Evaluation of the Freshwater Influx Associated with a
Heinrich Event: Example of a Joint Model-Data
Approach
The thickness and location of the detritic levels found in the
North Atlantic open the question of the volume of icebergs
released in this region, especially as the records of sea level
change during the deglaciation do not show the Heinrich 1
event having any impact. This would lead us to assume that
the volume of icebergs is low (a maximum of 3–5 m of sea
level). However, estimates achieved using other methods
(Roche et al. 2004) have quite different results. Yet, in order
to reproduce the evolution of these events in climate models,
we must know the influx of freshwater represented by the
corresponding armada of icebergs. It is therefore necessary
to determine not only the volume of icebergs released into
the ocean, but also the timeframe over which they were
released. The classic
14 C dating method presents a major
problem related to the reorganization of the ocean. Indeed,
the postulated shutdown of the thermohaline circulation is
causing a change in the distribution of carbon within the
ocean reservoir, the largest in the Earth system. Basically,
the stopping of the thermohaline circulation would make the
surface waters appear younger, and the bottom waters appear
older during the event, and it would make the surface waters
appear older, and the bottom waters younger after the event,
making it difficult to evaluate the duration of the event.
As the results from conventional methods assessing the
duration and volume of icebergs emitted during these events
have a very high degree of uncertainty, a more detailed
estimation using different methods is necessary. An
approach combining model and data was undertaken by
Roche et al. (2004), based on the simulation of d
18 O of water
in the ocean. In fact, there is a large number of d
18 O records
of foraminiferal calcite from marine sedimentary cores from
the North Atlantic, which constrain tightly the evolution of
this indicator during the Heinrich 1 and 4 events. The basis
of this new evaluation method is to consider that the geographical distribution of the d
18 O anomaly recorded in
planktonic foraminifera provides information about the
duration and volume of icebergs emitted. If the thermohaline
circulation slows down, the slight anomaly in d
18
O created
by the melting of the armadas of icebergs will tend to remain
longer at the surface (and vice versa). The maximum
recorded anomaly should also be related to the maximum of
the iceberg flux. To assess this relationship, Roche et al.
(2004) performed a large set of simulations by varying the
duration and influx of additional freshwater for all of the
feasible values (determined from the data). Then, the simulated d
18 O anomaly was compared with the anomaly measured in the marine sediment cores. The simulations that best
represent the distribution described by the data were selected
to provide a new estimate of the volume of icebergs discharged and the duration of these events: for the Heinrich 4
event (about 45 ka BP) the most likely duration is
300 ± 100 years and the volume of icebergs discharged is
equivalent to about 3 m of sea level. These results were later
confirmed by a method based on a sediment/iceberg model
(Roberts et al. 2014).
Interpretation of the Isotopic Signal Measured in the
North Atlantic
We have seen that the d
18 O records from Greenland cores
and from ocean sediment cores appear to have very similar
signals during periods of rapid climate variability. These
similarities should not lead to the conclusion that both signals have the same climatic cause. Indeed, d
18 O records
from foraminiferal calcite are complex, sensitive to both
temperature and hydrological changes. Hydrological changes have a particularly strong impact. To separate out the
influences of the different causes of d
18 O variations in calcite, one solution is to simulate this indicator within a climate model to analyze the importance of the various
processes at work. Roche and Paillard (2005) performed this
type of simulation for the series of Dansgaard-Oeschger
events surrounding the Heinrich 4 event (about 40 ka BP).
The result is shown in Fig. 29.4. The first outcome is the
accuracy of the model in reproducing the variations measured in the ocean record. The advantage of using a climate
model is that not only the d
18 O of calcite can be simulated
but also temperature and salinity, prognostic variables for the
climate model. The ‘temperature’ and ‘hydrological cycle’
components can then be extracted from the d
18 O signal of
the calcite. The surprising result from Fig. 29.4 is that, in the
model, the strongest changes in both temperature and d
18 O
of the water over the entire simulated period are shown to be
during the Heinrich 4 event. In the d
18 O record in calcite, the
greatest variation is associated with the Dansgaard-Oeschger
events. The reason for this difference is that since the variations in temperature and in the d
18 O of water have, from the
point of view of the calcite d
18 O, opposing signatures, their
effects partially cancel each other out in the d
18 O of the
calcite, hiding the strongest signal in the marine sedimentary
data. This example shows how difficult it is to interpret
isotopic paleoclimate indicators in terms of climate and how
an integrated data-model approach can provide a better
understanding of climate dynamics in the past.
Transmission of the Signal to the Continents
The pollen records extracted from the marine cores provide
information on the evolution of the vegetation contemporaneous with the evolution of the oceanic conditions. The two
412
M. Kageyama et al.
Heinrich Event: Example of a Joint Model-Data
Approach
The thickness and location of the detritic levels found in the
North Atlantic open the question of the volume of icebergs
released in this region, especially as the records of sea level
change during the deglaciation do not show the Heinrich 1
event having any impact. This would lead us to assume that
the volume of icebergs is low (a maximum of 3–5 m of sea
level). However, estimates achieved using other methods
(Roche et al. 2004) have quite different results. Yet, in order
to reproduce the evolution of these events in climate models,
we must know the influx of freshwater represented by the
corresponding armada of icebergs. It is therefore necessary
to determine not only the volume of icebergs released into
the ocean, but also the timeframe over which they were
released. The classic
14 C dating method presents a major
problem related to the reorganization of the ocean. Indeed,
the postulated shutdown of the thermohaline circulation is
causing a change in the distribution of carbon within the
ocean reservoir, the largest in the Earth system. Basically,
the stopping of the thermohaline circulation would make the
surface waters appear younger, and the bottom waters appear
older during the event, and it would make the surface waters
appear older, and the bottom waters younger after the event,
making it difficult to evaluate the duration of the event.
As the results from conventional methods assessing the
duration and volume of icebergs emitted during these events
have a very high degree of uncertainty, a more detailed
estimation using different methods is necessary. An
approach combining model and data was undertaken by
Roche et al. (2004), based on the simulation of d
18 O of water
in the ocean. In fact, there is a large number of d
18 O records
of foraminiferal calcite from marine sedimentary cores from
the North Atlantic, which constrain tightly the evolution of
this indicator during the Heinrich 1 and 4 events. The basis
of this new evaluation method is to consider that the geographical distribution of the d
18 O anomaly recorded in
planktonic foraminifera provides information about the
duration and volume of icebergs emitted. If the thermohaline
circulation slows down, the slight anomaly in d
18
O created
by the melting of the armadas of icebergs will tend to remain
longer at the surface (and vice versa). The maximum
recorded anomaly should also be related to the maximum of
the iceberg flux. To assess this relationship, Roche et al.
(2004) performed a large set of simulations by varying the
duration and influx of additional freshwater for all of the
feasible values (determined from the data). Then, the simulated d
18 O anomaly was compared with the anomaly measured in the marine sediment cores. The simulations that best
represent the distribution described by the data were selected
to provide a new estimate of the volume of icebergs discharged and the duration of these events: for the Heinrich 4
event (about 45 ka BP) the most likely duration is
300 ± 100 years and the volume of icebergs discharged is
equivalent to about 3 m of sea level. These results were later
confirmed by a method based on a sediment/iceberg model
(Roberts et al. 2014).
Interpretation of the Isotopic Signal Measured in the
North Atlantic
We have seen that the d
18 O records from Greenland cores
and from ocean sediment cores appear to have very similar
signals during periods of rapid climate variability. These
similarities should not lead to the conclusion that both signals have the same climatic cause. Indeed, d
18 O records
from foraminiferal calcite are complex, sensitive to both
temperature and hydrological changes. Hydrological changes have a particularly strong impact. To separate out the
influences of the different causes of d
18 O variations in calcite, one solution is to simulate this indicator within a climate model to analyze the importance of the various
processes at work. Roche and Paillard (2005) performed this
type of simulation for the series of Dansgaard-Oeschger
events surrounding the Heinrich 4 event (about 40 ka BP).
The result is shown in Fig. 29.4. The first outcome is the
accuracy of the model in reproducing the variations measured in the ocean record. The advantage of using a climate
model is that not only the d
18 O of calcite can be simulated
but also temperature and salinity, prognostic variables for the
climate model. The ‘temperature’ and ‘hydrological cycle’
components can then be extracted from the d
18 O signal of
the calcite. The surprising result from Fig. 29.4 is that, in the
model, the strongest changes in both temperature and d
18 O
of the water over the entire simulated period are shown to be
during the Heinrich 4 event. In the d
18 O record in calcite, the
greatest variation is associated with the Dansgaard-Oeschger
events. The reason for this difference is that since the variations in temperature and in the d
18 O of water have, from the
point of view of the calcite d
18 O, opposing signatures, their
effects partially cancel each other out in the d
18 O of the
calcite, hiding the strongest signal in the marine sedimentary
data. This example shows how difficult it is to interpret
isotopic paleoclimate indicators in terms of climate and how
an integrated data-model approach can provide a better
understanding of climate dynamics in the past.
Transmission of the Signal to the Continents
The pollen records extracted from the marine cores provide
information on the evolution of the vegetation contemporaneous with the evolution of the oceanic conditions. The two
412
M. Kageyama et al.