考虑参数和边界条件不确定性的地下水污染随机模拟

为了分析模型参数的随机变化和边界条件的随机变化对地下水溶质运移模型输出结果的不确定性影响,采用蒙特卡洛模拟对一假想算例展开研究,并结合风险评估阐述不确定性分析结果.首先,建立研究区地下水溶质运移数值模拟模型,并综合利用局部灵敏度分析和全局灵敏度分析方法筛选出对模型输出结果影响较大的参数,连同模型的边界条件(第一类边界条件-水头值)一起作为随机变量.然后,利用优化超参数的高斯过程回归(GPR)方法建立模拟模型的替代模型,代替模拟模型完成蒙特卡洛随机模拟.最后,对随机模拟的结果进行统计分析和区间估计,并利用污染物浓度的概率分布函数对1、2、3号观测井进行地下水污染风险评价.结果表明:置信水平>80%时,1,2,3号观测井污染物浓度值的置信区间分别为34.77~35.03,57.74~58.04,100.07~100.69mg/L.此外,1,2,3号观测井为轻度污染的风险分别为6%,100%,100%;为中度污染的风险分别为0%,0%,99.6%;为重度污染的风险分别为0%,0%,0.5%,藉此为地下水污染修复防治和地下水的合理利用提供可靠参考依据.

Stochastic simulation of groundwater pollution considering uncertainty of parameters and boundary conditions

In order to investigate the influences of random changes in model parameters and boundary conditions on the output uncertainty of groundwater solute transport model, combined application of Monte Carlo simulation and risk assessment were applied to illustrate the uncertainty analysis results of a hypothetical example. Firstly, a numerical simulation model of groundwater solute transport was established, then the parameters with greater impacts on the model output screened by local and global sensitivity analysis, together with the boundary conditions of the model (the first type of boundary conditions—head value) were set as random variables.Then the Gaussian Process Regression (GPR) method of optimizing hyperparameters was employed to establish an alternative model of the simulation model to complete the Monte Carlo stochastic simulation. Finally, statistical analysis and interval estimate of the results of random simulation were carried out, and the probability distribution function of pollutant concentration was used to estimate the risk of different degrees of pollution of observation wells 1, 2, and 3. The results show that when the confidence level was greater than 80%, the confidence intervals of the pollutant concentration values in observation wells 1, 2, and 3 were 34.77~35.03, 57.74~58.04, and 100.07~100.69mg/L, respectively. In addition, in observation wells 1, 2, and 3, the risk of slight pollution was 6%, 100% and 100%, respectively; the risk of moderate pollution was 0%, 0% and 99.6%, respectively; the risk of heavy pollution was 0%, 0%, and 0.5%, respectively. The present study can provide a reliable reference for pollution remediation and rational utilization of groundwater.