Groundwater flow in the Brunswick/Glynn County area, Georgia, 2000-04

Abstract

An existing regional steady-state model for coastal Georgia, and parts of South Carolina and Florida, was revised to evaluate the local effects of pumping on the migration of high chloride (saline) water in the Upper Floridan aquifer located in the Brunswick/Glynn County, Georgia (Ga.) area. Revisions were focused on enhancing the horizontal and vertical resolution of the regional model grid in the vicinity of saline water. Modifications to the regional model consisted of (1) limiting grid size to a maximum of 500 feet (ft) per side in the vicinity of chloride contamination; (2) representing the upper and lower Brunswick aquifers with distinct model layers; (3) similarly, representing upper and lower water-bearing zones of the Upper Floridan aquifer with distinct model layers in Glynn and Camden Counties, Ga.; and (4) establishing new hydraulic-property zones in the Upper Floridan aquifer. The revised model simulated steady-state conditions that were assumed to exist during 2000 and 2004. Calibration of the revised steady-state model using pumping rates from 2000 indicates a "good" match (±10 ft) based on 181 observations, with median residuals (simulated minus observed water levels) in each of the active model layers ranging from -8.62 to 4.67 ft, and root mean square error (RMSE) ranging from 10.9 to 11.4 ft. In the Brunswick/Glynn County area, groundwater-level residuals in the upper water-bearing zone of the Upper Floridan aquifer (layer 7) indicate an "excellent" match (±5 ft) based on 41 observations with a median residual of -0.35 ft and RMSE of 4.32 ft. Calibration of the revised steady-state model using 2004 pumping rates and adjusted specified-head input values in the Floridan aquifer system indicates a "good" match (-10 ft) based on 88 observations, with median residuals in each of the active model layers ranging from -6.31 to -2.05 ft, and RMSE ranging from -6.95 to 14.5 ft. In the Brunswick/Glynn County area, groundwater-level residuals in the upper water-bearing zone of the Upper Floridan aquifer (layer 7) indicate an "excellent" match (±5 ft) based on 32 observations with a median residual of -1.50 ft and RMSE of 5.34 ft. Simulated potentiometric surfaces for 2000 and 2004 indicate coastward groundwater flow in the Upper and Lower Floridan aquifers influenced by pumping centers at Savannah, Jesup, and Brunswick, Ga., and indicate steep potentiometric gradients to the west and north of the Gulf Trough. In the Brunswick/Glynn County area, simulated industrial production wells located north of downtown Brunswick intercept local groundwater flow in the upper and lower water-bearing zones of the Upper Floridan aquifer and have created a cone of depression that locally alters the regional coastward flow direction. Maps of simulated water-level change during the 2000-04 period show differences in groundwater levels in the Upper Floridan aquifer that range from -2.5 ft to more than 5 ft in areas of coastal Georgia, and more than 20 ft near the Georgia-Florida State Line. Positive values indicate higher simulated water levels during 2004 than during 2000, which were caused by reduced pumping in the Upper Floridan aquifer prompted by the shutdown of a paper mill near the southern model boundary in 2002 and increased recharge following a prolonged drought during 1998-2002. Simulated potentiometric profiles for 2000 and 2004 were used to evaluate the potentiometric gradients in the upper water-bearing zone of the Upper Floridan aquifer (layer 7) near the chloride plume in the downtown Brunswick area. Four potentiometric profiles were constructed for 2000 to compare the simulated and observed water levels in 13 wells and were oriented outward from a primary well field. The simulated potentiometric gradients from the four profiles for 2000 ranged from 3.6 to 5.2 feet per mile (ft/mi) compared to observed values ranging from 4.1 to 5.6 ft/mi. The five potentiometric profiles constructed for 2004 allowed for a similar comparison using simulated and observed water levels in 18 wells. The simulated potentiometric gradients from the five profiles for 2000 ranged from 3.6 to 11.1 ft/mi compared to observed values ranging from 3.8 to 10.2 ft/mi. Simulated potentiometric gradients were higher for 2004 than for 2000 because of the inclusion of a well located within the cone of depression near downtown Brunswick. Composite-scaled sensitivities of the model parameters indicate the revised model is most sensitive to pumping rates, followed by the horizontal hydraulic conductivity in the Upper Floridan aquifer for zones along coastal Georgia. The revised model is least sensitive to the horizontal hydraulic conductivity of the confining units and vertical hydraulic conductivity of the aquifers. For parameters defined by hydraulic-property zones in the upper and lower water-bearing zones of the Upper Floridan aquifer, such as horizontal hydraulic conductivity, model sensitivity was not as great in the Brunswick/Glynn County area as other areas along coastal Georgia. The model exhibited more sensitivity to these parameters however, than to parameters representing the majority of zones defining the vertical hydraulic conductivity of the confining units, which originally were assumed to govern upward migration of chloride contamination into this aquifer. Analysis of simulated water-budget components for 2000 and 2004 indicate that specified-head boundaries in the Floridan aquifer system to the south and southwest of the regional model area control about 70 percent of inflows and nearly 50 percent of outflows to the model region. Other water-budget components indicate an 80-million-gallon-per-day decrease in pumping from the Floridan aquifer system during this period.