Probabilistic nowcasting and forecasting of extreme precipitation for impact-based warning systems. (EXPRIMA#2)

This research profile aims to address this growing need by providing the scientific basis for impact-based warning systems that cover a range of time scales while capturing the associated uncertainties. It will accomplish this by means of a unique integrative approach that considers the whole chain from the observations to the risk assessment for the end user, fully exploiting the probabilistic information at every step. This approach ensures that the uncertainties are correctly modelled and propagated through the highly nonlinear local impact models. Finally, it aims to deliver information that end users can combine with exposure and vulnerability maps to fully asses the risks and their probabilities. In order to meet the users’ evolving needs, the envisioned system will be the product of continuous co-design with stakeholders. This research profile aims to address the following key research questions: 1. Seamless prediction: how to optimally combine the latest observations and model output to develop a seamless, rapidly updating weather prediction system? 2. Data assimilation: how to overcome the challenges of assimilating high-resolution and novel meteorological observations at convection-permitting spatial scales? 3. Ensemble prediction: how to fully leverage the probabilistic information of an ensemble prediction system to generate (impact-based) warnings? 4. Insights from dynamical systems theory: can we employ these recent insights to characterize the predictability of precipitation and use this information to improve precipitation forecasts? 5. Towards impact-based warnings: how to design a fast, fully integrated impact-based warning system, covering multiple aspects of extreme precipitation such as pluvial flooding, hail, and droughts? 6. Societal and public service relevance: how to advise stakeholders in the integration of probabilistic forecasts with their exposure and vulnerability information to develop effective crisis management, prevention, relief and information strategies? To tackle these questions, we will leverage the RMIB’s knowledge of meteorological modelling and KU Leuven’s expertise on impact modelling and machine learning, as well as the shared expertise on meteorological observations, probabilistic methods and stochastic modelling of hydro-meteorological extremes to fully exploit probabilistic forecasts for impact-based warnings.