Ensemble predictions and perceptions of risk, uncertainty, and error in flood forecasting

Abstract

Under the auspices of the World Meteorological Organization, there are a number of international initiatives to promote the development and use of so-called ensemble prediction systems (EPS) for flood forecasting. The campaign to apply these meteorological techniques to flood forecasting raises important questions how the probabilistic information these systems provide can be applied for what in operational terms is typically a binary decision of whether or not to issue a flood warning. To explore these issues, we report on the results of a series of focus group discussions conducted with operational flood forecasters from across Europe on behalf of the European Flood Alert System. Working in small groups to simulate operational conditions, forecasters engaged in a series of carefully designed forecasting exercises. using various different combinations of actual data from real events. Supplemented by results from a follow-up questionnaire survey, these exercises suggest that flood forecasters may understand risk and uncertainty differently than promoters of EPS intend. The paper concludes by exploring the implications of these divergent ‘epistemic cultures’ for efforts to apply ensemble prediction techniques developed in the context of weather forecasting to the rather different one of flood forecasting.