时间序列模型在吉林西部地下水动态变化预测中的应用

Application of time-series model to predict groundwater regime

运用时间序列分析理论对吉林西部地下水位动态变化进行了分析和预报.首先采用多项式拟合提取水位动态的趋势分量,而后运用频谱分析方法中的谐波分析提取其中的周期成分,利用自回归(AR)模型模拟随机分量,最后将三者线性叠加建立预报模型,并给出了模型精度检验方法.通过模型分析,可知该区地下水位变化存在两个主要周期(1年和7～9年),揭示了地下水位的季节性变化和多年变化规律.2002年以后的预报结果表明部分地区的地下水位将持续下降,应及时加以控制.

The time-series analysis theory is applied to analyze and forecast the dynamic variation of groundwater in west part of Jilin Province, China. First, the trend component of groundwater level dynamic variation is picked up by polynomial calibration, the periodic component is extracted by spectrum analysis and the stochastic component is simulated by using autoregression （AR） modeI. Finally, a forecasting model is established through linear superposition of these components and the method for verifying the accuracy of the model is suggested. The analysis of this model shows that there are two major periods in the variation of groundwater level in this area by which seasonal and secular variation of groundwater level is revealed. The prediction for years after 2002 indicates that a continuing decline of groundwater level exists in some district of this area, which must be controlled in time.