Influence of conceptual model uncertainty on contaminant transport forecasting in braided river aquifers
Influence of conceptual model uncertainty on contaminant transport forecasting in braided river aquifers
复制标题
概念模型不确定性对辫状河含水层污染物迁移预测的影响
DOI：
10.1016/j.jhydrol.2015.07.036
复制
发表时间：
2015
期刊：
影响因子：
--
通讯作者：
P. Huggenberger
中科院分区：
文献类型：
--
作者：
G. Pirot;P. Renard;E. Huber;J. Straubhaar;P. Huggenberger
水文地质学家经常面临野外资料缺乏的问题。弥补这种数据不足的一个有趣的方法是使用模拟数据。然后提出了使用模拟数据时的预测精度和不确定度评估问题。本文以辫状河含水层污染物运移预测为例，对这些问题进行了研究。在使用文献中的模拟数据时，根据不同的地质概念模型（多高斯，基于对象的，伪遗传）产生了多个无条件的地质实现。在基于MADE-II示踪剂实验数据集的污染物运移问题中，对这些岩石物理实现进行了测试。模拟结果表明，利用模拟数据可以实现合理的污染物输运预测。电导率非均质场的初始浓度条件和位置对羽流行为的影响大于等效渗透率。结果还强调了包括各种各样的地质概念模型的必要性，并且只要没有现场数据允许概念模型或参数值伪造，就不要在每个概念中限制参数空间探索。
Hydrogeologist are commonly confronted to field data scarcity. An interesting way to compensate this data paucity, is to use analog data. Then the questions of prediction accuracy and uncertainty assessment when using analog data shall be raised. These questions are investigated in the current paper in the case of contaminant transport forecasting in braided river aquifers. In using analog data from the literature, multiple unconditional geological realizations are produced following different geological conceptual models (Multi-Gaussian, Object-based, Pseudo-Genetic). These petrophysical realizations are tested in a contaminant transport problem based on the MADE-II tracer experiment dataset. The simulations show that reasonable contaminant transport predictions can be achieved using analog data. The initial concentration conditions and location regarding the conductivity heterogeneity field have a stronger influence on the plume behavior than the resulting equivalent permeability. The results also underline the necessity to include a wide variety of geological conceptual models and not to restrain parameter space exploration within each concept as long as no field data allows for conceptual model or parameter value falsification.