基于时间序列模型的济南趵突泉地下水位预测

本文对2012年5月至2017年12月济南泉域趵突泉的地下水位波动规律进行了分析,并采用3种灰色时间序列模型,评价了保泉形势,通过模型的拟合效果检验可知:对于存在周期性波动趋势的地下水位观测值,GM(1,1)模型优势未能得到充分体现,ARIMA与Holt-Winters模型的预测结果较为接近,整体精度较高,表现出明显的预测优势。深入比选ARIMA与Holt-Winters模型的8项拟合优度指标,最终确定利用预测效果最优的Holt-Winters模型,从时间序列的角度对趵突泉地下水位进行分析,并预测2018—2019年趵突泉的年平均水位分别为27.734m和27.605m,泉水位波动的峰值为28.215 m,谷值将出现于2019年的6月份,数值仅为27.124m,逼近27.01m的泉水停喷线,为近7年来的最低水位。预测结果表明:现状降雨与开采条件下,2019年6月趵突泉将面临潜在的停喷的危机,保泉形势不容乐观。

Prediction of groundwater table based on time series models in Baotu Spring of Jinan

In this paper, the fluctuating trend of groundwater table in Baotu Spring of Jinan spring catchment was analyzed from May 2012 to December 2017. Three gray time series models were developed to evaluate the current situation of spring protection. The results show that the GM (1, 1) model is not really suitable for the groundwater table with obvious fluctuation. Meanwhile, Holt-Winters model is determined as optimal predictive model according to an in-depth comparison of eight goodness-of-fit indexes between ARIMA and Holt-Winters models. Accordingly, the groundwater table in Baotu Spring was predicted using Holt-Winters model. This concluds that the annual average groundwater table in Baotu Spring respectively is 27.734m and 27.605m in 2018 and 2019,and the peak value of spring water table is 28.215m. The valley value of 27.124m will appear in June 2019, which is close to the drying-up groundwater table (27.01m) of Baotu Spring and is the lowest value in recent years. Overall, a potential crisis of stopping spraying in Baotu Spring will occurr in June 2019. The situation of spring protection is not optimistic under the current rainfall and exploitation conditions.