基于时间序列分析改进法的地下水位动态预报——以吉林省桦甸市为例

采用逐步回归法替代传统时间序列分析模型中的多项式拟合法和自回归法,仅选用对因变量影响较大的自变量建立方程,提高了模型的精度。依据桦甸市26130018号观测井2000—2008年逐月的地下水位埋深资料,分别利用传统的、改进的时间序列分析法建立模型,以2009—2010年的观测数据进行精度检验,选择较优的改进模型预测了2011—2013年逐月的地下水位。结果表明:2个模型均满足精度要求,但经过改进后,随机方程和趋势方程的相关系数分别由0.830 7和0.803 9增至0.913 5和0.970 9,拟合结果明显提高。

Dynamic Forecasting of Groundwater Table Based on the Improved Time Series Analysis Method:A Case Study of Huadian City,Jilin Province,China

The stepwise regression method was employed to replace the polynomial fitting method and autoregressive model in the traditional time series analysis model.Only use the significant variables to establish the prediction equation instead of taking into account all the variables,so the accuracy was improved.The traditional and improved models were built respectively according to the monthly groundwater table observation data of No.26130018,the observation well in Huadian city,Jilin Province,China from 2000 to 2008,and they were tested by the data from 2009 to 2010.The monthly groundwater table of 2011 and 2012 was predicted by the improved model.The result shows that both the two models meet the accuracy requirements,but the correlation coefficients of trend equation and stochastic equation of the improved model increase from 0.830 7 and 0.803 9 to 0.913 5 and 0.970 9,the fitting results are significantly improved.