APPROACHES TO OPTIMAL AQUIFER MANAGEMENT AND INTELLIGENT CONTROL IN A MULTIRESOLUTIONAL DECISION SUPPORT SYSTEM

Despite remarkable new developments in stochastic hydrology and adaptations of advanced methods from operations research, stochastic control, and artificial intelligence, optimal solutions of complex real-world problems in hydrogeology have been quite limited. The main reason is the ultimate reliance on first-principle models that lead to complex, distributed-parameter partial differential equations (PDE) on a given scale. While the addition of uncertainty, and hence, stochasticity or randomness has increased insight and highlighted important relationships between uncertainty, reliability, risk, and their effect on the cost function, it has also (a) introduced additional complexity that results in prohibitive computer power even for just a single uncertain/random parameter, and (b) it led to the recognition in our inability to assess the full uncertainty even when including all model uncertainties. Thus, we are left with the need in deterministic solutions in a stochastic world, with limited ability to assess the uncertainty. We suggest a paradigm shift - from reliance on rigid, limited and uncertain PDE models at the heart of the optimization scheme to a goal-oriented multiresolutional decision support system (MRDS) that continually integrates all the knowledge about the hydrogeologic system (geophysics, geology, geochemistry, biology, thermodynamics, hydrology) in multiresolutional (MR) knowledge representations, and uses MR search to optimize complex aquifer management systems in real time. Intelligent control combines methods of operations research (OR), control theories, and artificial intelligence (AI). Intelligent control is a computationally efficient procedure of directing a complex system with incomplete and inadequate representation, and under incomplete specifications of an uncertain environment, towards a certain goal. The main function of intelligence is global optimization of complex systems that are not sufficiently represented. An intelligent control agent such as MRDS, which could combine optimal management with site characterization, would benefit from the advantages and eliminate the limitations of current models (deterministic or stochastic) by integrating them as interpretations and initial guides within ever-changing MR knowledge representations.