Megadroughts and megapluvials are multi-year dry and wet events of exceptional intensity and duration. There is strong paleoclimate evidence for the existence of such events in various regions of the globe. However, the mechanisms how these anomalies can be sustained for long periods of time have not been elucidated and reproduced in climate models. Here we address this question using large-ensemble simulations with the CESM2 fully-coupled climate model. We argue that, outside ENSO-influenced regions, meteorological megadroughts and megapluvials in the simulations are mainly caused by natural variability in the atmosphere with limited influence from the oceans.

We first show that interannual correlations in accumulated precipitation are weak: multi-year extreme precipitation events arise as a succession of independent yearly events. Secondly, anomalous SST patterns do not explain the intensity of dry and wet years. Finally, we propose a simple, purely stochastic model of atmospheric precipitation that can reproduce most of the key characteristics of simulated extreme dry and wet years. These results imply that, outside ENSO-influenced regions, megadroughts and megapluvials in the climate model are caused by the succession of independent dry and wet years, which are themselves stochastically expected large deviations arising from the natural variability of the atmosphere, and not from the interaction with the slower ocean. However, the intensity and frequency of recorded megadroughts and megapluvials are not compatible with these model results. This strongly suggests that key physical mechanisms are missing in the model to reproduce these peculiar events and advocates for caution in estimating the probability of multi-year dry and wet events from climate model simulations.

Robin Noyelle, ETH Zürich

Lieu
École normale supérieure – PSL
24 rue Lhomond – aile Erasme
salle Claude Froidevaux – E314