Geoscience Reference
In-Depth Information
Labrador Sea or the Greenland-Iceland-Norwegian (GIN)
seas. Despite the dependence of model formulation and
internal variability of the AMOC
asynchronous or not. A detailed investigation of the AMOC
recovery process should take a broader view of the geograph-
ical heterogeneity of the NADW formation in the North
Atlantic and the regional contributions to the total AMOC
intensity.
Another de
fluctuation under control
runs, the AMOC changes in perturbation experiments (water-
hosing experiments) are mainly associated with the con-
vection changes in the Labrador Sea and GIN seas with
asynchronous and different contributions [Krebs and Tim-
mermann, 2007a, 2007b; Mignot et al., 2007; Vellinga and
Wood, 2002]. This mechanism of AMOC variability is
found in the widely used Community Climate System
Model version 3 (CCSM3) under control and water-hosing
experiments in glacial, preindustrial, and present-day cli-
mate conditions [Hu et al., 2007; Renold et al., 2009]. So
the details of NADW reinitiation in multiple areas of origin
are important for the understanding of the AMOC recovery
process. Other works have touched upon the asynchronicity
of the reinitiation of deepwater formation at different loca-
tions, but these studies provide divergent conclusions.
Starting with an arti
ciency in studies of the AMOC recovery is
that idealized water-hosing experiments are not typically run
for long enough to study the entire recovery process. Previ-
ous modeling studies usually emphasized the prerecovery
period of the simulation, ending when the AMOC intensity
resumes its unperturbed level for the first time and neglecting
whether the AMOC system has fully recovered or not [Vel-
linga and Wood, 2002; Bitz et al., 2007; Hu et al., 2008;
Renold et al., 2009]. Actually, numerous experiments with
different models have shown that the AMOC will keep
increasing after its initial recovery and continue to increase
in intensity before returning to its unperturbed level, exhibit-
ing an
phenomenon [Manabe and Stouffer,
1997; Knutti et al., 2004; Stouffer et al., 2006; Mignot et al.,
2007; Krebs and Timmermann, 2007a, 2007b; Schmittner
and Galbraith, 2008; Arzel et al., 2008; Liu et al., 2009].
The historic occurrence of an AMOC overshoot during the
BA event has been validated with combined observational
and model evidence. That this is the effect of AMOC over-
shoot in deep currents of the North Atlantic regionally is
shown in model simulations, and this feature is consistent
with the distribution of reconstructed proxies such as cores
GGC5 and TTR-451 (J. Cheng et al., Model-proxy compar-
ison for overshoot phenomenon of Atlantic thermohaline
circulation at Bølling-Allerød, unpublished manuscript,
2011). So the postrecovery stage, or overshoot stage, of the
AMOC system should not be ignored.
The mechanisms for AMOC recovery are still not well
understood. With an intermediate complexity model
(CLIMBER-3
α
), Wright and Stocker [1991] and Mignot et
al. [2007] thought the recovery of AMOC derived from the
destabilized strati
“
overshoot
”
state of AMOC in
model HadCM3 under modern climate conditions, Vellinga
and Wood [2002] found that NADW formation reinitiates
cially
“
turned off
”
first in the GIN seas and then moves southward to the
Labrador Sea. This work also proposed that the northward
salt transport determines the order of the NADW reinitia-
tion. In contrast, Renold et al. [2009] found that the re-
initiation of the NADW first occurred in the Labrador Sea
and then in the GIN seas, based on a series of water-hosing
experiments with CCSM3. Renold et al. proposed that the
northward salt and heat transport determines the northward
reinitiation of NADW beneath the retreat of sea ice. The
contradictory results of Vellinga and Wood [2002] and
Renold et al. [2009] rely upon the attendance of sea ice
change during the AMOC recovery period and maybe the
dependence of the models. However, the results of Renold
et al. [2009] seem to simulate the real world more closely
because of the more realistic experiment scheme.
Renold et al. [2009] also found two stages in the recovery
of AMOC, as suggested by the evolutionary characteristics
of the AMOC intensity. However, in many studies using
different models with different complexities, a two-stage
mechanism of AMOC recovery is not always evident in a
single time series of AMOC intensity [Manabe and Stouffer,
1997; Knutti et al., 2004; Mignot et al., 2007; Arzel et al.,
2008]. The two-stage feature of AMOC recovery seems to
rely on the asynchronicity of NADW reinitiation in multiple
origins according to Renold et al. [2009]. The asynchronous
feature of the NADW reinitiation in multiple origins could
induce the intensity series to present the two-stage feature but
may not always do so. This suggests that a single measure of
AMOC intensity maybe not be a good choice for judging
whether the reinitiation of NADW in multiple origins is
cation with warmed abyssal water in the
North Atlantic. Using the intermediate complexity model
ECBILT-CLIO, Goosse et al. [2002] proposed that the re-
covery of AMOC comes from a stochastic process of air-sea
interaction. With a coupled general circulation model (GCM)
such as LSG-ECHAM3/T42, Knorr and Lohmann [2003]
proposed that a weakened Antarctic Bottom Water cell con-
tributes to the recovery of AMOC. With different complexity
models, Vellinga and Wood [2002] (HadCM3 and coupled
GCM), Yin et al. [2006] (UIUC and coupled GCM), and
Krebs and Timmermann [2007a, 2007b] (intermediate com-
plexity model and ECBILT-CLIO) hypothesized that the
northward advection of the salinity anomaly is the key factor
for AMOC recovery, and this hypothesis is supported by
paleorecords [Carlson et al., 2008].
Search WWH ::
Custom Search