A Comparison of Six Transport Models of the MADE-1 Experiment implemented with Different Types of Hydraulic Data

Abstract

Modeling contaminant transport by groundwater is a topic of great interest that stimulated intensive research in the last four decades due to its relevance to aquifer pollution (Dagan, 1989;Fetter et al., 2018;Gelhar, 1993;Rubin, 2003). The task of predicting transport faces several difficulties: the processes are of long duration, measurements are scarce, the subsurface medium is of complex heterogeneous structure subjected to uncertainty and many times, the geometry and the mass content of the contaminant source is also uncertain. Under these circumstances, models play an important role: they help understanding the involved processes, analyzing field data, and making long term predictions. Models developed in the past differ in conceptualization of the aquifer structure, in the required data, in quantification of transport, in the formulation of the governing equations and mechanisms they represent, in computational complexity, and in their goals. We focus on transport of plumes of conservative solutes in steady flow, driven by the natural head gradient. At the considered aquifer scale, the main mechanism responsible for plume spreading is macrodispersion (Zech et al., 2015(Zech et al., , 2019 due to the spatial variability of the hydraulic conductivity   K x . The effect increases with higher K heterogeneity as quantified for instance by the log-conductivity variance  2 Y . We focus on the quantification of transport by the spatial relative mass distribution in mean flow direction x.