Enhancements to the Mississippi Embayment Regional Aquifer Study (MERAS) groundwater-flow model and simulations of sustainable water-level scenarios

Abstract

Arkansas continues to be one of the largest users of groundwater in the Nation. As such, long-term planning and management are essential to ensure continued availability of groundwater and surface water for years to come. The Mississippi Embayment Regional Aquifer Study (MERAS) model was developed previously as a tool to evaluate groundwater availability within the Mississippi embayment, which encompasses much of eastern Arkansas where the majority of groundwater is used. The Arkansas Water Plan is being updated for the first time since 1990 and serves as the State’s primary, comprehensive water-resources planning and guidance document. The MERAS model was selected as the best available tool for evaluation of specific water-use pumping scenarios that are currently being considered by the State of Arkansas. The model, developed as part of the U.S. Geological Survey Groundwater Resources Program’s assessment of the Nation’s groundwater availability, is proving to be invaluable to the State as it works toward development of a sustained yield pumping strategy. One aspect of this investigation was to evaluate multiple methods to improve the match of observed to simulated groundwater levels within the Mississippi River Valley alluvial and middle Claiborne (Sparta) aquifers in the MERAS model. Five primary methods were evaluated: (1) explicit simulation of evapotranspiration (ET), (2) upgrade of the Multi-Node Well (MNW2) Package, (3) geometry improvement within the Streamflow Routing (SFR) Package, (4) parameter estimation of select aquifer properties with pilot points, and (5) modification of water-use estimates. For the planning purposes of the Arkansas Water Plan, three scenarios were developed to evaluate potential future conditions: (1) simulation of previously optimized pumping values within the Mississippi River Valley alluvial and the middle Claiborne aquifers, (2) simulated prolonged effects of pumping at average recent (2000–5) rates, and (3) simulation of drawdown constraints on most pumping wells. The explicit simulation of ET indicated little, if any, improvement of model fit at the expense of much longer simulation time and was not included in further simulations. Numerous attempts to fully utilize the MNW2 Package were unsuccessful in achieving model stability, though modifications made to the water-use dataset remained intact. Final improvements in the residual statistics may be attributed to a single method, or a cumulative effect of all other methods (geometry improvement with the SFR Package, parameter estimation with pilot points, and modification of water-use estimates) attempted. The root mean squared error (RMSE) for all observations in the model is 22.65 feet (ft) over a range in observed hydraulic head of 741.66 ft. The RMSE for water-level observations in the Mississippi River Valley alluvial aquifer is 14.14 ft (an improvement of almost 3 ft) over a range in observed hydraulic head of 297.25 ft. The RMSE for the Sparta aquifer is 32.02 ft (an improvement of approximately 3 ft) over a range in observed hydraulic head of 634.94 ft. Three scenarios were developed to utilize a steady-state version of the MERAS model. Scenario 1 was developed to use pumping values resulting from the optimization of baseline rates (typically 1997 pumping rates) from previous optimization modeling of the alluvial aquifer and the Sparta aquifer. Scenario 2 was developed to evaluate the prolonged effects of pumping from the alluvial aquifer at recent pumping rates. Scenario 3A was designed to evaluate withdrawal limits from the alluvial aquifer by utilizing drawdown constraints equal to an altitude of approximately 50 percent of the predevelopment saturated thickness of the alluvial aquifer or 30 ft above the bottom of the alluvial aquifer, whichever was greater. The results of scenario 1 indicate large water-level declines throughout the area of the alluvial aquifer, regardless of the substitution of the optimized pumping values from earlier model simulations. The results of scenario 2 also indicate large areas of water-level decline, as compared to half of the saturated thickness, throughout the alluvial aquifer. The results of scenario 3A reveal some effects from the inclusion of multiple aquifers in a single simulation. The initial configuration of scenario 3A resulted in water levels well below the defined drawdown constraint, and some areas of depleted aquifer (water levels that are near or below the bottom of the aquifer) in east-central Arkansas. A fourth simulation (scenario 3B) was configured to apply the same drawdown constraints from the alluvial aquifer wells to the Sparta aquifer wells in the depleted area. These drawdown constraints reduce leakage from the alluvial aquifer to the underlying Sparta aquifer. This configuration did not produce depleted areas within the alluvial aquifer. Scenarios 3A and 3B indicate that even when pumping is limited in the alluvial aquifer, water levels in the alluvial aquifer may continue to decline in some areas because of pumping in the underlying Sparta aquifer.