Are dependencies of extreme rainfall on humidity more reliable in convection-permitting climate models?

Abstract. Convection-permitting climate models (CPMs) are becoming increasingly used in climate change studies. These models show greatly improved convective rainfall statistics compared to parameterized-convection regional climate models (RCMs), but are they also more reliable in a climate change setting? Increases of rainfall extremes are generally considered to be caused by increases in absolute humidity, primarily following from the Clausius-Clapeyron relation, while the influence of relative humidity changes is uncertain and not systematically explored. Quantifying these humidity dependencies in the present-day climate may help the interpretation of future changes, which are driven by increases in absolute humidity, but also decreases in relative humidity in most continental areas in summer. Here, we systematically analyse hourly rainfall extremes and their dependencies on 2 m dew point temperature (absolute humidity) and dew point depression (relative humidity) in 7 RCM and 5 CPM simulations for the present-day climate. We compare these to observations from The Netherlands (a moderate moist climate) and Southern France (a warmer and drier climate). We find that the RCMs display a large spread in outcomes, in particular in their relative humidity dependence, with a strong suppression of hourly rainfall extremes in low relative humidity conditions. CPMs produce better overall rainfall statistics, show less inter-model spread, and have absolute and relative humidity dependencies more consistent with the observations. In summary, our results provide evidence that future changes in convective rainfall extremes in CPMs are more reliable compared to RCMs, whereas the discussed dependencies also provide a metric to evaluate and further improve model performance as well as improving convection schemes.