Nitrate represents the most widespread contaminant in shallow aquifers, especially in urban areas, and poses risks to human health, when the contaminated groundwater is ingested. In urban environments, the release of nitrate in groundwater can occur from multiple sources and is frequently associated with sewage leakage and septic tank infiltration. The Rio Claro Aquifer, located on the campus of the São Paulo State University at Rio Claro, offers an attractive example of a shallow aquifer impacted by nitrate contamination. Old sewage spills are considered to be the main sources of contamination; however, their locations remain largely unknown. Because of the scarce data and heterogeneous aquifer geology, the direct backward location approach is unsuitable in this case. Aiming to predict the probable locations of contamination sources, we developed a probabilistic backward location approach to identify the backward location in multiple geological scenarios using stochastic simulations. The numerical flow simulation and backward particle tracking were conducted based on 100 stochastic scenarios generated with Markov chains using lithological data from core descriptions. The multiple backward locations generated by stochastic simulations allowed us to build a density map to identify the region most likely to contain the contamination sources, thus simplifying the investigation and mitigation of the sewage spills.

硝酸盐是浅层含水层中最普遍的污染物，尤其是在城市地区，摄入被污染的地下水会危害人体健康。在城市环境中，地下水中硝酸盐的释放可能有多种来源，并且经常与污水泄漏和化粪池渗透有关。位于里约克拉罗市圣保罗州立大学校园内的里约克拉罗含水层，是受到硝酸盐污染影响的浅层含水层的一个很好的例子。旧的污水泄漏被认为是主要的污染源；然而，他们的位置仍然未知。由于数据稀少和含水层地质不均，在这种情况下不适合采用直接向后定位方法。旨在预测污染源的可能位置，我们开发了一种概率随机向后定位方法，可以使用随机模拟在多个地质场景中识别反向向后位置。基于岩心描述的岩性数据，基于由马尔可夫链产生的100个随机情景进行了数值流模拟和向后粒子跟踪。随机模拟产生的多个向后位置使我们能够构建密度图，以识别最有可能包含污染源的区域，从而简化了污水泄漏的调查和缓解。基于岩心描述的岩性数据，基于由马尔可夫链产生的100个随机情景进行了数值流模拟和向后粒子跟踪。随机模拟产生的多个向后位置使我们能够构建密度图，以识别最有可能包含污染源的区域，从而简化了污水泄漏的调查和缓解。基于岩心描述的岩性数据，基于由马尔可夫链产生的100个随机情景进行了数值流模拟和向后粒子跟踪。随机模拟产生的多个向后位置使我们能够构建密度图，以识别最有可能包含污染源的区域，从而简化了污水泄漏的调查和缓解。