The Mississippi Embayment Regional Aquifer Study (MERAS): Documentation of a groundwater-flow model constructed to assess water availability in the Mississippi embayment

Abstract

The Mississippi Embayment Regional Aquifer Study (MERAS) was conducted with support from the Groundwater Resources Program of the U.S. Geological Survey Office of Groundwater. This report documents the construction and calibration of a finite-difference groundwater model for use as a tool to quantify groundwater availability within the Mississippi embayment. To approximate the differential equation, the MERAS model was constructed with the U.S. Geological Survey's modular three-dimensional finite-difference code, MODFLOW-2005; the preconditioned conjugate gradient solver within MODFLOW-2005 was used for the numerical solution technique. The model area boundary is approximately 78,000 square miles and includes eight States with approximately 6,900 miles of simulated streams, 70,000 well locations, and 10 primary hydrogeologic units. The finite-difference grid consists of 414 rows, 397 columns, and 13 layers. Each model cell is 1 square mile with varying thickness by cell and by layer. The simulation period extends from January 1, 1870, to April 1, 2007, for a total of 137 years and 69 stress periods. The first stress period is simulated as steady state to represent predevelopment conditions. Areal recharge is applied throughout the MERAS model area using the MODFLOW-2005 Recharge Package. Irrigation, municipal, and industrial wells are simulated using the Multi-Node Well Package. There are 43 streams simulated by the MERAS model. Each stream or river in the model area was simulated using the Streamflow-Routing Package. The perimeter of the model area and the base of the flow system are represented as no-flow boundaries. The downgradient limit of each model layer is a no-flow boundary, which approximates the extent of water with less than 10,000 milligrams per liter of dissolved solids. The MERAS model was calibrated by making manual changes to parameter values and examining residuals for hydraulic heads and streamflow. Additional calibration was achieved through alternate use of UCODE-2005 and PEST. Simulated heads were compared to 55,786 hydraulic-head measurements from 3,245 wells in the MERAS model area. Values of root mean square error between simulated and observed hydraulic heads of all observations ranged from 8.33 feet in 1919 to 47.65 feet in 1951, though only six root mean square error values are greater than 40 feet for the entire simulation period. Simulated streamflow generally is lower than measured streamflow for streams with streamflow less than 1,000 cubic feet per second, and greater than measured streamflow for streams with streamflow more than 1,000 cubic feet per second. Simulated streamflow is underpredicted for 18 observations and overpredicted for 10 observations in the model. These differences in streamflow illustrate the large uncertainty in model inputs such as predevelopment recharge, overland flow, pumpage (from stream and aquifer), precipitation, and observation weights. The groundwater-flow budget indicates changes in flow into (inflows) and out of (outflows) the model area during the pregroundwater-irrigation period (pre-1870) to 2007. Total flow (sum of inflows or outflows) through the model ranged from about 600 million gallons per day prior to development to 18,197 million gallons per day near the end of the simulation. The pumpage from wells represents the largest outflow components with a net rate of 18,197 million gallons per day near the end of the model simulation in 2006. Groundwater outflows are offset primarily by inflow from aquifer storage and recharge.