The 2017 RAE/CAE Symposium (pfdf) on Urban Flooding in Nanjing underlined the major challenges faced with urban flooding in China. Several cities have suffered severe flooding in recent years. While flooding is a function of a number of interacting factors specific to each city, pluvial flooding from intense rainfall has often been a common factor. Real-time flood forecasting RTFF systems have a critical role to play in managing the evolving flood risk during an event, providing information needed to support decision-making in issuing warnings and implementing various Risk Reduction Actions (RRAs) through mobilizing the emergency services, and organizing flood resistance/evacuation measures to protect citizens. However, forecasts are inherently uncertain, and this uncertainty should be explicitly quantified and used within appropriate decision-making procedures that can minimize costs, economic losses, loss of life and social disruption. Yet, many flood forecasting systems employ a deterministic forecasting approach in which forecast errors are not acknowledged, leading to sub-optimal decisions and consequent costs and losses could be reduced.

Research on RTFF over the past decade has demonstrated that probabilistic forecasts can achieve higher success rates if the forecast predictive uncertainty is used in decision-making. Yet, forecasting policy and practice remains rooted in the traditional deterministic approach whereby a model forecast is treated as reality rather than a virtual reality subject to error. This error is a function of lead time and emanates from errors in model inputs (space-time measurements of precipitation using radar/raingauges), model structure error, parameter estimation error, etc. Where Ensemble Quantitative Precipitation Forecasts are used to extend lead time, the ensemble forecast error is invariably not correctly quantified through a proper Bayesian predictive uncertainty procedure. Flood forecasting agencies are aware of the problem, but are unsure how to use uncertainty information in their warning decision-making procedures, or do not wish to acknowledge uncertainty explicitly. The barrier to progress lies in a failure to convince policy-makers and practitioners of the benefits that can be achieved through the use of uncertainty information in decision-making which is well established in other fields.

This web site explores uncertainty by replicating the components of an end-to-end flood forecasting system.

Please read our project Overview where our research goals are defined.

This project has been funded by the Royal Academy of Engineering under the UK China Urban Flooding Research Impact Programme.