Recommended citation: Pérez-Alarcón, A.; Coll-Hidalgo, P.; Nieto, R., Gimeno-Sotelo, L., Gimeno, L. (2023). Historical dataset of Mediterranean cyclones from the high-resolution ECMWF ERA5 reanalysis. MedGU Annual Meeting. Istanbul, Türkiye, 26-30 November 2023.

Abstract

The Mediterranean region (MR) shows a high annual frequency of cyclone activity, causing property damages and human losses in the densely populated coastal areas of the Mediterranean Sea. Unlike their tropical counterparts, there are no historical records of the intensity and trajectories of medcyclones from observations. Examining their features is a challenging subject of atmospheric sciences. Although previous climatological analyses have applied cyclone detection and tracking methods, most of the resulting long-term datasets, including dates of occurrence, position in latitude and longitude, intensity, size and cyclone type (i.e. with extratropical, hybrid or tropical features), are not public available. Therefore, this work proposes a historical dataset of medcyclones (MEDCYDAT) from the high-resolution ECMWF ERA5 reanalysis from 1950 to 2022. The critical cyclone centres were identified based on the local mean sea level pressure (MSLP), while the next track location at t + 6 hours was the nearest low to the original low at time t. The tracking method produced 73-year of cyclone tracks within the MR defined by a rectangular domain between 25°N-50°N and 5°W-40°E. The resulting database is a comma-delimited text format following a similar structure to the HURDAT2 dataset provided by the US National Hurricane Center for tropical cyclones. Although it is difficult to assess the ability of the tracking methods for detecting cyclones in the MR due to no reference dataset for benchmarking (i.e., best-track for tropical cyclones), the resulting tracks from this work mostly agree with previous climatological studies on the annual and seasonal cyclone frequency, genesis areas and track density. Therefore, MEDCYDAT can be useful for different applications, such as mapping potential storm surge risk in coastal regions, benchmarking the performance of different artificial neural networks to detect and track medcyclones, masking cyclone areas to identify the origin of moisture that support their development and associated precipitation, and assessing their contribution to monthly and annual precipitation totals in the MR.