A Matter of Scale: Thermodynamic and Large-Scale Constraints in Extreme Rainfall Under a Changing Climate

Abstract

Changes in the hydrological cycle and, in particular, in rainfall extreme events induced by global warming are expected to pose significantly increased hazards in the coming decades. However, changes in the probability of occurrence of intense precipitation remain poorly understood even in observations. Here we investigate the thermodynamic and large-scale constraints to the generation of extreme rainfall at both hourly and daily scales. To this aim, we address some of the ambiguities intrinsic to the traditional definition of the dependence of extreme rainfall on temperature as mediated by the Clausius-Clapeyron (CC) relation. For this purpose, we use a non-asymptotic extreme value distribution (Marani and Ignaccolo, 2015) as a basis for our analysis. In this framework, the distribution of extremes emerges from the distribution of the ordinary events, here allowed to vary under climate change. The distribution of annual maxima is expressed as a function of the probability distribution of all events (that may be inferred using most of the available data, rather than just on yearly maxima) and of the number of event occurrences per year. The rationale here is that a warming of the atmosphere will affect the distribution of all rainfall events, i.e. the shape of the ordinary event distribution, rather than just rainfall extremes as in traditional CC arguments. Based on this approach, we then analyze the relation between the parameters of the probability distribution of ordinary precipitation events and temperature at the daily and hourly scales, using observational data in Padova, Italy (where almost 300 years of observations are available) and multiple stations in the continental US. While local temperature is widely considered to be a major driver of change in rainfall regimes, changes in large-scale circulation are also expected to play a significant role in shaping future rainfall regimes. In order to represent the effects of large-scale circulation, and analyze changes that remain unexplained by local temperature, we compute here the Vertically Integrated Moisture Convergence, derived from the ECMWF Reanalysis v5 (ERA5) dataset. Our results indicate that hourly precipitation is mainly controlled by thermodynamics, with the scale parameter of the probability distribution of hourly precipitation intensity showing a CC dependence. Conversely, at the daily scale, we show that precipitation variability is not explained by temperature changes but is rather driven by other factors such as large-scale circulation. These results support the need for an integrated approach, which quantitatively accounts for both local thermodynamics and large-scale circulation to estimate future changes in daily precipitation extremes under a climate change.