中文摘要：

地下水流数值模拟过程中,水文地质参数的不确定性对模拟结果影响很大。以内蒙古鄂尔多斯市某水源地为例,利用拉丁超立方抽样（LHS）方法获得了含水层渗透参数的随机组合,进行地下水流随机模拟。通过对观测资料与计算水位的绝对值平均（MAE）误差和误差均方根（RMSE）的统计分析,获得了模型较为稳定的随机模拟次数是243次。利用该随机模型对水源地设计开采量进行水位预测,并给出允许降深的风险性分布图。结果表明,预测水位和标准差分布符合实际情况,水位降深大于35 m的风险性最大达到75%。

英文摘要：

The uncertainty of aquifer parameter has great impact on simulation results during setting up a numerical simulation model of groundwater flow.We have done a stochastic simulation of hydraulic conductivity with a source field in Ordos City,Inner Mongolia.Random combinations of hydraulic conductivity were obtained with Latin Hypercube Sampling（LHS） method.The proper number of stochastic realizations was 243 according to error statistics of mean absolute error（MAE） and root mean square error（RMSE） of water levels between simulated and observed in observation wells,which could better describe spatial variability of aquifer.With this stochastic model,the designed exploitation amount was predicted and the risk distribution of allowed drawdown was computed.Results show that the mean water level and standard deviation of realizations are appropriate and the biggest risk reaches 75% when water level drawdown is greater than 35 meters.