A Field Evidence Model: How to Predict Transport in a Heterogeneous Aquifers at Low Investigation Level?

Abstract

Aquifer heterogeneity in combination with data scarcity is a major challenge for reliable solute transport prediction. Velocity fluctuations cause non-regular plume shapes with potentially long tailing and/or fast travelling mass fractions.High monitoring cost and presumably missing simple concepts have limited the incorporation of heterogeneity to many field transport models up to now.We present a hierarchical aquifer model which combines large-scale deterministic structures and simple stochastic ap-5 proaches. Such a heterogeneous conductivity can easily be integrated into a numerical models. Depending on the modelling aim, the required structural complexity can be adapted. The same holds for the amount of available field data. The conductivity model is constructed step-wise following field evidence from observations; though relying on as minimal data as possible.Starting point are deterministic blocks, derived from head profiles and pumping tests. Then, sub-scale heterogeneity in form of random binary inclusions are introduced to each block. Structural parameters can be determined e.g. from flowmeter measure-10 ments.As proof of concept, we implemented a predictive transport model for the heterogeneous MADE site. The proposed hierarchical aquifer structure reproduces the plume development of the MADE-1 transport experiment without calibration. Thus, classical ADE models are able to describe highly skewed tracer plumes by incorporating deterministic contrasts and effects of connectivity in a stochastic way even without using uni-modal heterogeneity models with high variances. The reliance of 15 the conceptual model on few observations makes it appealing for a goal-oriented site specific transport analysis of less well investigated heterogeneous sites.

A Comparison of Six Transport Models of the MADE‐1 Experiment Implemented With Different Types of Hydraulic Data

“…Thus, the FOA (Fiori et al., 2017), the MIM and self‐consistent approximation (MIMSCA) (Fiori et al., 2013), and the TDRW (Dentz et al., 2020) models utilize extensive hydraulic profiling data (Bohling et al., 2016), for inferring geostatistical parameters. In contrast, the binary inclusions model (Zech et al., 2021) uses pumping tests and a few flowmeter data whereas the Binary Facies model relies on granulometric data.…”

“…We describe here briefly the application of the model to MADE presented in Zech et al. (2021), in which the stochastic conceptualization of the binary hydraulic conductivity took large scale deterministic information into account.…”

“…Application to MADE: Zech et al. (2021) adopted the K ‐structure for MADE and its characterizing parameter values based on the inspection of the piezometric surface map of Boggs et al. (1992, Figure 3), on two large scale pumping tests and on a few flow meter logs (Boggs et al., 1992).…”

“…We examine the ability of six transport models of different aquifer conceptualization to predict the observed relative mass distribution without calibration of transport parameters: (a) the first order approximation (FOA) (Fiori et al., 2017); (b) the multi‐indicator model and self‐consistent approximation (MIMSCA) (Fiori et al., 2013); (c) the time‐domain random walk (TDRW) of Dentz et al. (2020); (d) the Binary Inclusions model (Zech et al., 2021); (e) a Binary Facies model, and (f) the flow reactors model (see Section 3 for details). We further examine the predictive power of the models by extending the time ( t = 1,000 days) beyond MADE observations ( t max = 503 days).…”