The Role of Sub-Seasonal Tropical Convective Variability for the Extratropical Response to ENSO

Abstract

The boreal winter extratropical response to tropical heating associated with a significant El Niño Southern Oscillation (ENSO) event has been understood primarily in terms of a seasonal average. However, the development and variability of the response forced by a highly sub-seasonally variable atmospheric forcing is still unclear. Significant modes of sub-seasonal tropical convective variability exist that have welldocumented transient impacts on the extratropics, e.g. the Madden-Julian Oscillation (MJO). Does sub-seasonal tropical convective variability impact the seasonal mean extratropical response? If so, what matters for the response: short-lived strong events or more persistent moderate events? Does sub-seasonal convective variability have implications for predictability during El Niño? Using the Community Atmospheric Model v. 4.0 (CAM4) of the National Center for Atmospheric Research, the response to ENSO is simulated using large ensembles of seasonal integrations forced with observed SST from multiple El Niño events. We decompose the diabatic heating rate (Q) across the tropical Indo-Pacific in terms of temporal and intra-ensemble variability. We then repeat the simulations by prescribing subsets of Q in an effort to diagnose the impact of sub-seasonal and intra-ensemble Q variability. Neglecting sub-seasonal Q variability has a systematic impact on the response, generally extending the Pacific jet and deepening upper-level heights across the North Pacific. A local enhancement is simulated in tropical upper-level divergence, likely related to the vertical redistribution of Q resulting from averaging together vertical profiles associated with moderate and deep convection. Persistence may also play a role. When prescribing nearly all Q variability (for frequencies greater than one day), we find that Q-circulation coupling is locally unimportant, i.e. the temporal evolution, mean, and variance of upper-level divergence from a coupled control simulation are reproducible. However, interestingly the amplification of the extratropical response to El Niño is still simulated. It follows that these differences are attributed to the neglect of tropical-extratropical two-way coupling. This is the primary finding of the study and suggests that the extratropical response to a significant El Niño event is not simply a forced response to tropical heating. Regarding the variability of the response to El Niño, we find a robust relationship in the seasonal mean ensemble spread, linking Q variations in the west/central tropical Pacific and extratropical fluctuations projecting onto the Arctic Oscillation (AO). This mode accounts for about a third of the ensemble spread (in 200 hPa geopotential height). Previous studies have revealed a similar relationship in observed interannual variability and have simulated the tropically forced component. In CAM4, this relationship in the ensemble spread is only evident during El Niño and is the dominant contributor to the enhanced ensemble spread simulated during El Niño, acting to lower predictability of the response. Extratropical forcing of the tropics is important for this mode, as prescribing Q variability weakens the relationship and cannot reproduce the enhancement in spread. We find that this relationship emerges as residue of a low frequency sub-seasonal tropical-extratropical coupled mode for which a significant extratropical influence is evident. Over the Pacific, coherent interaction is simulated between evolving persistent planetary waves and convectively coupled waves bearing a resemblance to equatorial Kelvin waves or MJO-like behavior. We hypothesize that the two-way tropicalextratropical interaction for this mode acts to weaken the extratropical response to El Niño.