Is the fate of Mesoscale Convective Systems written from the start?

Mesoscale Convective Systems (MCSs) that become large or have long lifespans contribute disproportionately to extreme rainfall. Gaining a better understanding of the factors that determine whether a system will become large could improve our understanding of extreme weather phenomena. The recent emergence of high-resolution global simulations from the DYAMOND project, coupled with a storm tracking algorithm called TOOCAN, provides a groundbreaking opportunity to study the factors controlling the maximum area of MCSs. In this study we use machine learning algorithms to predict the maximum area of convective systems based on their early development stages and initial environmental conditions. The results reveal that the initial evolution of the system anticipates its maximum area. Factors such as the presence of ice in the system's environment, proximity to surrounding systems, intensity of vertical velocity at 500 hPa, and the migration distance, have been identified as significant factors in improving the accuracy of the prediction. Using a linear model, we investigate the relative role of the environment and of the system itself, in the growth of the system.