The Use of Numerical Modeling Techniques to Optimize Groundwater Withdrawls and Minimize Streamflow Depletion
David P. Ahlfeld
Groundwater reserves can be a valuable resource in alleviating rising water demands worldwide. The groundwater is generally pumped out of surficial aquifers using withdrawal wells. However, unregulated groundwater withdrawals can impact the flow in nearby streams negatively, causing streamflow depletions. This can hinder the ecological balance of streams and endanger the aquatic life fonTIs that call such streams home. Therefore, groundwater withdrawals must be carried out using a well managed withdrawal strategy. Rhode Island has a growing annual water demand that has been projected to exceed current supply in parts of the state by 2020. As a result, several new withdrawals have been proposed to be built in the Big River area. However, an optimized well withdrawal strategy needs to be developed with an eye towards minimizing any potential streamflow depletion that such activities might cause. Ground-Water Management (GWM), a groundwater optimization process for the three-dimensional groundwater modeling software MODFLOW 2000, is a tool that can help in developing such strategies. This study details the use of MOD FLOW 2000 to develop a three dimensional numerical model of the Big River area to examine how the withdrawals from thirteen hypothetical wells can be maximized while minimizing corresponding streamflow depletion at four specific stream sites within the model domain. Results provided include optimal attainable annual withdrawals under several different imposed instream flow criteria, sensitivity analysis, and comparisons between solutions obtained by two different MOD FLOW solver packages, Preconditioned Conjugate Gradient (PCG) and Geometric Multigrid (GMG). Results showed that to yield an optimal solution for the Big River problem with a satisfactory degree of precision, a planning horizon of three years and above using the Geometric Multigrid solver package of MOD FLOW 2000 is advisable. PCG failed to solve the Big River problem for values of the head close convergence criterion HCLOSE less than 0.01, whereas GMG yielded solutions for HCLOSE values as small as 0.0001. A planning horizon of three year or more was necessary for the withdrawals from the wells within the model domain to attain a state of dynamic equilibrium and thus make sense in terms of long tenn sustainability.