Simulated Response of the Atlantic Meridional Overturning Circulation and Northern Hemisphere Climate to NAO Variations on Interannual to Centennial Time Scales

Thursday, 18 December 2014: 9:00 AM
Thomas L Delworth, NOAA, Princeton, NJ, United States and Fanrong Jenny Zeng, NOAA Princeton, Princeton, NJ, United States
We use large ensembles of coupled model experiments to assess the contribution of observed variations of the North Atlantic Oscillation (NAO) to AMOC variability and to Northern Hemisphere climate variations over the period 1901-2013. We use multiple versions of GFDL coupled models, and conduct ensembles of simulations in which the ocean component of the coupled model is forced with anomalous NAO related fluxes of heat, water, and momentum. The spatial pattern of these fluxes is derived based on the regression of ECMWF-Interim reanalysis fluxes versus observed NAO variations. The flux patterns are modulated in time according to observed variations over 1901-2013, idealized sinusoidal variations with periods from 2 to 100 years, or abrupt changes.

We first show that NAO-induced AMOC changes have a significant impact on hemispheric-scale temperature variations. For example, aspects of the widespread Northern Hemisphere cooling of the late 1960s and 1970s, as well as rapid warming in the 1980s and 1990s, are reproduced in these NAO-forced simulations. The hemispheric temperature changes occur in response to the observed switch from a strong negative phase of the NAO to a strong positive phase of the NAO, with associated weakening and strengthening of the AMOC. The NAO-induced AMOC changes translate into hemispheric scale temperature variations by modulating poleward oceanic heat transport, which is amplified through climate feedbacks involving shortwave radiation fluxes. For example, a warmed North Atlantic reduces sea ice and snow cover, thereby leading to enhanced absorption of shortwave radiation and further warming.

In a more idealized setting we show that the AMOC responds most clearly to NAO forcing with timescales comparable to, or longer than, the inherent AMOC variability of the model. This dependence is important in evaluating the impact of the NAO on decadal-scale temperature variations. Further, model adjustment to the imposed NAO forcing shares many aspects with the internal AMOC variability of the model, including modulation of Labrador Sea convection and the propagation of density anomaly signals.