E9.10 The Madden Julian Oscillation: CESM1 and E3SMv1

Abstract

The Madden Julian Oscillation (MJO) is a key mode of sub-seasonal variability and predictability, not just in the tropics, where it is reflected in convective organization time and space scales, but also in the remote influences affecting seasonal meteorology: features such as monsoons, Atmospheric Rivers and the NAO. Changes in the local and remote influence of the MJO in a warmer future could manifest in a number of ways including amplitude, frequency and phase modifications, impacting climate and predictability profoundly.

To assess potential future impacts, we examine a number of CESM1 large-ensemble experiment sets. MJOs produced by the large-ensemble have a number of deficiencies, but they do exhibit intra-seasonal, large-scale coupled variability that explains a significant portion of the total variability. To detect a real change in this metric in the future, we examine its spread from all large-ensemble members in 1° historical and RCP8.5 experiments. It is clear that to discount MJOs from similar ensemble members being from the same population, and reject the influence of internal variability, warming has to be at the level predicted to occur at the very end of this century.

Addressing the robustness of this result to resolution we examine a similar, but smaller, subset of simulations at a much higher 0.25° resolution. Increased resolution gives rise to a stronger and more realistic MJO and equivalently a future MJO that is stronger; more so that for the future stronger MJO seen at low-resolution.

Finally, we show preliminary results from the most recent releases of CMIP6 models, E3SMv1 and CESM2. These models share many atmospheric physics elements, and correspondingly the characteristics of the MJO are similarly much improved. In particular, there is now propagation of events extending well out into the central Pacific. By contrasting the behavior of un-coupled versus coupled behavior there are significant differences in performance, where MJO events are better represented in the fully coupled model versions. This model result re-ignites the question regarding intra-seasonal surface coupling and its role in at least sustaining (and maybe initiating) the MJO.