MULTIOBJECTIVE RISKS IN RESERVOIR OPERATION

Reservoir operation objectives are considered by maximizing an objective function while maintaining various minimum values of a second objective in an optimization constraint. This is often referred to as the ϵ‐constraint approach; a priori estimates of objective worth are thereby eliminated. The constrained optimizations are repeated for many synthetic sequences of inputs to evaluate the risk associated with attainment of one objective for a given value of another objective. A trade‐off level for design of the reservoir operation rule is then subjectively selected based upon consideration of objective trade‐offs and their risks. An operation rule is estimated from the optimum decision set associated with the selected trade‐off level, by using Implicit Stochastic Optimization methods already available. These techniques are applied to the practical problem of the Coralville reservoir, near Iowa City, Iowa, for flood control and recreation operating purposes and resulting operation rules are compared to the existing operation rule. Copyright 1979 by the American Geophysical Union.