中文摘要：

地下水污染事件难以被直接观测,其发生环境通常也具有多种不确定性,风险评价可以量化地下水污染事件的危害性。利用随机配点模型和多项式抽样技术,建立了高效的风险评价模型;考虑了渗透系数、孔隙率和弥散度等多种不确定性因子,探讨了复杂条件下的风险函数的分布特征。研究结果表明：基于随机配点法的风险评价模型避免了多次重复求解对流弥散方程,通过计算成本低廉的随机配点技术得到浓度随机场的拉格朗日多项式,进行多项式抽样获取浓度样本并得到风险函数;与传统的解析算法相比,该方法无需对输入参数和浓度的分布形态做出假设;与传统的蒙特卡罗（Monte Carlo）算法相比,该模型具有明显的效率优势和优越的收敛速度;输入参数的分布类型对风险分布产生显著影响。

英文摘要：

Groundwater contamination event is hardly observed, and its circumstance has various uncertainties. Risk assessment provides a quantitative approach to evaluate the potential hazard of the contamination. This study develops a new method for risk assessment by using the stochastic collocation method and the polynomial sampling technique. The uncertainties from hydraulic conductivity, porosity and dispersivity are considered simultaneously. The risk distri- bution is assessed in a complex groundwater system. It is shown that our proposed model avoids the large number of repetitive solutions to the diffusion-dispersion equation. The low-cost stochastic collocation method is able to determine the Lagrange polynomials expression of the concentration field. The concentration samples are thus obtained by zero- cost polynomial sampling technique. Comparing to the traditional analytical methods, the new model does not adopt any assumption to the concentration distribution and the proposed model has superior efficiency and convergence prop- erty compared to the Monte Carlo simulation. The distribution type of input variable has significant influence to risk distribution.