二维强非均质含水层中水力传导度空间变异对污染物迁移的影响

对于介质的非均质性相对较小的情况，可将水力传导度对数(lnK) 的空间分布概化为正态分布的统计均匀随机场。而对于强非均质介质，lnK的空间分布可表征为其增量遵从Levy稳定分布的非平稳随机场。本研究假设二维承压含水层内面积为128m x 128 m的研究区域中,在(10m,64m)处瞬时注入浓度为1000 mg/L的污染物：利用改进的连续随机增加（successive random additional, SRA) 方法生成含水层lnK增量具有Levy稳定分布的统计特征的随机场；采用Monte-Carlo数值实验方法并结合MODFLOW与MT3DMS分别模拟研究区域内水流和溶质迁移过程，分析水力传导度强变异性对污染羽状物浓度空间矩及宏观弥散的影响。结果表明：Levy指数a决定lnK的空间分布形状，宽度参数C则影响lnK的取值范围；污染羽的质心位置(一阶矩)与C、a的大小无关；C越大、a 越小，污染羽形状越不规则，污染羽状物的扩散范围(二阶矩)越大；且a 越小，污染羽状物具有明显的拖尾现象；纵向宏观弥散度随时间呈幂函数递增趋势。

Effect of hydraulic conductivity on contaminant transport in a highly heterogeneous two-dimensional aquifer

Hydraulic conductivity (K) in slightly heterogeneous media is currently treated as a lognormally distributed statistically homogeneous random field . But for the highly heterogeneous media, the lnK is considered as a non-stationary random field with its increments following a Levy-stable distribution. In this paper, a confined two-dimensional aquifer with an area of 128 m x 128 m was considered as a case of study, with which a pulse contamination input with a concentration of 1000 mg/L was introduced at the position x=10m, y=64m. The successive random additional (SRA) algorithm was used to create the lnK field of the aquifer with Levy-stable distribution. A Monte-Carlo simulation, with which the MODFLOW and MT3DMS were used to simulate water flow and contaminant transport respectively, was performed to investigate the highly heterogeneous properties on water flow and contaminant transport. And the spatial moments and macrodispersity were calculated. Results indicated that the Levy index a is mainly responsible for the shape of the lnK distribution, while the width parameter C has significant effect on the range of lnK values. As a becomes smaller, contaminant transport becomes more anomalous with a long time tailing part plume. It is also found that the spread of the contaminant plume and its second moments increase as the increase of width parameter C , but the centroid position of the plume is independent of the values of parameters C and a. . The longitudinal macrodispersivity is scale-dependent and increases with time in a power law function.