Simulation of Projected Water
Demand and Ground-Water Levels in the Coffee Sand and Eutaw-McShan Aquifers
in Union County, Mississippi, 2010 through 2050
MODEL LIMITATIONS
The results of this investigation, to project future water demand in New Albany and to evaluate the response of the aquifers to increased pumping to meet that demand, are based on the IWR-MAIN and MODFLOW models. The results of both models depend on the data used to define model conditions and on the assumptions made to simulate actual conditions. As with any model, the degree of uncertainty increases the further out in time that projections are made. A projection of 52 years (1998-2050) assumes many political, environmental, economic, and technical factors will not shift radically.
IWR-MAIN is used primarily to test assumptions and the effect various assumptions or changes would have on water use in the WSA rather than as a predictive tool to generate absolute amounts in the future. This fact and basic assumptions about growth, land use, population, and technology determine the results. If the assumptions change (for example, if population decreases in the area), water-demand results will change. The accuracy of the results depends on the validity of the assumptions.
The accuracy of ground-water models is limited by the assumptions made in formulating the governing flow equations and in the assumptions made during model construction. Models are limited by cell size, number of layers, boundary conditions, time discretization, hydraulic values, accuracy of calibration, verification data, and parameter sensitivity. Models also are limited by the availability of data and the interpolations and extrapolations of available data in a model. The model may be calibrated and verified, but the calibrated parameter values may not be unique in satisfying a particular distribution of hydraulic head.
The model used in this study is suitable for analyzing ground-water flow on a regional scale. Site-specific analysis is limited by horizontal and vertical discretization of the model and the availability of site-specific data. The model calculates an average head for the entire cell area (1 square mile), which may not be a good approximation for the water level in an individual well. The transmissivity and other hydraulic data for an aquifer are assumed constant in each 1-mi2 grid cell.
The ground-water flow model should not be used for analysis if large pumpages are placed near boundaries representing the outcrop area (a head-dependent flux boundary), near a ground-water flow divide (the Tombigbee River), or near the downdip limit of freshwater. The assumption of a fixed freshwater-saltwater interface boundary used in the downdip areas of the aquifers may not be valid if large-capacity pumping wells were placed nearby. The model is not designed to estimate movement of the freshwater-saltwater interface or to evaluate any change in water quality.
Sand and clay thickness maps used to develop the calibrated model are based on total thicknesses for the units derived from borehole-geophysical log analyses that were gridded to a 1-mi2 grid. In some areas, sand and clay thicknesses for the aquifers vary greatly over short lateral distances, and thicknesses may actually vary substantially within a grid cell. In some areas, data points were widely spaced, and sand and clay thicknesses had to be extrapolated or interpolated over a broad area, possibly misrepresenting actual conditions where data were not available.
The best available historical pumpage estimates were used in the simulations (J.H. Hoffmann and A.J. Warner, Mississippi Office of Land and Water Resources, written commun., 1997); however, determining the exact values of historical pumpage for the aquifers is difficult. Reported pumpage values for recent years could not be verified for accuracy. If large inaccuracies in modeled pumpage exist, the model would not be considered properly calibrated.
Projected water withdrawals and the locations of these withdrawals may change over 50 years. Distributions will likely change to some degree; some large pumping centers may be added while others may cease pumping altogether. If large unforeseen changes in pumpage occur in the Union County area, the projections as simulated would be inaccurate. However, the farther away from Union County that such unforeseen changes occur, the less likely these changes will have a large affect on projections in Union County. Models are imperfect representations of a complex natural system; however, if used with caution and good judgment, they can be valuable tools.