The investigations described in this review demonstrate the accelerated pace of research into simulation-optimization groundwater management methods. Major strides have been made to allow the formulation and solution of complex groundwater management problems. The five key aspects of recent research include: (1) improved formulation and solution techniques for solving stochastic groundwater management problems, (2) methods that link groundwater management modeling with parameter estimation and sampling network design, (3) dynamic groundwater management methods that enable the design of efficient time-varying groundwater control strategies, (4) methods for solving combinatorial groundwater management problems, and (5) groundwater quality management models that account for nonequilibrium transport of multiple contaminants.
Most research reviewed in this paper focused on simulation and optimization methods for attacking the important problem of aquifer remediation design. Many types of problems related to aquifer remediation remain to be solved. These include optimal management of contaminant transport with non-linear multicomponent reactions; remediation design for multiphase flow and transport; and optimal design involving in situ chemical and biological remediation. Naturally, these problems will have to be formulated within a stochastic framework to account for model uncertainty, so continued advancement in the area of stochastic simulation-optimization methods is very important.
Field studies have demonstrated the applicability of simulation-optimization methods to a variety of local and regional groundwater quantity and quality issues. It is interesting to note that a number of the studies cited [e.g. Varljen and Shafer, 1992; Chau, 1992; Finney et al., 1992; Marryott et al., 1993; Gharbi and Peralta, 1994;] identified management strategies that were superior to the strategy currently in operation. However, there remains an absence of published reports describing the implementation of groundwater management strategies derived from combined simulation and optimization models. That is not to say that simulation-optimization methods are not being used to aid in real-world decision making. One example is a large-scale, nonlinear management model that has been developed and implemented to manage groundwater pumpage and surface water deliveries in the Turlock and Modesto Irrigation Districts in California [ L.J. Lefkoff, personal communication]. Retrospective analyses of such studies would be invaluable because they would provide researchers with feedback regarding the practical limitations and needed extensions in groundwater management techniques.
Acknowledgments. The author is grateful for the helpful review comments provided by Paul Barlow, U.S. Geological Survey, and Roy Haggerty, Stanford University.