Yesterday, I had the opportunity to attend a colloquium seminar where Zhengyu Liu of University of Wisconsin-Madison gave a presentation on a couple of topics. One was on his recent paper which I discussed not too long ago concerning transient simulations of the the deglacial climate evolution. He also discussed the stability issues of the Thermohaline circulation and how intermediate models of AMOC tend to exhibit hysteresis behavior while fully-coupled AOGCM’s do not. I won’t touch on that but I’ll touch briefly on a few key points concerning the time period of roughly LGM to Bolling Allerod and some background. The regional and global-scale responses of atmospheric warming and their causes are explained better in my first post which is linked above. I also discuss the time period of LGM-BA in a bit more detail in case readers are unfamiliar with these events.:
— At present, the THC is characterized by a strong North Atlantic Deep Water (NADW) and a moderate Antarctic Bottom Water. In contrast, the LGM is characterized by dominant Antarctic bottom water and shallower and weaker NADW (See Shin et al 2003, GRL).
— A number of climate models gets the thermohaline depth at the LGM wrong (without cheating a little bit of course). Some models even show enhanced LGM NADW circulation, which is inconsistent with paleoclimatic data. One model exception is CCSM which was the model used by Liu et al. and discussed in the above link.
— At the LGM, sea ice extent in the Northern Hemisphere actually changes relatively little. Sea ice thickness changes significantly, perhaps doubling, but in Antarctica the Southern Ocean is characterized by deep convective mixing, which prevents thickening of sea-ice. Instead sea ice extent changes significantly in Antarctica at the LGM. This is true even today where Antarctic sea ice extent shows much more interannual and seasonal variability than the North (and also why global sea ice extent is not really a good proxy for the cyrosphere response to global warming).
— When ice is formed from sea water, salt is rejected from the crystal structure which results in the formation of brine and adds salt to the water underneath the ice and increasing the density. A significant fraction ( > 75%) of the increase in Antarctic Bottom Water at the LGM (relative to present) is explained by increased brine injection in the Southern Ocean. This makes the deep ocean saltier relative to present day. This denser Atlantic Bottom Water penetrates into the North Atlantic along the ocean bottom and shallows the North Atlantic THC
Liu et al 2009
The above also shows that the well-known “bipolar see-saw response” applies to SST’s while the subsurface ocean warms throughout the Atlantic. An increase in meltwater flux starting around 20kya induces a gradual decrease in the AMOC and freshens glacial bottom water. From 17 ka to the Bolling-Allerod, the meltwater flux decreases and thus AMOC shows a gradual recovery but resumes abruptly (see part D).
Liu et al 2009
— Dr. Liu proposes that the North Atlantic THC is controlled predominantly by the climate forcing over the Southern Ocean at the long glacial-interglacial cycle timescales, but by the North Atlantic climate forcing at the short timescales. This idea is best illustrated in his 2006 paper although the idea itself is older.
— For a similar forcing (say a globally mixed tracer like carbon dioxide), the initial response is to favor a much more pronounced Northern warming due to higher heat capacity and the buffering effect of the Southern Ocean. On longer timescales, the cooling becomes more important in the South because of the stronger ice-albedo feedback, which depends on extent and not thickness. This thin and vast sea ice is more sensitive to forcing than in the North and enhanced formation leads to enhanced brine input. Just for illustrative purposes, we can look at the temperature response to 2xCO2 in a slab ocean GCM compared to a model with a full-depth ocean GCM. Without an ocean GCM, the warming is nearly symmetric (except for Antarctica). Slab ocean models just have a shallow upper ocean, so the atmosphere and ocean come into equilibrium very quickly. Models with a deep ocean take thousands of years to come into equilibrium because heat is “sequestered” into the deep ocean. Presumably the two curves would look the same if we could run the full depth ocean GCM out for thousands of years. See Bitz et al 2006 for further details.
Cecilia Bitz, personal correspondence.
— CO2 is a primary forcing during the LGM and the leading causes of glacial THC anomalies. I also posted a while ago on the role of the Southern Ocean on atmospheric CO2 levels and the effects of the westerlies on upwelling, so that paper may be worth going over again.
— Under this view, the shallow/weak NADW is caused by strong Atlantic Bottom Water, caused by stronger sensitivity of sea ice in the SH to lower CO2. Faster responses (usually associated with abrupt climate changes) are dominated in the North, since it takes too much time for the Southern Ocean to change much for some perturbation. However, on much longer timescales, we have a proposed dominant Southern Ocean forcing of the North Atlantic THC.