Welcome to NOAH
Balancing Water Supply with Vulnerability to Contamination
- Toms River, New Jersey

Problem Definition

As competition for increasingly scarce ground water resource grows, many decision makers may come to rely upon rigorous multiobjective techniques for helping to identify appropriate and defensible policies, particularly when disparate stakeholder groups are involved. In this problem, the coastal New Jersey community, Toms River, derives a large portion of its drinking water from the Parkway Well Field, which withdraws water from an unconfined aquifer.

As shown by Figure 1, a nearby Superfund site, formerly known as Reich Farm, has produced a ground water contaminant plume in the aquifer, contaminating two municipal supply wells in the Parkway Wellfield, with four nearby supply wells remaining vulnerable to contamination. Because of high water demand during summer tourism months, and the availability of only a few alternative water sources without significant capital investment, the water utility must balance water supply with well vulnerability to contamination. The stakes of this decision-making problem are compounded by the results of a six year, $10 million dollar epidemiological study completed on the Township in 2002, which concluded that prenatal exposure to contaminated Parkway well water over the years 1982 to 1996 was a risk factor for leukemia in female children living in the community (New Jersey Department of Health and Senior Services, 2003). Artificial neural networks (ANNs) were developed with simulation data from a numerical ground water flow model developed for the study area to accurately predict dynamic groundwater levels in response to variable wellfield pumping and regional recharge rates. The ANN-derived state-transition equations were embedded into a multiobjective optimization model, from which the Pareto frontier or tradeoff curve between water supply and wellfield vulnerability was identified. Relative preference values and power factors were subsequently assigned to the three stakeholders, namely the company whose waste contaminated the aquifer, the community supplied by the wells, and the water utility that owns and operates the wells. A compromise pumping policy that effectively balances the two conflicting objectives in accordance with the preferences of the three stakeholder groups was then identified using various distance-based methods.



The multiobjective analysis presented here considered the conflicting objectives of maximizing water supply volume via pumping while minimizing well vulnerability to contamination over a 12-month planning horizon. Using MODFLOW simulation results from a ground water model of the system developed by the New Jersey Geological Survey (NJGS), an ANN was trained to accurately predict ground water levels at points of interest in the aquifer in response to monthly changes in pumping and areal recharge rates. Difference in water levels between paired cell locations located on opposite sides of the plume boundary was used to quantify wellfield vulnerability to contamination.

Because higher pumping by the "clean" wells makes them more vulnerability to contamination (i.e. contaminant capture), the objectives of maximizing water supply via the supply wells while minimizing their vulnerability to contamination are conflicting. The resulting ANN-derived state-transition equations were then embedded into LINGO (LINDO, 1999), a commercial optimization program, to generate the Pareto frontier. A comparison between the ANN-derived and MODFLOW-verified non-normalized Pareto frontiers is shown in Figure 2 and shows a very close match

Various multiobjective methodologies were then applied to identify the best compromise solution for balancing water supply with vulnerability to contamination. The optimal policy, identified in accordance with assumed preferences and priorities of the stakeholders, the water utility can operate the well field at approximately 60 percent of its maximum, while satisfying customer concerns that the existing supply is being protected to the extent practicable. That is, as shown by Figure 3, the flow paths of the contaminant plume clearly converge to the three recovery wells inside the plume, protecting the nearby clean supply wells from contamination.

A more detailed overview of this work can be found in the following journal articles:

  • Coppola, E., M. Poulton, E. Charles, J. Dustman, and F. Szidarovszky. (2003). Application of Artificial Neural Networks to Complex Groundwater Management Problems. Journal of Natural Resources Research, 12, no. 4: 303-320.
  • Coppola E., F. Szidarovszky, D. Davis, S. Spayd, M. Poulton and E. Roman. Multiobjective Analysis of a Public Wellfield Using Artificial Neural Networks, Journal of Ground Water, in press.