Precursors and pathways: dynamically informed extreme event forecasting demonstrated on the historic Emilia-Romagna 2023 flood

Abstract

The ever-increasing complexity and data volumes of numerical weather prediction demand innovations in the analysis and synthesis of operational forecast data. Here we show how dynamical thinking can offer directly applicable forecast information, taking as a case study the extreme northern Italy flooding of May 2023. We compare this event with long-lasting historical northern Italy rainfall events in order to determine (a) why it was so extreme, (b) how well it was predicted, and (c) how we may improve our predictions of such extremes. Lagrangian analysis shows, in line with previous work, that 48-hourly extreme rainfall in Italy can be caused by moist air masses originating from the North Atlantic; North Africa; and, to a lesser extent, eastern Europe, with compounding moisture contributions from all three regions driving the May 2023 event. We identify the large-scale precursors of typical northern Italy rainfall extremes based on geopotential height and integrated vapour transport fields. We show in European Centre for Medium-Range Weather Forecasts (ECMWF) operational forecasts that a precursor perspective was able to identify the growing possibility of the Emilia-Romagna extreme event 8 d beforehand – 4 d earlier than the direct precipitation forecast. Such dynamical precursors prove to be well suited for identifying and interpreting predictability barriers and could help build forecasters' understanding of unfolding extreme scenarios in the medium range. We conclude by discussing the broader implications and operational potential of dynamically rooted metrics for understanding and predicting extreme events, both in retrospect and in real time.