Geoscience Reference
In-Depth Information
Figure 15. Schematic illustration of the proposed positive feedback process controlling reinitiation of NADW formation in
Labrador Sea and GIN seas. Meridional pressure difference is indicated by δ P.
Atlantic from meltwater discharge, the whole North Atlantic
Basin is nearly homogeneous for salinity, so the salinity
gradient at the south boundary of NADW regions is impor-
tant for salt transport. During the AMOC recovery period,
the maximum meridional salinity gradient within the upper
layers of the North Atlantic moves northward.
On the basis of our reconsidered mechanism of NADW
reinitiation, only after NADW formation is fully reinitiated in
the Labrador Sea, does the dynamical environment of the GIN
seas result in full convective reinitiation. So the combination
of local and nonlocal mechanisms causes NADW formation
reinitiation in the Labrador Sea and GIN seas to be asynchro-
nous, and the sequence of reinitiation is northward.
tion in the two regions of origin with a reconsidered mech-
anism. We also suggest that the two-stage feature of AMOC
recovery should not just depend on the evolutionary char-
acteristics of a time series of AMOC intensity because a
single time series is not sufficient to present the reinitiation
process of NADW formation in multiple origins, synchro-
nous or otherwise. Comparing the results of Renold et al.
[2009], our results provide more physical and geographical
analysis with the same model CCSM3 by providing a clearer
and more complete mechanism, and these results are doubly
interesting because they are based on a particular historical
event of abrupt climate change.
Our results, in addition to previous work such as that of
Vellinga and Wood [2002] and Renold et al. [2009], point out
that the Labrador Sea and GIN seas both provide a compa-
rable robust contribution to the AMOC recovery. However,
some recent work also shows that the role of the multiple
origins of NADW formation in AMOC variability and sta-
bility, as derived from different models [Bentsen et al., 2004;
Mignot and Frankignoul, 2005; Latif et al., 2006; Spence et
al., 2008], is uncertain. Despite the dependence of model,
5. DISCUSSIONS AND CONCLUSIONS
In this analysis, we found that the simulated AMOC re-
covery during the last deglaciation occurred in two stages:
first, in the Labrador Sea and then in the GIN seas. We
described how the two-stage feature of AMOC recovery
depends on the asynchronous reinitiation of NADW forma-
Search WWH ::




Custom Search