基于ARIMA模型的灌区地下水动态预报  

Groundwater dynamic prediction in irrigation area based on ARIMA Model

作　　者：李娜[1,2] 周维博[1,2] 董起广[1,2] 李慧[1,2] 马聪[1,2] 

机构地区：[1]长安大学环境科学与工程学院,陕西西安710054 [2]长安大学旱区地下水文与生态效应教育部重点实验室,陕西西安710054

出　　处：《人民长江》2014年第9期43-47,共5页Yangtze River

基　　金：国家"十二五"科技支撑计划资助项目(2011BAD29B0104)

摘　　要：针对陕西省泾惠渠灌区地下水不合理开采引起的环境地质恶化问题,首先采用平稳性检验、纯随机性检验以及BIC准则对影响灌区地下水位的降雨量、渠灌用水量、地下水开采量以及渠井用水比4个主要因子进行特征分析和预处理,然后利用ARIMA模型对这4个因子的未来变化进行模拟预测。在此基础上,采用Visual MODFLOW三维可视化软件对灌区的含水层结构、水文地质参数、研究区边界条件及源汇项等进行概化。利用灌区2001～2008年各观测井点地下水位对模型进行模拟识别,采用灌区2009～2010年的地下水位进行模型验证。结果表明,模型训练结果与实测水位较吻合,误差满足要求,对未来地下水位的预测也较为合理。Aiming at the deteriorating environmental geological problems of Jinghuiqu irrigation area in Shaanxi Province induced by irrational underground water mining,we use stationary test,pure randomness test and BIC criteria to analyze and pre-treat the four factors affecting the underground water level,including precipitation,amount of irrigation water,amount of groundwater extraction and the water use ratio of channel to well,and then apply ARIMA model to forecast the tendency of four main factors in the future. Then the aquifer structure,hydrogeological parameters,boundary conditions and source- sink term of the irrigation area are generalized by Modflow visualization software. We simulate and identify the model using the level data of groundwater observation wells from 2001 to 2008,and further verify the model using the underground level from 2009 to 2010. The results show that the training results are in accordance with the actual values,and so is the predicted groundwater level.

关 键 词：泾惠渠灌区 ARIMA模型 VISUAL MODFLOW 地下水预测 

分 类 号：P641.2[天文地球—地质矿产勘探]