The Mediterranean region is experiencing pronounced aridification and in certain areas higher occurrence of intense precipitation. In this work, we analyze the evolution of the rainfall probability distribution in terms of precipitating days (or “wet-days”) and all-days quantile trends, in Europe and the Mediterranean, using the ERA5 reanalysis. Looking at the form of wet-days quantile trends curves, we identify four regimes. Two are predominant: in most of Northern Europe the rainfall quantiles all intensify, while in the Mediterranean the low-medium quantiles are mostly decreasing as extremes intensify. The wet-days distribution is then modeled by a Weibull law with two parameters, whose changes capture the four regimes. Assessing the significance of the parameter changes over 1950–2020 shows that a signal on wet-days distribution has already emerged in Northern Europe (where the distribution shifts to more intense rainfall), but not yet in the Mediterranean, where the natural variability is stronger. We extend the results by describing the all-days distribution change as the wet-days’, plus a contribution from the dry-days frequency change, and study their relative contribution. In Northern Europe, the wet-days distribution change is the dominant driver, and the contribution of dry-days frequency change can be neglected for wet-days percentiles above about 50\%. In the Mediterranean, however, the contribution to all-days change of wet-days distribution change is much smaller than the one of dry-days frequency. Therefore, in the Mediterranean the increase of dry-days frequency is crucial for all-days trends, even when looking at heavy precipitations.

Deep convection occurs periodically in the Gulf of Lion, driven by the seasonal atmospheric change and Mistral winds. To determine the variability and drivers of the seasonal and Mistral forcing, 20 years of ocean simulations were run. Two sets of simulations were performed: a control set, forced by unfiltered atmospheric forcing, and a seasonal set, forced by filtered forcing. The filtered forcing retained the seasonal aspects but removed the high frequency phenomena. Assuming the Mistral acts primarily in the high frequency, comparing the two sets allows for distinguishing the effects of the Mistral on the ocean response. During the preconditioning phase, the seasonal forcing was found to be the main destratifying process, removing on average 45.7% of the stratification, versus the 28.0% removed by the Mistral. Despite this difference, at the time of deep convection, both the seasonal and Mistral forcing each triggered deep convection in roughly half of the events. Larger sensible and latent heat fluxes were found in the seasonal forcing of the years with deep convection, acting as the main drivers (removing 0.17 m2s-2 and 0.43 m2s-2 of stratification, respectively). They are themselves driven by increased wind speeds, believed to be the low frequency signal of the Mistral, as more Mistral events occur during winters with deep convection (34.3% versus 28.6%). The evolution of the seasonal forcing in a changing climate may have a significant effect on the future deep convection cycle of the Gulf of Lion.