Application of Numerical Tools to Investigate a Leaky Aquitard beneath Urban Well Fields

Abstract

Memphis aquifer is the primary drinking water source in Shelby County (Tennessee, USA), and it supplies industrial, commercial, and residential water. Memphis aquifer is separated from the Shallow aquifer by a clayey layer known as the Upper Claiborne confining unit (UCCU). All of the production wells in the Memphis area are screened in the Memphis aquifer, or even deeper in the Fort Pillow aquifer. Traditionally, it was assumed that the UCCU could fully protect the Memphis aquifer from the contaminated Shallow aquifer groundwater. However, recent studies show that at some locations, the UCCU is thin or absent, which possibly leads to the contribution of Shallow aquifer to the Memphis aquifer. Accurately locating the breaches demands expensive and difficult geological or geochemical investigations, especially within an urban area. Hence, a pre-field investigation to identify the locations where the presence of breaches is likely can significantly reduce the cost of field investigations and improve their results. In this study, to identify the locations where the presence of breaches in the UCCU is likely, we develop a reliable MODFLOW-based numerical model, and use three different analyses: (1) pilot-point calibration (PPC), (2) velocity and flow budget (VFB), and (3) particle tracking (PT), to post-process the developed groundwater model results. These pre-field numerical investigations provide relevant and defensible explanations for groundwater flow anomalies in an aquifer system for informed decision-making and future field investigations. In this study, we identify five specific zones within the broad study area which are reasonable candidates for the future field investigations. Finally, we test the results of each analysis against other evidence for breaches, to demonstrate that the results of the numerical analyses are reliable and supported by previous studies.

Development of a Numerical Multi-Layered Groundwater Model to Simulate Inter-Aquifer Water Exchange in Shelby County, Tennessee

“…The correspondence of localized leakage in parts of Shelby County with previously identified breach areas and anomalous water-table depressions suggests that determinative methods of assessing heterogeneity in confining units may provide useful model constraints for modeling inter-aquifer water exchange. Although other methods also have proven useful for identifying locations and extent of inter-aquifer exchange through the UCCU in Shelby County [22,61,96], determinative methods have the advantage of potentially providing more predictive behavior.…”

“…It underlies the shallow aquifer and overlays much of the Memphis aquifer with clay, silt, and sand, ranging in thickness from 0 to 60 m but thins toward the eastern part of Shelby County [4] (See Figure 3). Several investigations have shown the presence of hydrogeologic breaches where the shallow aquifer is in direct hydraulic connection with the Memphis aquifer owing to thin or absent clay intervals [4,6,7,9,12,13,22,32,33,40,[59][60][61].…”

“…Local variability in inter-aquifer exchange can be evaluated numerically [22], but the impacts on water budget under transient conditions in a complex hydrogeologic and pumping setting are not well established. Previous numerical modeling studies of the Memphis aquifer in Shelby County [23][24][25][26][27][28] have not adequately addressed inter-aquifer [4], as well as the area where the Memphis aquifer is under unconfined conditions.…”

Parameterization of Pilot Point Methodology for Supplementing Sparse Transmissivity Data

“…This solution though essentially implying optimal reproduction of the historic hydraulic head fields, is assumed to lead to optimal model predictive capability. The reported studies have generally used kriging as the T interpolator employing the sampled T values [17,[23][24][25][26]. The other IP based approach for parameter estimation is to predict (model) the parameter values at the sampled locations and assimilate the sampled values into the predicted ones by ensembled Kalman filters (EnKF) [27][28][29].…”

“…Other reported subjective criteria are placement of pre-stipulated number of the pilot points in areas of high hydraulic gradients [31], uniformly distributed [25,26,[32][33][34][35], in areas of lower monitoring density [20], or at arbitrary locations [23]. These heuristic approaches may compromise the potential gain from the pilot point data with respect to the envisaged objectives.…”