Influence of Different Conceptual Models on Contaminant Transport Uncertainty in Braided River Aquifers, Using Data from Analog Sites

Braided-river aquifers represent a major source of water supply in inneralpine regions such as the Swiss Alps. In order to improve the protection of this resource or its consumers from any contaminant, one can use heterogeneous geological models and run flow and transport simulations to be able to predict contaminant migration and to quantify uncertainty. An important issue -due to field, time or money constraints- is that hydrogeological and geophysical information on the study site are often scarce. Meanwhile, analog study sites exist and can provide porosities and permeabilities from boreholes data, precise description and length scale characteristics of deposited structures from outcrops or Ground Penetrating Radar (GPR) surveys and interpretations. Here we test a procedure that includes the integration of analog site data to build geological models based on different conceptual models to provide a prediction of a contaminant migration as well as its uncertainty. The exercise is performed on the dataset from the MAcro Dispersion Experiment (MADE) site in Columbus (USA). GPR lines from the MADE site and analog site informations are used to calibrate the geological model parameters. The different geological modeling approaches are based on Sequential Gaussian Simulations, Multiple-Point Statistics, pseudo-genetic algorithm or object-based algorithm. The quality of the predictions is evaluated by comparing the simulated plume characteristics to the MADE plume characteristics.Preliminary results show that the 4 different types of geological models display very different connectivity patterns. For this specific exercise, the object-based model produced the largest uncertainty space that seems to encapsulate the tracer measurements, whereas the other methods generated a narrower uncertainty space that lay away from the measurements. We address the impact of additional information on the relevance of the methods.