The essential characteristics of this ‘pure warming’
process are that at a given depth (1) the salinity of
the water column is unchanged and (2) the potential temperature has increased.
Pure freshening
Instead of warming the surface waters, suppose
now that the water in the mixed layer has been
freshened by increased precipitation–evaporation
and there is no change in the heat flux from the
atmosphere (Fig. 7.3.3a) or in the winds. Following
the same scenario as in the ‘pure warming’ process,
after subduction and transport to the hydrographic
cast between the two material surfaces, the potential temperature is unchanged, the salinity has
decreased, and density surfaces are now slightly
deeper (Fig. 7.3.3c). Along the isopycnal, 1 , the
freshening on pressure surfaces (SЈΗ z :0) has
resulted in both a cooling and a freshening on density surfaces (respectively ЈΗ n , SЈΗ n :0).
Thus changes on density and pressure surfaces
must be treated carefully if they are to be interpreted in terms of changes in surface water masses.
The distinguishing feature between the warming
and freshening processes is how they affect temperature and salinity on both pressure and density
surfaces. It is possible to separate formally the
relative roles of pure warming and freshening
(Bindoff and McDougall, 1994). This is important
because models of climate change for increased
greenhouse gases show increased precipitation at
high latitudes as well as increased surface temperatures (Gordon and O’Farrell, 1997), both of which
contribute significantly to the modelled buoyancy
changes. Such changes may equally well occur as a
result of natural climate variability.
Pure heave
Pure heave occurs where water masses are
unchanged (ЈΗ n ,:SЈΗ n :0) but isopycnals have been
displaced vertically. Although this process has no signature on the –S diagram, it can lead to large temperature and salinity changes on pressure surfaces.
Experience has shown that temperature and salinity differences on pressure surfaces in deep waters
between repeated hydrographic sections taken 6
months apart are characterized by the pure heave
process (Bindoff and McDougall, 1994) and probably resulted from variations in the seasonal winds.
Thus, the principal advantage of using temperature and salinity changes on density surfaces lies in
their ability to discriminate between changes due to
pure heave (resulting mainly from short-term fluctuations due to dynamical processes such as propagating Rossby waves or seasonal variations in
gyre circulation) and those due to surface forcing.
A secondary advantage is that on density surfaces
the prior noise in the thermocline is much smaller
and changes of temperature and salinity on density
surfaces tend to have larger signal-to-noise ratios.
The analysis that follows for the South Atlantic
is essentially similar to that for the Pacific (Section
7.3.9). However, for the Atlantic Ocean basins:
(1) there are multiple sources for the intermediate
water; and (2) isopycnal mixing along the tenuous
paths from the sources is likely to be important. As
a result, the complete analysis, decomposing the
observed changes into freshening and warming, is
not warranted, and the decompositions of the
observed changes on isobaric surfaces were instead
limited to changes along neutral surfaces and those
due to heave.
7.3.4 Climatic warming of Atlantic
Intermediate Waters
In this section, the interdecadal temperature variability of the Atlantic Ocean is investigated by
comparing hydrographic sections worked in the
1920s, 1950s, 1980s and 1990s. Previous studies
(Roemmich and Wunsch, 1984; Parrilla et al.,
1994; Bryden et al., 1996; Joyce and Robbins,
1996; Arhan et al., 1998; Levitus, 1989; Joyce
et al., 1999) have concentrated on changes in the
North Atlantic since the 1950s. Here, we reanalyse these changes and extend the analysis to
the South Atlantic and back to the 1920s.
For this study we extracted data from the following sources: three South Atlantic sections from
the Meteor expedition of 1925–27 (Defant, 1936),
plus a total of seven North Atlantic and four South
Atlantic sections from the International Geophysical Year (IGY, 1956–60), (Fuglister, 1960;
Worthington and Wright, 1970), five North
Atlantic sections from 1981–85 (Roemmich and
Wunsch, 1984; McCartney, 1992; Knapp, 1988;
Hendry, 1989), one from 1992 (Parrilla et al.,
1994), and one from 1993 (Arhan et al., 1998).
These were supplemented by two 1983 sections
(McCartney and Woodgate-Jones, 1991; Warren
and Speer, 1991) and one 1994 section (Holfort
et al., 2000) from the South Atlantic. All section data
7.3 The World during WOCE
565
Dickson et al.
process are that at a given depth (1) the salinity of
the water column is unchanged and (2) the potential temperature has increased.
Pure freshening
Instead of warming the surface waters, suppose
now that the water in the mixed layer has been
freshened by increased precipitation–evaporation
and there is no change in the heat flux from the
atmosphere (Fig. 7.3.3a) or in the winds. Following
the same scenario as in the ‘pure warming’ process,
after subduction and transport to the hydrographic
cast between the two material surfaces, the potential temperature is unchanged, the salinity has
decreased, and density surfaces are now slightly
deeper (Fig. 7.3.3c). Along the isopycnal, 1 , the
freshening on pressure surfaces (SЈΗ z :0) has
resulted in both a cooling and a freshening on density surfaces (respectively ЈΗ n , SЈΗ n :0).
Thus changes on density and pressure surfaces
must be treated carefully if they are to be interpreted in terms of changes in surface water masses.
The distinguishing feature between the warming
and freshening processes is how they affect temperature and salinity on both pressure and density
surfaces. It is possible to separate formally the
relative roles of pure warming and freshening
(Bindoff and McDougall, 1994). This is important
because models of climate change for increased
greenhouse gases show increased precipitation at
high latitudes as well as increased surface temperatures (Gordon and O’Farrell, 1997), both of which
contribute significantly to the modelled buoyancy
changes. Such changes may equally well occur as a
result of natural climate variability.
Pure heave
Pure heave occurs where water masses are
unchanged (ЈΗ n ,:SЈΗ n :0) but isopycnals have been
displaced vertically. Although this process has no signature on the –S diagram, it can lead to large temperature and salinity changes on pressure surfaces.
Experience has shown that temperature and salinity differences on pressure surfaces in deep waters
between repeated hydrographic sections taken 6
months apart are characterized by the pure heave
process (Bindoff and McDougall, 1994) and probably resulted from variations in the seasonal winds.
Thus, the principal advantage of using temperature and salinity changes on density surfaces lies in
their ability to discriminate between changes due to
pure heave (resulting mainly from short-term fluctuations due to dynamical processes such as propagating Rossby waves or seasonal variations in
gyre circulation) and those due to surface forcing.
A secondary advantage is that on density surfaces
the prior noise in the thermocline is much smaller
and changes of temperature and salinity on density
surfaces tend to have larger signal-to-noise ratios.
The analysis that follows for the South Atlantic
is essentially similar to that for the Pacific (Section
7.3.9). However, for the Atlantic Ocean basins:
(1) there are multiple sources for the intermediate
water; and (2) isopycnal mixing along the tenuous
paths from the sources is likely to be important. As
a result, the complete analysis, decomposing the
observed changes into freshening and warming, is
not warranted, and the decompositions of the
observed changes on isobaric surfaces were instead
limited to changes along neutral surfaces and those
due to heave.
7.3.4 Climatic warming of Atlantic
Intermediate Waters
In this section, the interdecadal temperature variability of the Atlantic Ocean is investigated by
comparing hydrographic sections worked in the
1920s, 1950s, 1980s and 1990s. Previous studies
(Roemmich and Wunsch, 1984; Parrilla et al.,
1994; Bryden et al., 1996; Joyce and Robbins,
1996; Arhan et al., 1998; Levitus, 1989; Joyce
et al., 1999) have concentrated on changes in the
North Atlantic since the 1950s. Here, we reanalyse these changes and extend the analysis to
the South Atlantic and back to the 1920s.
For this study we extracted data from the following sources: three South Atlantic sections from
the Meteor expedition of 1925–27 (Defant, 1936),
plus a total of seven North Atlantic and four South
Atlantic sections from the International Geophysical Year (IGY, 1956–60), (Fuglister, 1960;
Worthington and Wright, 1970), five North
Atlantic sections from 1981–85 (Roemmich and
Wunsch, 1984; McCartney, 1992; Knapp, 1988;
Hendry, 1989), one from 1992 (Parrilla et al.,
1994), and one from 1993 (Arhan et al., 1998).
These were supplemented by two 1983 sections
(McCartney and Woodgate-Jones, 1991; Warren
and Speer, 1991) and one 1994 section (Holfort
et al., 2000) from the South Atlantic. All section data
7.3 The World during WOCE
565
Dickson et al.