lamination (Alefs et al. 1996; Czymzik et al. 2013) and the
time represented by a sediment sample 1 cm thick was on
average 2.5 years. A first, direct comparison of the
youngest 200 years of the d
18 O L with the adequately
averaged mean annual air temperature (MAAT) record from
Hohenpeissenberg (within the catchment of the lake) gave a
reasonable correlation (r
2 = 0.91), but an apparent d
18 O L /
MAAT sensitivity, which, with 0.38‰/°C, was significantly
below the d
18 O P /MAAT sensitivity derived from three
decades of observation in the region and for the rest of
Europe (0.58‰/°C). In addition, all d
18 O L values younger
than 1920, while still showing the same temperaturedependence, were systematically displaced by −0.15‰ with
respect to the older part of the record. The smallerthan-expected amplitude of d
18 O L can clearly be attributed
to oversampling of the record compared to the reactivity of
the entire retention system and, in a next step, can be used to
better quantify the catchment residence time. The abrupt
relative shift of d
18 O L is an indication of a change of the
d
18 O P -d
18 O L difference or of the catchment effect in its
wider sense, related to the regulation of the Ammer River,
which started in 1920 and ended in 1922. This regulation
was designed to accelerate the runoff of storm-related flood
events (which occur primarily in summer) and to allow
cultivation of the extended river plains. Thus, after the regulation, a larger portion of isotopically-enriched summer
water was transferred directly into the lake, was mixed into
the epilimnion, and became partially lost via the outlet,
whereas before regulation, those flood waters could infiltrate
into the river plain aquifers and more efficiently change the
longer term d
18 O I and, in consequence, the d
18 O L .
After correction for this ‘summer bypass effect’, the entire
d
18 O L record was compared to the d
18 O L calculated using a
very basic lake model, by assuming a linear relation between
d
18 O P and the measured air temperature and by taking into
account changes in the seasonal distribution of precipitation
and the most obvious features of lake water mixing. Figure 15.4 shows the results of this modelling approach, which
gives the best fit between calculated and observed d
18 O L , if
the mean drainage basin retention time is set to 3 years and
if the temperature dependence of d
18 O P is assumed to be
0.58‰/°C, thus confirming the assumption that the correlation based on direct observation during the last decades
was valid for the last two centuries.
The Record of d
18
O P in Central Europe Over
the Past 15,000 Years
The longer term history of d
18 O P reconstructed from
Ammersee deep-lake cores has been discussed in two publications, the first of which (von Grafenstein et al. 1998)
concentrates on the synchrony of a cold event around
8200 years BP in Europe and Greenland and its probable
forcing by the collapse of the Hudson Bay ice dome. The
second (von Grafenstein et al. 1999a, b) makes a comparison
between d
18 O P in Europe and in Greenland over the period
between 15,000 and 5500 years B.P. The most striking
feature is probably the great similarity in the records, even at
high frequencies, providing evidence that the climate of both
regions (and most likely of the entire North Atlantic
perimeter) experienced the same decadal variations, governed by the variability of the heat flux from the North
Atlantic Ocean. Independent confirmation of the details of
the d
18 O from ice in Central Greenland in a European record
significantly increases confidence in the quantitative interpretation of changes of the regional air temperatures and
helps to exclude alternative explanations such as significant
changes in the prevailing water vapor sources for precipitation in both regions.
Like the 8.2-ka-event, the short, abrupt, cold oscillations
during the relatively warm periods of the late Glacial and
early Holocene were probably forced by cataclysmic
fresh-water discharges into the Atlantic Ocean as a consequence of the disintegration of the continental ice sheets.
Despite the high correlation at high frequencies, the quantitative comparison of both d
18 O P records reveals periods of
systematic change in the differences between Europe and
Central Greenland around the Younger Dryas cold period.
These might indicate slow systematic changes in the surface
conditions and circulation in the Greenland-Norwegian Sea,
governed by the persisting meltwater flux from the Scandinavian ice sheet during the relatively warm periods. The
proposed mechanism (von Grafenstein et al. 1998) is also an
attractive hypothesis for ‘Dansgaard-Oeschger-events’,
occurring frequently during the period from *80,000 to
*25,000 yr B.P.
Figure 15.5 shows an overview of the state of d
18 O P
reconstruction from Ammersee deep-lake ostracods. The
lack of resolution for the last 5500 years is evident, even if
the available data already show some similarity with the
corresponding part of the d
18 O P records from Greenland.
The highest resolved parts of the record provide probably the
best constrained and resolved d
18 O P record existing in
Europe. However, there is still uncertainty ranging between
+0.6‰ and −0.3‰ from potential hydrological changes that
have not been considered. In addition, the record might not
be representative of the entire continent, at least during the
late Glacial, when stronger and highly dynamic longitudinal
climate gradients were very likely to have prevailed. Both
the errors from hydrological change and those from regional
European climatic gradients may partially explain the
apparent cross North Atlantic differences. Another problem
of the existing record from Ammersee is the quality of the
age model which is based on the few radiocarbon measurements of macro plant and insect remains found in the
188
U. von Grafenstein and I. Labuhn
time represented by a sediment sample 1 cm thick was on
average 2.5 years. A first, direct comparison of the
youngest 200 years of the d
18 O L with the adequately
averaged mean annual air temperature (MAAT) record from
Hohenpeissenberg (within the catchment of the lake) gave a
reasonable correlation (r
2 = 0.91), but an apparent d
18 O L /
MAAT sensitivity, which, with 0.38‰/°C, was significantly
below the d
18 O P /MAAT sensitivity derived from three
decades of observation in the region and for the rest of
Europe (0.58‰/°C). In addition, all d
18 O L values younger
than 1920, while still showing the same temperaturedependence, were systematically displaced by −0.15‰ with
respect to the older part of the record. The smallerthan-expected amplitude of d
18 O L can clearly be attributed
to oversampling of the record compared to the reactivity of
the entire retention system and, in a next step, can be used to
better quantify the catchment residence time. The abrupt
relative shift of d
18 O L is an indication of a change of the
d
18 O P -d
18 O L difference or of the catchment effect in its
wider sense, related to the regulation of the Ammer River,
which started in 1920 and ended in 1922. This regulation
was designed to accelerate the runoff of storm-related flood
events (which occur primarily in summer) and to allow
cultivation of the extended river plains. Thus, after the regulation, a larger portion of isotopically-enriched summer
water was transferred directly into the lake, was mixed into
the epilimnion, and became partially lost via the outlet,
whereas before regulation, those flood waters could infiltrate
into the river plain aquifers and more efficiently change the
longer term d
18 O I and, in consequence, the d
18 O L .
After correction for this ‘summer bypass effect’, the entire
d
18 O L record was compared to the d
18 O L calculated using a
very basic lake model, by assuming a linear relation between
d
18 O P and the measured air temperature and by taking into
account changes in the seasonal distribution of precipitation
and the most obvious features of lake water mixing. Figure 15.4 shows the results of this modelling approach, which
gives the best fit between calculated and observed d
18 O L , if
the mean drainage basin retention time is set to 3 years and
if the temperature dependence of d
18 O P is assumed to be
0.58‰/°C, thus confirming the assumption that the correlation based on direct observation during the last decades
was valid for the last two centuries.
The Record of d
18
O P in Central Europe Over
the Past 15,000 Years
The longer term history of d
18 O P reconstructed from
Ammersee deep-lake cores has been discussed in two publications, the first of which (von Grafenstein et al. 1998)
concentrates on the synchrony of a cold event around
8200 years BP in Europe and Greenland and its probable
forcing by the collapse of the Hudson Bay ice dome. The
second (von Grafenstein et al. 1999a, b) makes a comparison
between d
18 O P in Europe and in Greenland over the period
between 15,000 and 5500 years B.P. The most striking
feature is probably the great similarity in the records, even at
high frequencies, providing evidence that the climate of both
regions (and most likely of the entire North Atlantic
perimeter) experienced the same decadal variations, governed by the variability of the heat flux from the North
Atlantic Ocean. Independent confirmation of the details of
the d
18 O from ice in Central Greenland in a European record
significantly increases confidence in the quantitative interpretation of changes of the regional air temperatures and
helps to exclude alternative explanations such as significant
changes in the prevailing water vapor sources for precipitation in both regions.
Like the 8.2-ka-event, the short, abrupt, cold oscillations
during the relatively warm periods of the late Glacial and
early Holocene were probably forced by cataclysmic
fresh-water discharges into the Atlantic Ocean as a consequence of the disintegration of the continental ice sheets.
Despite the high correlation at high frequencies, the quantitative comparison of both d
18 O P records reveals periods of
systematic change in the differences between Europe and
Central Greenland around the Younger Dryas cold period.
These might indicate slow systematic changes in the surface
conditions and circulation in the Greenland-Norwegian Sea,
governed by the persisting meltwater flux from the Scandinavian ice sheet during the relatively warm periods. The
proposed mechanism (von Grafenstein et al. 1998) is also an
attractive hypothesis for ‘Dansgaard-Oeschger-events’,
occurring frequently during the period from *80,000 to
*25,000 yr B.P.
Figure 15.5 shows an overview of the state of d
18 O P
reconstruction from Ammersee deep-lake ostracods. The
lack of resolution for the last 5500 years is evident, even if
the available data already show some similarity with the
corresponding part of the d
18 O P records from Greenland.
The highest resolved parts of the record provide probably the
best constrained and resolved d
18 O P record existing in
Europe. However, there is still uncertainty ranging between
+0.6‰ and −0.3‰ from potential hydrological changes that
have not been considered. In addition, the record might not
be representative of the entire continent, at least during the
late Glacial, when stronger and highly dynamic longitudinal
climate gradients were very likely to have prevailed. Both
the errors from hydrological change and those from regional
European climatic gradients may partially explain the
apparent cross North Atlantic differences. Another problem
of the existing record from Ammersee is the quality of the
age model which is based on the few radiocarbon measurements of macro plant and insect remains found in the
188
U. von Grafenstein and I. Labuhn