参数不确定性对地下水数值模拟结果可靠性的影响

以标准矩形为研究区,按介质渗透系数差异将研究区划分为4个单元,应用HydroGeoSphere模拟系统建立数值模型,以含水层渗透系数为目标参数,分析不同参数分区的灵敏度系数;选择高灵敏度的分区建立渗透系数随机场,进行参数不确定随机模拟,分析模拟结果相对确定性模型的可靠度.结果表明:位于模型最顶层的单元1内渗透系数灵敏度最大,这与该单元的分布位置直接相关;随机模拟得到的压力水头按一定频率分布,确定性模型所得到的压力水头变化值290.52 cm出现的频率仅为7%;当认为确定性模型结果的可靠度为100%时,随机模拟结果相对于确定性模型结果的可靠性明显降低,这表明考虑到参数的不确定性时,模拟结果与确定性模型结果存在显著差异.随机模拟结果的概率性分布特征可以更加真实地表达实际情况,有利于水资源决策性管理的风险规避.

Influence of parameter uncertainty on reliability of groundwater numerical simulation results

A rectangular aquifer domain served as a study region for groundwater numerical simulations. The region was divided into four zones according to differences in hydraulic conductivity. A numerical simulation model was established by using the HydroGeoSphere simulation system. The sensitivity coefficient was analyzed in those aquifer zones when the permeability coefficient of the aquifer was used as an objective parameter. A stochastic simulation was conducted to evaluate the reliability of the results produced by deterministic model and stochastic model. Results showed that the permeability coefficient in top element 1 was the most sensitive. This fact may be related to the zone location directly. Based on the stochastic model, the pressure head was distributed in a certain probability. According to the deterministic model, the frequency of 290.52 cm pressure head variation was 7% only. When the reliability of deterministic model was considered as 100%, the reliability of stochastic model was reduced remarkably. This suggested that there was a significant difference in the results predicted by the stochastic model from those by the deterministic model. Since the probabilistic distribution characteristics in the results of stochastic model can represent the actual situation of groundwater more realistically, the stochastic model is favor of risk evasion in groundwater resources decision-making management.