210
Wolffetal.
salinity relationship. When global data are used a
regression line with a = 0.5 and b = -17 can be inferred (Broecker 1986). However, different regions in the oceans are governed by different equations (Berger and Gardner 1975; Fairbanks et a!.
1992). While the slope of the 1)180w-salinity relationship approaches 1.0 towards the polar regions
(Vetshteyn eta!. 1974, Fairbanks eta!. 1992), it becomes smaller towards the equator. Fairbanks computed a slope of 0.19 for the western equatorial
Atlantic. In the eastern equatorial Atlantic it is as
low as 0.08.
Why is there a relationship between 1) 18 0 of
sea-water and salinity? Both the 18 0/ 16 0 ratio and
salinity are controlled by the evaporation-precipitation regime. Excess of evaporation over precipitation drives both salinity and 1) 18 0 W up, because
salt is left behind during evaporation and oxygen
isotopes are fractionated with water vapor being
depleted in 1) 18 0 relative to sea-water.
But, while the salt content of water vapor is
constant and essentially equal to zero, the 1) 18 0 of
water vapor can vary considerably depending on
its origin and its residence time in the atmosphere.
While evaporating and precipitating waters have
quite similar isotopic signatures in the tropics, precipitation in higher latitude is increasingly depleted
in 180 due to further fractionation (Craig and
Gordon 1965; Wefer and Berger 1991). This is the
reason why the slope ofthe straight line in equation (5) increases at higher latitudes.
Various other influences complicate the system,
e.g. fractionation during evaporation and during
formation of rainfall is influenced by windspeed and
temperature, respectively (Merlivat and Jouzel
1979). In fact, the relation between isotope enrichment and salinity increase is characterized by nonlinearity (Ferron sky and Brezgunov 1989).
In a regional context, however, isotopic signatures of precipitation and evaporation remain rather
constant in the long-term annual mean. A linear
1) 18 0 w -salinity relationship as is expressed in equation (5) can be assumed (Epstein and Mayeda
1953). This straight line represents a continuum
between dry and moist conditions in a given region.
To reconstruct absolute salinity values from seawater 1) 18 0, a 8 18 0 w -salinity relationship has to be
adopted and, if we ignore regional changes in the
evaporation-precipitation regime, modifications
have to be performed accounting for global effects
due to water storage in the continental ice-caps.
These corrections reflecting past variations of both
the 8180 of sea-water and salinity can either be
made as individual computational steps or can be
included in equation (5). We prefer the second strategy because it allows the formulation of a paleo1) 18 0 W salinity relationship which is useful in the
view of numerical models that include the transport of 1) 18 0 in both atmosphere (Hoffmann 1995)
and ocean (Paul et a!. this volume). Also, it allows
the evaluation of the hydrological cycle at a certain time in the past (Wolff et a!. 1998).
Rewriting equation (5) gives:
(6)
where d is the correction for 1) 18 0 of sea-water
g
and S g is the correction term for globally changing
salinities. Figure 2 graphically illustrates the shift
of the 1)180w-salinity relationship neglecting
changes in the slope of the straight line.
Solving equation (6) for salinity yields:
(7)
How do we compute the correction terms d.
and S.? Maximum global glacial-interglacial
changes in 1) 18 0 W and S have been estimated as
1.2 %0 (see Wefer et al. 1996) and 1.1 %0 (Rostek
et al. 1993), respectively .The correction terms d g
can be approximated from a global 1) 18 0 W time
series, e.g. from the SPECMAP stack or the curve
given by Labeyrie et al. (1987) which have to be
adjusted to a maximum glacial-interglacial~1)180w
of 1.2 %0. A gioball) 18 0 w -curve based on Labeyrie
et a!. (1987) and Vogelsang (1990) for the last 360
kyrs is shown in Fig. 9d. The correction term for
salinity Sg can be obtained in the same way by
normalizing the global curve to 1.1 %0. Another
approach is to relate Sg to changes in sea level
relative to the modern should a sea level curve be
available (Rostek et al. 1993).
Wolffetal.
salinity relationship. When global data are used a
regression line with a = 0.5 and b = -17 can be inferred (Broecker 1986). However, different regions in the oceans are governed by different equations (Berger and Gardner 1975; Fairbanks et a!.
1992). While the slope of the 1)180w-salinity relationship approaches 1.0 towards the polar regions
(Vetshteyn eta!. 1974, Fairbanks eta!. 1992), it becomes smaller towards the equator. Fairbanks computed a slope of 0.19 for the western equatorial
Atlantic. In the eastern equatorial Atlantic it is as
low as 0.08.
Why is there a relationship between 1) 18 0 of
sea-water and salinity? Both the 18 0/ 16 0 ratio and
salinity are controlled by the evaporation-precipitation regime. Excess of evaporation over precipitation drives both salinity and 1) 18 0 W up, because
salt is left behind during evaporation and oxygen
isotopes are fractionated with water vapor being
depleted in 1) 18 0 relative to sea-water.
But, while the salt content of water vapor is
constant and essentially equal to zero, the 1) 18 0 of
water vapor can vary considerably depending on
its origin and its residence time in the atmosphere.
While evaporating and precipitating waters have
quite similar isotopic signatures in the tropics, precipitation in higher latitude is increasingly depleted
in 180 due to further fractionation (Craig and
Gordon 1965; Wefer and Berger 1991). This is the
reason why the slope ofthe straight line in equation (5) increases at higher latitudes.
Various other influences complicate the system,
e.g. fractionation during evaporation and during
formation of rainfall is influenced by windspeed and
temperature, respectively (Merlivat and Jouzel
1979). In fact, the relation between isotope enrichment and salinity increase is characterized by nonlinearity (Ferron sky and Brezgunov 1989).
In a regional context, however, isotopic signatures of precipitation and evaporation remain rather
constant in the long-term annual mean. A linear
1) 18 0 w -salinity relationship as is expressed in equation (5) can be assumed (Epstein and Mayeda
1953). This straight line represents a continuum
between dry and moist conditions in a given region.
To reconstruct absolute salinity values from seawater 1) 18 0, a 8 18 0 w -salinity relationship has to be
adopted and, if we ignore regional changes in the
evaporation-precipitation regime, modifications
have to be performed accounting for global effects
due to water storage in the continental ice-caps.
These corrections reflecting past variations of both
the 8180 of sea-water and salinity can either be
made as individual computational steps or can be
included in equation (5). We prefer the second strategy because it allows the formulation of a paleo1) 18 0 W salinity relationship which is useful in the
view of numerical models that include the transport of 1) 18 0 in both atmosphere (Hoffmann 1995)
and ocean (Paul et a!. this volume). Also, it allows
the evaluation of the hydrological cycle at a certain time in the past (Wolff et a!. 1998).
Rewriting equation (5) gives:
(6)
where d is the correction for 1) 18 0 of sea-water
g
and S g is the correction term for globally changing
salinities. Figure 2 graphically illustrates the shift
of the 1)180w-salinity relationship neglecting
changes in the slope of the straight line.
Solving equation (6) for salinity yields:
(7)
How do we compute the correction terms d.
and S.? Maximum global glacial-interglacial
changes in 1) 18 0 W and S have been estimated as
1.2 %0 (see Wefer et al. 1996) and 1.1 %0 (Rostek
et al. 1993), respectively .The correction terms d g
can be approximated from a global 1) 18 0 W time
series, e.g. from the SPECMAP stack or the curve
given by Labeyrie et al. (1987) which have to be
adjusted to a maximum glacial-interglacial~1)180w
of 1.2 %0. A gioball) 18 0 w -curve based on Labeyrie
et a!. (1987) and Vogelsang (1990) for the last 360
kyrs is shown in Fig. 9d. The correction term for
salinity Sg can be obtained in the same way by
normalizing the global curve to 1.1 %0. Another
approach is to relate Sg to changes in sea level
relative to the modern should a sea level curve be
available (Rostek et al. 1993).