Heavy water use from the Cretaceous Middendorf aquifer in South Carolina has created a large, regional cone of depression in the potentiometric surface of the Middendorf aquifer in Charleston and Berkeley Counties, South Carolina. Water-level declines of up to 249 feet have been observed in wells over the past 125 years and are a result of ground-water use for public-water supply, irrigation, and private industry. To address the concerns of users of the Middendorf aquifer, the U.S. Geological Survey, in cooperation with Mount Pleasant Waterworks, updated an existing ground-water flow model to incorporate additional data that have been compiled since 1989. The updated ground-water flow model incorporates water-level data collected from 349 wells in 2004, baseflow data measured at 17 streams, hydraulic property data from 265 wells, and water-use data compiled for more than 2,700 wells for the period between the early 1900s to 2004.
The ground-water flow system of the Coastal Plain physiographic province of South Carolina and parts of Georgia and North Carolina was simulated using the U.S. Geological Survey finite-difference code MODFLOW-2000. The model was vertically discretized into nine layers to include the five aquifers of the surficial, the combined Floridan aquifer system and Tertiary sand aquifer, Black Creek, Middendorf, and Cape Fear, separated by four intervening confining units. Specified-head boundary conditions were used at the lateral boundaries of the model and for the lower Coastal Plain part of the surficial aquifer; no-flow boundary conditions were used at the updip and downdip extent of the model layers and at the base of the Cape Fear aquifer.
Ground-water conditions for predevelopment and 2004 were simulated using steady-state and transient approximations, respectively. Simulated water levels generally matched the observed conditions, plus or minus a 20-foot calibration target, with 56.4 and 64.8 percent of the simulated values approximating the measured values for predevelopment and 2004 hydrologic conditions, respectively. The root-mean-square error of the water-level residuals for the various model layers varied between 20.2 and 34.4 feet for predevelopment and 18.2 and 36.7 feet for 2004. The general goodness of fit also was apparent in the calculation of the ratio of standard deviation of residuals to range of observations for each modeled aquifer layer. The calculated ratios for the predevelopment and 2004 hydrologic conditions were less than 0.10 for all model layers except for the Cape Fear aquifer in both predevelopment and 2004 simulations.
The Mount Pleasant model was most sensitive to changes in simulated specific storage of most model layers, vertical anisotropy of the confining units above and below the Middendorf aquifer, hydraulic conductivity of the confining units, and the specified-head boundary conditions for the surficial aquifer. The model also is sensitive to horizontal hydraulic conductivity of the Floridan aquifer system and Tertiary sand aquifer and the Black Creek and Middendorf aquifers. Simulated water budgets indicate that the primary sources of water to the model are recharge and the specified-head boundaries in layers 1 and 3. More than 88 percent of the water that discharges from the model discharges from layers 1-3 through specified-head boundaries and rivers. Approximately 11 percent of the water budget was discharged through wells for the 2004 budget. In 2004, 8.11 million gallons of water per day was discharged from wells in the Mount Pleasant area. Water to these wells is provided predominantly by lateral flow within the Middendorf aquifer. Additional water is provided from aquifer storage and leakage from confining units located above and below the Middendorf aquifer. Downward flow through the Middendorf confining unit is a reversal of the predevelopment flow direction.
Five predictive water-management scenarios were simulated to determine the effects on the