径生长序列进行分析,结果如下:(1)对于1a(年)为间隔的短叶松林分平均直径生长序列,提取其趋势函数,使其平稳化后,为MA(4)模型。做未来4a(年)以内的直径生长量预报,相对误差小于3％。(2)对于5a(年)为间隔的短叶松林分平均直径生长序列进行模拟,ARIMA(1,2,0)为最好的模型。对未来20年以内的直径生长量做预报,相对误差小于5％。

In this paper,time series analyses are used for modeling stand growth and predicting yield.The stochastic time dependent structures of 9 average diameter growth series of even-aged Jackpine stands were analysed using autoregressive average moving (ARMA) models.The conclusions are as follow:(1) After removing the growth trend from the average diameter growth series,the overall best model for the residual series is MA (4).The relative error is less than 3% for predicting future diameter growth within 4 years.(2) ARTMA(1,2,0) is the best choise for the five-year interval average diameter growth series,and predicting relative error is less than 5%within 20 years.