SIMULATION OF PROJECTED GROUND-WATER LEVELS

A calibrated ground-water flow model was used to simulate projected water levels for the Cretaceous-Paleozoic aquifer system in northeastern Mississippi. A thorough description of model construction and boundary conditions is presented in Strom (1998); however, an abbreviated model description pertinent to the Coffee Sand and Eutaw-McShan aquifers is presented below.

Ground-Water Model Description

The ground-water model was constructed using the finite-difference computer code MODFLOW (McDonald and Harbaugh, 1988; Harbaugh and McDonald, 1996). The model grid covered 34,960 mi² and was oriented north-south because no predominant axes of transmissivity for the aquifers were indicated by the data. A lateral anisotropy ratio of 1 was used in the simulations. Each grid layer consisted of 230 rows and 152 columns with each grid cell 1 mile square. The model was vertically discretized into 6 layers resulting in a total of 209,760 grid cells. Layers 1, 2, and 3 represented the Coffee Sand, Eutaw-McShan, and Gordo aquifers, respectively. The Coker and Iowa aquifers were represented by layer 4, and the massive sand and Devonian aquifers were represented by layer 5. Although the Coker and Iowa, and the massive sand and Devonian aquifers are not stratigraphically related, those aquifers can be simulated on shared layers because their boundaries do not areally coincide. The Lower Cretaceous aquifer was represented by layer 6.

 

Model boundaries determine where and how much water enters and leaves the model; therefore, the selection of appropriate boundaries for the aquifers is a major concern. The selection of model boundaries for the aquifers in this model was based on a conceptual interpretation of the flow system developed by using information reported by Boswell (1963, 1978); Boswell and others (1965); Cushing (1966); Hardeman (1966); Bicker (1969); Gandl (1982); Wasson (1986); Davis (1987); E.H. Boswell, J.F. Everett, D.L. Hardin, J.H. Hoffman, S.P. Jennings, P.A. Phillips (Mississippi Office of Land and Water Resources, oral commun., 1993); Jennings (1994); J.H. Hoffmann (Mississippi Office of Land and Water Resources, oral commun., 1997), and S.P. Jennings (Mississippi Office of Land and Water Resources, written commun., 1997).

 

The Coffee Sand aquifer is overlain by a thick, relatively impermeable sequence of units in the Selma Group; therefore, the area overlying the Coffee Sand aquifer was simulated as a no-flow boundary. Layer 1 represents the Coffee Sand aquifer in the northern part of the model, but also represents an upper constant-head boundary for the Eutaw-McShan aquifer (layer 2) in the southeastern part of the model area. The constant heads overlying the Eutaw-McShan in this region represent surficial water levels in the chalk and clay overlying the Eutaw-McShan aquifer. However, most of this potential water is hydraulically separated from the Eutaw-McShan by the clay and chalk confining unit that sharply thickens westward, limiting most vertical flow due to the low vertical hydraulic conductivity of the confining unit.

 

The downdip extent of freshwater (defined for the purposes of this study as a concentration of 10,000 milligrams per liter of dissolved solids) represents no-flow lateral boundaries for all of the aquifers because of the contrast in density across the freshwater-saltwater interface. Previous investigations (Mallory, 1993; Arthur, 1994; Strom and Mallory, 1995) have indicated that this contrast in density effectively eliminates horizontal movement. A no-flow boundary at this location assumes a stable downdip, freshwater-saltwater interface. For many of the aquifers, the region where the dissolved-solids concentrations are between 1,000 to 10,000 milligrams per liter is relatively small, which also implies there is little mixing of water and flow is parallel to the freshwater-saltwater interface. If flow were to occur across the interface in the downdip direction, flow would eventually move upward at some point to discharge; flow upward is unlikely, however, because the confining units above the Eutaw-McShan thicken to the southwest in the downdip direction to more than 1,500 feet near the freshwater-saltwater interface. Any substantial upward flow would be through secondary structural features, such as faults.

 

The northern or northwestern boundaries of the Coffee Sand and Eutaw-McShan aquifers represent the limits of the sediments and are simulated as no-flow boundaries (figs. 3 and 4, respectively). The southeastern boundary of the Eutaw-McShan is also simulated as a no-flow boundary. The southeastern boundary is at a lateral ground-water flow divide formed by the Tombigbee and Black Warrior Rivers. Water-level data indicate that these rivers, particularly near their confluence, are major discharge areas for the Eutaw-McShan aquifer, with lateral flow converging from both the east and the west captured by the river channels (Gardner, 1981). Consequently, no lateral flow is assumed to move beneath the Tombigbee and Black Warrior Rivers. Ground water in this area is discharged by upward leakance through the confining units.

 

An average of about 52 inches per year of precipitation falls on the aquifer outcrop areas in northeastern Mississippi (National Oceanic and Atmospheric Administration, 1981). Only a small fraction of this amount enters the ground-water flow system as recharge. Some of the water that enters the ground-water flow system travels only a short distance before being discharged locally into streams and other drains. The digital model does not simulate all the localized flow because of the 1-mile grid discretization. The model simulations represent the intermediate and regional scale flow system. The outcrop areas of the Coffee Sand and Eutaw-McShan aquifers were simulated with head-dependent flux boundaries. This was implemented using the river package in MODFLOW (Harbaugh and McDonald, 1996). The large base flows observed in even small streams in the outcrop area indicate that recharge from the precipitation-rich environment is more than sufficient to provide all the recharge that the aquifers can accept; however, much of the potential recharge is rejected by the aquifers and diverted into surface runoff due to the limited lateral transmissivities of the aquifers. The minimum land-surface altitude in each outcrop grid cell, which approximates stream base-flow water-level elevations, represents the river stages in the river package.

Projected Ground-Water Levels

Projection simulations were made using baseline-, normal-, and high-growth water-use demands. The simulated potentiometric surfaces of the Coffee Sand and Eutaw-McShan aquifers for year 2000 (figs. 7 and 8, respectively) serve as a reference for near-current conditions from which water-level changes are discussed.