Connecting science, operations and decision-making when communicating uncertainty in hydro-meteorological forecasting

Almost ten years ago, we published a paper where we raised the question whether effective communication of uncertainty in hydro-meteorological forecasts was an impossible mission (Ramos et al. Meteorol. Appl. 17: 223–235, 2010, DOI: 10.1002/met.202). We wanted to understand if the multiple ways of interpreting uncertainty, as well as the multiple users and forecasting situations affecting forecast display and confidence, could hamper probabilistic forecast communication in operational hydrological forecasting. We looked at the main general interconnections present in a typical flood forecasting and alert chain, the challenges of extracting meaningful information from probabilistic forecasts and the way ensemble forecasts were effectively used in flood warning and decision-making. At the end, we were optimistic to say that the “mission is not impossible, although the tasks to be executed might be difficult to accomplish.” Here, we discuss a follow-up question: what have we accomplished in terms of communicating uncertainty in hydrological forecasts in practice, and contributing to better inform decision-making? The impact of forecasts, in terms of, for instance, anticipation of extreme events and crisis management, depends on how good they are but also on how they are understood and used in practice. This requires connecting science, operations and decision-making through the forecasting chain. We present some experiments with role-play games and benchmarking skilful streamflow forecasts developed to better understand the way probabilistic predictions can support decisions, and discuss where successes were achieved and challenges remain.