Bayesian inference and decision making for extreme hydrologic events

Hydrologic decision making usually occurs in an uncertain world and should combine inferences about uncertain outcomes with the decision makers’ preferences toward these outcomes. The decisions associated with flood protection are considered in detail where there exists uncertainty in the frequency of future flood discharges. Procedures are developed for analyzing and accounting for both statistical uncertainty of competing flood frequency models and statistical parameter uncertainty for the individual models. Inferences about flood frequency are combined with a decision model for which the decision rule is the maximization of expected monetary benefits. A case study to determine the optimal size of local flood protection by using prior information from a regional regression model, historical data, and realistic flood cost and damage functions is presented.