Atmospheric and Oceanic Response to Southern Ocean Deep Convection Oscillations on Decadal to Centennial Time Scales in Climate Models

Abstract:

Many current coupled global climate models simulate open ocean deep convection in the Southern Ocean as a recurring event with time scales ranging from a few years to centennial (de Lavergne et al., 2014, Nat. Clim. Ch.). The only observation of such event, however, was the occurrence of the Weddell Polynya in the mid-1970s, an open water area of 350 000 km² within the Antarctic sea ice in three consecutive winters. Both the wide range of modeled frequency of occurrence and the absence of deep convection in the Weddell Sea highlights the lack of understanding concerning the phenomenon. Nevertheless, simulations indicate that atmospheric and oceanic responses to the cessation of deep convection in the Southern Ocean include a strengthening of the low-level atmospheric circulation over the Southern Ocean (increasing SAM index) and a reduction in the export of Antarctic Bottom Water (AABW), potentially masking the regional effects of global warming (Latif et al., 2013, J. Clim.; Martin et al., 2014, Deep Sea Res. II). It is thus of great importance to enhance our understanding of Southern Ocean deep convection and clarify the associated time scales.

In two multi-millennial simulations with the Kiel Climate Model (KCM, ECHAM5 T31 atmosphere & NEMO-LIM2 ~2˚ ocean) we showed that the deep convection is driven by strong oceanic warming at mid-depth periodically overriding the stabilizing effects of precipitation and ice melt (Martin et al., 2013, Clim. Dyn.). Sea ice thickness also affects location and duration of the deep convection. A new control simulation, in which, amongst others, the atmosphere grid resolution is changed to T42 (~2.8˚), yields a faster deep convection flip-flop with a period of 80-100 years and a weaker but still significant global climate response similar to CMIP5 simulations. While model physics seem to affect the time scale and intensity of the phenomenon, the driving mechanism is a rather robust feature. Finally, we compare the atmospheric and oceanic responses among CMIP5 models. Since open ocean convection is the dominant mode of AABW formation in these models, the northward extent and strength of the AABW cell in the Atlantic correlates with the deep convection intensity but varies between models. Likewise, atmospheric response patterns outside the Southern Ocean region are not consistent among models.