Cyclonic development in the Mediterranean Basin in CMIP6 models using a neural network approach

The Mediterranean basin is located between the subtropical high-pressure belt and the mid-latitude westerlies and is characterized by complex topography. Its orography, the relatively warm Mediterranean Sea, which is a source of energy and moisture, and the land-sea interactions result in significant cyclonic behavior in the synoptic and sub-synoptic scales. Due to its high sensitivity to climate change forcings, the Mediterranean region is considered a climate change hot spot, with impacts, such as the decline in the projected precipitation, leading to increasing aridification in an already water-stressed area. The above highlight the importance of examining the cyclonic development in the area and assessing the respective changes under different climate change scenarios. In this research, unsupervised machine learning algorithms are used in order to objectively identify cyclonic development in the Mediterranean basin using CMIP6 data for a subset of the different shared socio-economic pathways (SSP) that explore a wide range of possible future outcomes. In more detail, Sea Level Pressure from selected CMIP6 models is used as an input in a Self-Organizing Map (SOM) which is trained to identify the cyclone activity in the Mediterranean Basin for the 1981-2010 reference period. The ability of the network in terms of identifying effectively cyclogenesis regions and the transition probabilities is evaluated. The trained SOM is used to classify CMIP6 mid-century (2031-2060) projections and changes in the frequencies of occurrence of cyclonic development. These are evaluated in terms of physical drivers and regionally specific mechanisms. Examining the responses of cyclonic development to different forcing scenarios will not only shed light on the physical and dynamical processes that govern these circulations but will also allow identifying high–risk regions with potential socio-economic impacts.