Global Synthesis of Sea-Surface Temperature Trends During Marine Isotope Stage 11 - Integrated Analysis of Interglacial Climate Dynamics (INTERDYNAMIC) - page 15

first EOF follows a glacial-interglacial pattern with cold SSTs during MIS
12 and MIS 10, and a relatively long duration of warmer SSTs between 416 and
405 ka BP. This trend is similar to the record of Antarctic temperature during MIS
11 (Jouzel et al. 2007 ), although the interglacial temperature peak is leading the
Antarctic record with an offset of

The

4 ka (Fig. 1 b).This would indicate that during
MIS 11, the temperature over Antarctica was not closely coupled to the global mean
SST and might have re

*

ected an antiphased southern hemisphere insolation pattern
(Laepple et al. 2011 ). It further seems that the deglacial SST rise, indicated by
EOF1 preceded the reduction of the global ice volume (Elder

eld et al. 2012 )by

5 ka suggesting a faster reaction of the surface ocean to insolation and green-
house gas forcing than the response of the slowly melting ice sheets.
The second EOF

*

s scores indicate a later establishment of a relative SST
maximum and a longer-lasting period of warmer temperatures during late MIS 11
(Fig. 1 c). This regional trend is particularly re

'

ected in the SST records of the mid-
latitude North Atlantic Ocean and Mediterranean Sea. The apparently later onset of
the MIS 11 optimum and the longer duration of interglacial warmth have also been
observed by Voelker et al. ( 2010 ), who hypothesized that the associated sustained
meltwater input to the (sub-)polar regions may have resulted in a weaker Altantic
Meridional Overturning circulation (AMOC). Similarly, mean

13 C of benthic
foraminifera from deeper waters in the North Atlantic used as a proxy for North
Atlantic Deepwater (NADW) production (Lisiecki et al. 2008 ) show a trend of
increasing NADW production between 410 and 400 ka BP which is quite similar to
the EOF2 scores (Fig. 1 c).
The comparison of the proxy-based SST anomalies with CCSM3 model results
revealed a large difference in their variance. The range of proxy-based SST
anomalies is

δ

4 ° C, whereas modeled SST anomalies vary rarely by more than

*

1 C (Milker et al. 2013 ) (Fig. 2 a). The much lower variance in modeled temper-

ature trends might result from an underestimation of temperature changes in climate
models, an overestimated proxy SST variability, or from a combination of both.
Underestimation of climate variability in model simulations may be caused by
shortcomings in the model physics and/or missing climate components resulting in
a lack of potentially important feedback mechanisms. Higher variance in the proxy
data may result from noise and calibration uncertainties, from larger shifts in the
ecology of the microfossils or changes in seasonality and vertical habitats. Despite
the large differences in variance and considering all the potential sources of
uncertainty in the proxy-based SST values, it is remarkable that in several cases a
visual agreement between the direction of proxy and model SST changes emerged
(Fig. 2 b). Moreover for the boreal summer season of the 416 ka BP time slice as
well as for the boreal winter season of the 405 ka BP time slice, a statistically
signi

°

cant correlation between the proxy-based and modeled SST anomalies was
found (Milker et al. 2013 ).This indicates that orbital forcing, the major driver in the
CCSM3 experiments, has left a detectable signature in the global SST pattern
during MIS 11, despite its unusually low amplitude.