methane retroaction (less CH 4 leads to more OH, hence less
CH 4 ) estimate emissions that explain almost entirely the low
CH 4 value at LGM (Quiquet et al. 2015).
Beyond the problem of the large changes in glacialinterglacial GHG concentrations, another issue was raised in
the 2000s when data became available for periods older
than *430,000 years BP. As shown on Fig. 23.5, older
interglacials before *430,000 ka BP (before the “Mid
Brunhes Event”) are characterized by a colder climate than
more recent interglacials, associated to lower GHG concentrations. The colder interglacial climate can be attributed
to different orbital configurations and lower CO 2 (Yin and
Berger 2010, 2012) but the reason for the lower GHG
concentrations still remains to be explained (Bouttes et al.
2018).
Abrupt Changes
At the centennial to millennial scale, climate variability is
superimposed onto the orbital-scale glacial-interglacial
cycles and is recorded by specific expressions at different
latitudes and in different climate archives (Clement and
Peterson 2008; see Volume 2, Chap. 29). Indeed, Greenland
ice cores unveiled a succession of events called
Dansgaard-Oeschger (D-O) events during the last glacial
period (Dansgaard et al. 1993; North Greenland Ice Core
Project Members 2004). Typically, a DO event is depicted as
an abrupt warming of 5–16 °C of the mean annual surface
temperature within a few decades toward a relatively mild
phase. This phase is then usually characterised by a gradual
cooling over several centuries and its end is marked by a
rapid cooling leading to a relatively stable cold phase persisting over several centuries or even up to a thousand years.
The signature of DO events is recorded in continental and
marine records in the Northern Hemisphere. In the Southern
Hemisphere, there are counterparts to these DO events. Ice
cores indicate more gradual (millennial-scale) warming in
Antarctica during the Greenland cold phases (EPICA
Community Members 2006; Barker et al. 2009; WAIS
Divide Project Members 2015). The antiphase relationship
between the two hemispheres is attributed to the thermal
bipolar seesaw, a mechanism whereby heat is redistributed
in the Atlantic Ocean (Stocker and Johnsen 2003).
In addition to the succession of Dansgaard-Oeschger
events, another prominent feature identified in marine sediments from the North Atlantic is the occurrence of the
Heinrich events. Heinrich events are identified by the presence of debris in sediments, and were first discovered by
Ruddiman in 1977. These debris are too big to be transported by oceanic currents, and in 1988 Heinrich proposed
that they could have been brought by icebergs which melted
above the zone where these ice rafted debris (IRD) were
found in sediments (Hemming 2004).
Antarctic ice core records show a rapid rise in atmospheric CO 2 of around 15 ppm over 2000–4000 years
(Fig. 23.9), generally synchronous with the millennial-scale
Antarctic warming, followed by a more gradual decrease
than the Antarctic temperature drop (Ahn and Brook 2008;
Bereiter et al. 2012). Measurements have shown that the
CO 2 rise was not steady, but punctuated by events with a
rapid increase (Ahn et al. 2012). Other data measurements
have resulted in several carbon sources being suggested to
explain these increases in atmospheric CO 2 , such as the
Southern Ocean (Gottschalk et al. 2016) or the North
Atlantic (Ezat et al. 2017).
These rapid changes of atmospheric CO 2 have been
studied only relatively recently as previously, the temporal
resolution in the records was not sufficient. Temperature
changes have been studied for longer since more high resolution data were available. To replicate these rapid climate
changes in simulations, modellers have found that artificially
adding freshwater to the North Atlantic, for example theoretically due to the melting of numerous icebergs, could slow
down or even stop the Atlantic meridional overturning circulation. This then generally leads to warming in the North
Hemisphere and cooling in the South Hemisphere, in line
with changes observed in the data (see Volume 2, Chap. 29).
More recently, the impact of such hosing experiments on
the carbon cycle and atmospheric CO 2 evolution has also
been tested in carbon-climate models, to evaluate the role of
the terrestrial biosphere and the ocean, in particular, in
coupled ocean-atmosphere-terrestrial biosphere models. The
model response to the freshwater input appears to be very
dependent on the type, duration and amplitude of the
freshwater input, on the background climate (glacial vs
pre-industrial) and the model. For example, the LOVECLIM
model simulates a 15 ppmv increase in the context of a
pre-industrial climate, but a 10 ppm CO 2 decrease in a
glacial climate in response to a decrease in the AMOC driven by the same freshwater input (Menviel et al. 2008b). In
both cases, the ocean takes up more carbon and the terrestrial
biosphere loses carbon, but the balance between the two
outcomes results in opposite effects on the atmospheric CO 2 .
This balance also depends on the different time reactions of
the carbon reservoirs: vegetation reacts more rapidly than the
ocean.
In general, most models simulate an overall increase in
atmospheric CO 2 ranging from a few ppm up to more than
20 ppm, depending on the model and the size of the freshwater flux (Obata 2007; Schmittner and Galbraith 2008;
Menviel et al. 2008b; Bozbiyik et al. 2011; Bouttes et al.
2012; Matsumoto and Yokoyama 2013), but the causes are
different: in some models the ocean gains carbon and the
282
N. Bouttes et al.
CH 4 ) estimate emissions that explain almost entirely the low
CH 4 value at LGM (Quiquet et al. 2015).
Beyond the problem of the large changes in glacialinterglacial GHG concentrations, another issue was raised in
the 2000s when data became available for periods older
than *430,000 years BP. As shown on Fig. 23.5, older
interglacials before *430,000 ka BP (before the “Mid
Brunhes Event”) are characterized by a colder climate than
more recent interglacials, associated to lower GHG concentrations. The colder interglacial climate can be attributed
to different orbital configurations and lower CO 2 (Yin and
Berger 2010, 2012) but the reason for the lower GHG
concentrations still remains to be explained (Bouttes et al.
2018).
Abrupt Changes
At the centennial to millennial scale, climate variability is
superimposed onto the orbital-scale glacial-interglacial
cycles and is recorded by specific expressions at different
latitudes and in different climate archives (Clement and
Peterson 2008; see Volume 2, Chap. 29). Indeed, Greenland
ice cores unveiled a succession of events called
Dansgaard-Oeschger (D-O) events during the last glacial
period (Dansgaard et al. 1993; North Greenland Ice Core
Project Members 2004). Typically, a DO event is depicted as
an abrupt warming of 5–16 °C of the mean annual surface
temperature within a few decades toward a relatively mild
phase. This phase is then usually characterised by a gradual
cooling over several centuries and its end is marked by a
rapid cooling leading to a relatively stable cold phase persisting over several centuries or even up to a thousand years.
The signature of DO events is recorded in continental and
marine records in the Northern Hemisphere. In the Southern
Hemisphere, there are counterparts to these DO events. Ice
cores indicate more gradual (millennial-scale) warming in
Antarctica during the Greenland cold phases (EPICA
Community Members 2006; Barker et al. 2009; WAIS
Divide Project Members 2015). The antiphase relationship
between the two hemispheres is attributed to the thermal
bipolar seesaw, a mechanism whereby heat is redistributed
in the Atlantic Ocean (Stocker and Johnsen 2003).
In addition to the succession of Dansgaard-Oeschger
events, another prominent feature identified in marine sediments from the North Atlantic is the occurrence of the
Heinrich events. Heinrich events are identified by the presence of debris in sediments, and were first discovered by
Ruddiman in 1977. These debris are too big to be transported by oceanic currents, and in 1988 Heinrich proposed
that they could have been brought by icebergs which melted
above the zone where these ice rafted debris (IRD) were
found in sediments (Hemming 2004).
Antarctic ice core records show a rapid rise in atmospheric CO 2 of around 15 ppm over 2000–4000 years
(Fig. 23.9), generally synchronous with the millennial-scale
Antarctic warming, followed by a more gradual decrease
than the Antarctic temperature drop (Ahn and Brook 2008;
Bereiter et al. 2012). Measurements have shown that the
CO 2 rise was not steady, but punctuated by events with a
rapid increase (Ahn et al. 2012). Other data measurements
have resulted in several carbon sources being suggested to
explain these increases in atmospheric CO 2 , such as the
Southern Ocean (Gottschalk et al. 2016) or the North
Atlantic (Ezat et al. 2017).
These rapid changes of atmospheric CO 2 have been
studied only relatively recently as previously, the temporal
resolution in the records was not sufficient. Temperature
changes have been studied for longer since more high resolution data were available. To replicate these rapid climate
changes in simulations, modellers have found that artificially
adding freshwater to the North Atlantic, for example theoretically due to the melting of numerous icebergs, could slow
down or even stop the Atlantic meridional overturning circulation. This then generally leads to warming in the North
Hemisphere and cooling in the South Hemisphere, in line
with changes observed in the data (see Volume 2, Chap. 29).
More recently, the impact of such hosing experiments on
the carbon cycle and atmospheric CO 2 evolution has also
been tested in carbon-climate models, to evaluate the role of
the terrestrial biosphere and the ocean, in particular, in
coupled ocean-atmosphere-terrestrial biosphere models. The
model response to the freshwater input appears to be very
dependent on the type, duration and amplitude of the
freshwater input, on the background climate (glacial vs
pre-industrial) and the model. For example, the LOVECLIM
model simulates a 15 ppmv increase in the context of a
pre-industrial climate, but a 10 ppm CO 2 decrease in a
glacial climate in response to a decrease in the AMOC driven by the same freshwater input (Menviel et al. 2008b). In
both cases, the ocean takes up more carbon and the terrestrial
biosphere loses carbon, but the balance between the two
outcomes results in opposite effects on the atmospheric CO 2 .
This balance also depends on the different time reactions of
the carbon reservoirs: vegetation reacts more rapidly than the
ocean.
In general, most models simulate an overall increase in
atmospheric CO 2 ranging from a few ppm up to more than
20 ppm, depending on the model and the size of the freshwater flux (Obata 2007; Schmittner and Galbraith 2008;
Menviel et al. 2008b; Bozbiyik et al. 2011; Bouttes et al.
2012; Matsumoto and Yokoyama 2013), but the causes are
different: in some models the ocean gains carbon and the
282
N. Bouttes et al.