西安市城郊自备井开采区承压水位埋深的变化及预测  

The variation trend and prediction of the buried depth of confined water in suburb of Xi'an

作　　者：董艳慧[1] 周维博[2] 马东华[1] 

机构地区：[1]西安工业大学建筑工程学院,西安710032 [2]长安大学环境科学与工程学院,西安710054

出　　处：《干旱区资源与环境》2015年第6期158-162,共5页Journal of Arid Land Resources and Environment

基　　金：国家自然科学基金项目(51009008);西安工业大学校长科研基金(XAGDXJJ1123)资助

摘　　要：对西安市城郊自备井开采区1985-2005年的承压水开采量及水位埋深变化进行了分析,1985-1990年为开采量增加阶段,水位持续下降,1990年后开始减少开采量,至1997年开采量降到可开采量以下。减采5年后,即1995年后水位开始回升;至2005年,减采15年后,大多井点未回升至1985年的水位。在此基础上用径向基(RBF)神经网络模型,对减少承压水开采量后,该区4个承压水井点2006-2023年共18年的水位埋深进行了预测。1990-2002年的数据用于模型的训练,2003-2005年的数据用于校核。结果表明,2023年西安市城郊自备井开采区水位虽有适当恢复,但大多不能还原至1985年前未大规模超采时的水位。仅有降雨、径流和开采量20年左右的数据序列时,径向基神经网络模型能较准确地预测承压水位埋深。The changes of artificial exploitation quantity and buried depth of confined water from 1985 to 2005 in Xi＇an suburb own well working section were analyzed. As the exploitation increased,the confined water level declined from 1985 to 1990,exploitation reduced since 1990. From 1997 the exploitation quantity became under the allowable exploitation. After 5 years of exploitation reducing that was in 1995,the water level began to rebound. After 15 years exploitation reducing,most wells could not recover to the levels of 1985. A Radial Basis Function neural network model was established and was used to forecast the buried depth of confined water for2006 to 2023. The data of 1990 to 2002 were used to train,in order to get the RBF prediction model; the data of 2003-2005 were used to validate the model. The results indicate that the groundwater level will rise for some meters by 2023; however most wells could not recover to the level of 1985 that the lager-scale overdraft has not happened. The RBF neural network could accurately forecast the buried depth of confined water,based on rainfall,runoff and artificial exploitation with only about 20 years data.

关 键 词：承压水水位埋深 预测 径向基模型 降雨 径流 开采量 

分 类 号：P641[天文地球—地质矿产勘探]