Changes in probability of large precipitation events under global warming: Using theory to understand projections in the CMIP6 ensemble.

Abstract:

Observed probability distributions for precipitation exhibit two main regimes of behavior: a power law range for small events characterized by an exponent and a roughly exponential range for large events characterized by a precipitation scale, P_L. This scale provides a physically-motivated way of expressing the changes in extreme events, capturing behavior through a range of high percentiles. While water vapor increases tend to be governed by the Clausius-Clapeyron (CC) relationship to temperature, precipitation extreme changes are more complex due to dynamical feedbacks. In historical climate, Coupled Model Intercomparison Project Phase 6 (CMIP6) simulations exhibit familiar biases in the small-event regime, but reasonably capture the shape of the large-event regime and seasonal/spatial variations of P_L in historical climate, including effects of dynamical feedbacks. Under the SSP5-8.5 anthropogenic warming scenario, increases in P_L tend to be near or slightly greater than CC over broad regions in the multi-model ensemble. However, regions of substantially super-CC behavior are seen seasonally and in individual ensemble members, especially in the tropics. Considerations of the energy and moisture budgets simultaneously suggest that such departures from CC are to be expected, with moisture changes implying dynamical feedbacks except under special circumstances. Precipitation risk ratios increase steeply for the largest events, and higher percentiles tend to increase faster than moderate percentiles, in accordance with theory.