into a temperature curve (in the case of polar and tropical
glaciers). This approach is also possible with data from
historical documents, which, when they do not provide
direct climatic information, recount events related to a climate variable (floods, droughts, freezing) that may be standardized semi-quantitatively (Pfister 1980).
Another approach is required for analysis of the living
world. The climate signal recorded by microorganisms is
complex and is a reaction to a combination of several climatic variables; temperature, salinity, and nutrients in the
ocean; temperature and precipitation on land. It is therefore
not possible to decode this signal with a simple equation and
a multivariate approach is needed to interpret the changes in
sets of pollen and diatoms in a continental environment, and
of diatoms and foraminifera in a marine setting (Chap. 21).
On a smaller timescale, the thickness or density of a tree
ring, also influenced by a complex environment, can seldom
be interpreted with a simple calibration equation. Several
series from the same region need to be used to get a clear
climate signal.
Uncertainties on the Temporal Scale
Before discussing the uncertainties and limitations associated with the interpretation of series of proxies, we will
review the uncertainties related to their temporal resolution.
Figure 10.1 shows the temporal characteristics of climate
forcings (internal and external) and the different types of
supports of commonly studied proxies. Many forcings and
components of the climate system have characteristic times
of less than a year, while most proxies have longer characteristic times. In addition, with the uncertainty of dating (see
Part II), it is clear that time is an important factor of error in
the study of the interactions between climate and proxy.
Among proxies providing a seasonal resolution, tree rings
are prominent, but this is at the expense of the robustness of
the long term signal. Although it is possible in theory to
reach 10,000 years, trees have a much shorter lifespan and
the long extended data series are achieved by splicing many
short series together. The behavior of trees in the low frequency range is not exclusively due to climate, and this can
lead to significant disturbances in reconstructions. Glaciers,
another paleoclimate indicator, often have a high resolution
for recent periods, due to compaction of the ice, but this
reduces progressively as we go back in time. The same goes
for marine and lake cores. In some cases, the deposition of
their sediments may have an annual resolution (varved
sediments), but bioturbation (disturbance of the sediment by
small aquatic animals) often prevents this level being
achieved in practice.
Historical records are often very accurate, but they have
strong differences related to changes in instruments or
observers. This provides series with a high resolution but
over short periods. This review of climate records shows that
no single proxy is perfect, and that without multi-proxy
comparison, errors of interpretation could easily occur.
Uncertainties Associated with Geochemical
Indicators: The Specific Case of the Isotopic
Composition of Precipitations
Over the past forty years, quantitative reconstructions of
temperature changes have been obtained from the estimation
of past changes in the isotopic composition of precipitation.
These are measured in various continental archives which
hold ancient precipitation directly (glaciers, ice caps,
groundwaters) or indirectly. Indeed, past changes in the
isotopic composition of precipitation can leave a fingerprint
in the isotopic composition of molecules formed using this
water such as the calcite fossil skeletons of lake microorganisms, the calcite of speleothems, or the cellulose of tree
rings.
Different stable isotopic forms of the water molecule are
present on Earth. Their abundance is expressed by reference
to the international standard SMOW (Standard Mean Ocean
Water), which has 0.038% of H
17
2 O, 0.310% of HDO,
0.2005% of H
18
2 O and 99.762% of the principal form, H
16
2 O.
The different isotopic molecules are characterized by a different number of neutrons, and therefore, different masses;
different vapor saturation pressures; as well as differences in
symmetry. During each change of phase (condensation,
evaporation), the water molecules undergo isotopic fractionation, which includes processes at equilibrium (exchanges between infinite reservoirs) and kinetic processes:
during evaporation on the surface of the ocean or
re-evaporation of rain drops during their precipitation, or
during the formation of ice crystals in the clouds, the processes are faster than the diffusion time of water molecules
and cause kinetic effects associated with the molecular diffusivity of the different isotopic forms.
Since 1958, the International Atomic Energy Agency
coordinates a network of observations and a database of the
isotopic compositions of precipitations in modern times
(Fig. 10.2a). Since the 1960s, the measurements have revealed
a close relationship between the isotopic composition of precipitations and air temperature (the ‘isotopic thermometer’,
Fig. 10.2b). This relationship at a local level has been used
intensively to quantify the changes in past temperatures.
The use of natural archives to estimate variations in past
climates from the isotopic composition of oxygen or of
deuterium nevertheless poses many problems which are
sources of uncertainty for the quantification of climate
reconstructions:
10 Reconstructing the Physics and Circulation of the Atmosphere
139
glaciers). This approach is also possible with data from
historical documents, which, when they do not provide
direct climatic information, recount events related to a climate variable (floods, droughts, freezing) that may be standardized semi-quantitatively (Pfister 1980).
Another approach is required for analysis of the living
world. The climate signal recorded by microorganisms is
complex and is a reaction to a combination of several climatic variables; temperature, salinity, and nutrients in the
ocean; temperature and precipitation on land. It is therefore
not possible to decode this signal with a simple equation and
a multivariate approach is needed to interpret the changes in
sets of pollen and diatoms in a continental environment, and
of diatoms and foraminifera in a marine setting (Chap. 21).
On a smaller timescale, the thickness or density of a tree
ring, also influenced by a complex environment, can seldom
be interpreted with a simple calibration equation. Several
series from the same region need to be used to get a clear
climate signal.
Uncertainties on the Temporal Scale
Before discussing the uncertainties and limitations associated with the interpretation of series of proxies, we will
review the uncertainties related to their temporal resolution.
Figure 10.1 shows the temporal characteristics of climate
forcings (internal and external) and the different types of
supports of commonly studied proxies. Many forcings and
components of the climate system have characteristic times
of less than a year, while most proxies have longer characteristic times. In addition, with the uncertainty of dating (see
Part II), it is clear that time is an important factor of error in
the study of the interactions between climate and proxy.
Among proxies providing a seasonal resolution, tree rings
are prominent, but this is at the expense of the robustness of
the long term signal. Although it is possible in theory to
reach 10,000 years, trees have a much shorter lifespan and
the long extended data series are achieved by splicing many
short series together. The behavior of trees in the low frequency range is not exclusively due to climate, and this can
lead to significant disturbances in reconstructions. Glaciers,
another paleoclimate indicator, often have a high resolution
for recent periods, due to compaction of the ice, but this
reduces progressively as we go back in time. The same goes
for marine and lake cores. In some cases, the deposition of
their sediments may have an annual resolution (varved
sediments), but bioturbation (disturbance of the sediment by
small aquatic animals) often prevents this level being
achieved in practice.
Historical records are often very accurate, but they have
strong differences related to changes in instruments or
observers. This provides series with a high resolution but
over short periods. This review of climate records shows that
no single proxy is perfect, and that without multi-proxy
comparison, errors of interpretation could easily occur.
Uncertainties Associated with Geochemical
Indicators: The Specific Case of the Isotopic
Composition of Precipitations
Over the past forty years, quantitative reconstructions of
temperature changes have been obtained from the estimation
of past changes in the isotopic composition of precipitation.
These are measured in various continental archives which
hold ancient precipitation directly (glaciers, ice caps,
groundwaters) or indirectly. Indeed, past changes in the
isotopic composition of precipitation can leave a fingerprint
in the isotopic composition of molecules formed using this
water such as the calcite fossil skeletons of lake microorganisms, the calcite of speleothems, or the cellulose of tree
rings.
Different stable isotopic forms of the water molecule are
present on Earth. Their abundance is expressed by reference
to the international standard SMOW (Standard Mean Ocean
Water), which has 0.038% of H
17
2 O, 0.310% of HDO,
0.2005% of H
18
2 O and 99.762% of the principal form, H
16
2 O.
The different isotopic molecules are characterized by a different number of neutrons, and therefore, different masses;
different vapor saturation pressures; as well as differences in
symmetry. During each change of phase (condensation,
evaporation), the water molecules undergo isotopic fractionation, which includes processes at equilibrium (exchanges between infinite reservoirs) and kinetic processes:
during evaporation on the surface of the ocean or
re-evaporation of rain drops during their precipitation, or
during the formation of ice crystals in the clouds, the processes are faster than the diffusion time of water molecules
and cause kinetic effects associated with the molecular diffusivity of the different isotopic forms.
Since 1958, the International Atomic Energy Agency
coordinates a network of observations and a database of the
isotopic compositions of precipitations in modern times
(Fig. 10.2a). Since the 1960s, the measurements have revealed
a close relationship between the isotopic composition of precipitations and air temperature (the ‘isotopic thermometer’,
Fig. 10.2b). This relationship at a local level has been used
intensively to quantify the changes in past temperatures.
The use of natural archives to estimate variations in past
climates from the isotopic composition of oxygen or of
deuterium nevertheless poses many problems which are
sources of uncertainty for the quantification of climate
reconstructions:
10 Reconstructing the Physics and Circulation of the Atmosphere
139