A behavioural model for quantifying flood warning effectiveness

The extent of losses avoided as a result of a warning is a key measure of warning system effectiveness. Tools to estimate the impact of warnings on losses are limited to postflood analysis or estimates of potential rather than actual damages. This paper illustrates a method for the appraisal of actual damages when a flood warning is issued. The approach combines social science with engineering approaches to the problem of flood warning effectiveness. From a starting point of estimating potential damages by means of depth-damage curves, the method allows the identification of damage reduction by modelling how people respond to the warning. The model is in the form of an event tree representing human behavioural steps in the flood warning process. Two Australian case studies show how to apply the developed methodology. The results from these cases demonstrate the utility of the event-tree model that also allows the identification of weak links in the warning chain.