Improving forecast skills for MediterraneanTropical-like Cyclones using Data Assimilation: the case study of medicane Qendresa

This study investigates the meteorological characteristics of the tropical-like cyclone Qendresa, which occurred over the central Mediterranean Sea between 6 and 8 November 2014, using the WRF limited-area model. Qendresa originated during the early hours of 6 November through a “lee cyclogenesis” process east of the Algerian Atlas Mountains. The cyclone subsequently deepened as it moved towards the island of Pantelleria, then shifted southeastward, crossing Linosa Island, where its core reached the absolute minimum of sea level pressure. The system then continued towards Malta and, already weakened, moved off the eastern coast of Sicily, describing a loop-like trajectory on the morning of 8 November, before eventually moving eastward towards Greece. The event resulted in three fatalities and significant structural damage across the affected regions.
The primary aim of this work is to reproduce the cyclone’s structure and evolution, ensuring the best possible consistency with available observations, during the 36-hour period from 00 UTC on 7 November to 12 UTC on 8 November 2014. To enhance the identification of key physical processes and improve the accuracy of the simulated cyclone track, observational and reanalysis data at different vertical levels and time steps have been assimilated into the model. In order to do that, the Three-Dimensional Variational Data Assimilation (3DVar) technique was tested in different configurations, creating the Background Error matrix using 3 different strategies. Then, chosen the best performing one among them, three different experiments assimilating conventional MADIS data, Atmospheric Motion Vectors (AMVs) and other surface wind speed and direction datasets were performed, in order to assess the effects of assimilating conventional and unconventional data on the model's performance on this event. 
The results clearly show the low predictability of this event and a possible improvement may be achieved by using a multi-physics framework that explicitly resolves coupled air-sea-land processes. Also they have direct implications for coastal early warning systems, improving the representation of interaction-driven hazards and enhancing predictive capabilities in vulnerable coastal environments.