Influence of Ocean Diapycnal Diffusivity on the Amoc Stability and Variability

Thursday, 18 December 2014
Yonggang Liu1, Robert Hallberg2, Syukuro Manabe1, Bonnie Samuels2, Olga V Sergienko1 and Michael Oppenheimer1, (1)Princeton University, Princeton, NJ, United States, (2)NOAA GFD Lab-Princeton University, Princeton, NJ, United States
Spurious diapycnal mixing (κd) is often induced in the z-coordinate ocean model, which may be comparable in magnitude to the physical mixing. In order to investigate how the simulated climate may be affected by such spurious mixing, we use a fully coupled climate model, CM2G, in which the ocean module employs a density-coordinate in the vertical direction. The effect of spurious mixing in this model is much smaller than most z-coordinate models, allowing for a reliable specification of the values for κd. In a series of sensitivity tests, we increase κd by 1, 2, 4, 6 and 8 (unit: 10-5 m2 s-1), respectively, relative to the default value. In the first set of experiments, we run the model to equilibrium for each of the six κd (including the default value) under pre-industrial conditions, and investigate how the variability of Atlantic Meridional Ocean Circulation (AMOC) changes with κd. For the run with default κd, there is a sharp peak between period 20 to 30 years in the spectrum of AMOC variability. When κd is increased, we find that this peak generally becomes lower and lower, from ~45 Sv2/Hz when κd has the default value to only ~15 Sv2/Hz when κdis increased by 8 units.

 In the second set of experiments, we apply freshwater forcing (0.1 Sv) over the North Atlantic (50°N - 70°N) to each of the six runs obtained in the first set, and study how AMOC responds to the forcing. We find that the response of AMOC to freshwater forcing is significantly different for different κd. It weakens less and the response time becomes shorter when κd is increased. For example, AMOC weakens from ~24 Sv to ~18 Sv for the default κd, but weakens from ~35 Sv to ~32 Sv when κd is increased by 8 units. Our study therefore demonstrates that the AMOC stability and variability are strongly dependent on the value of κd, reducing the spurious component of this parameter in climate models is critical for simulating realistic climate variability, either internally or externally forced.