Predictability of extreme surface weather associated with Mediterranean cyclones in ECMWF ensemble forecasts – Part 1: Method and case studies

Authors:
Open Access
Publisher website
Google Scholar
Abstract
Extratropical cyclones are the main cause of extreme surface weather events in the Mediterranean such as heavy precipitation, floods, severe winds, and dust storms. However, the accuracy in predicting the timing, location, and intensity of such events is often insufficient, which is typically related to errors in cyclone position, propagation, and intensity. In this two-part study we use operational ensemble forecasts from the European Centre for Medium-Range Weather Forecasts to quantify the predictability of extreme surface weather conditions linked to Mediterranean cyclones. We apply an object-based approach to attribute events of extreme precipitation and surface winds to Mediterranean cyclones. Thereby, objects of extreme surface weather are identified at grid points that exceed the seasonal 99th percentile of these parameters and matched to cyclones based on their distance to the cyclone center. In this first part, we introduce the probabilistic method and three illustrative case studies of Mediterranean cyclones that occurred between November 2022 and September 2023, including the infamous Storm Daniel as well as Storms Denise and Jan. We find that the cyclones as well as their attributed objects of extreme surface weather are predicted well for lead times ≤ 48 h. However, for longer lead times there is large case-to-case variability in the ensemble performance. Predictions of extreme surface weather objects are found to be more uncertain (i) for smaller and less coherent objects, (ii) if the associated cyclone is captured by fewer ensemble members, and (iii) during the earlier stage of the cyclones' lifecycle. The methodological development and its application documented in this paper provide the basis for a multi-year investigation of the predictability of extreme weather linked to Mediterranean cyclones in the second part of this study.
All Related Versions
Funding Information
  • Schweizerischer Nationalfonds zur Förderung der Wissenschaftlichen Forschung (205419)

This publication has 48 references indexed in Scilit: