7.3.1 Assessing the representativeness of
the WOCE data set
From the start, it has been clear that the primary
WOCE goal of compiling a coherent global ocean
description would also require us to ask whether
the new WOCE dataset is in any sense ‘representative’ of the long-term behaviour of the ocean. This
secondary goal forms the subject of this chapter.
Equally clearly, we can expect to provide only a
partial answer, for two main reasons.
First, to assess how the WOCE period varied
from some pre-existing state, we need pre-existing
data, and that requirement simply can’t be built in
to the experiment design. The very fact that the
WOCE data set is of unprecedented scope, accuracy and quality means that comparable data will
not exist in many ocean areas and depths.
Second, even where such comparisons prove
possible, their value and validity will depend on
whether our past and present sampling was adequate to resolve the amplitudes and time scales of
ocean variability. Our longest time series may well
be adequate for this task, but these are few and
sparse. It will be much less obvious whether and to
what extent the relatively infrequent repeat hydrography of the WOCE Hydrographic Programme (for
example) was able to resolve the important variability of the upper, intermediate and deep ocean.
To illustrate the point, our comparison of section pairs from the mid- and South Atlantic (Section 7.3.4) will show that the variability signal was
large in the top 1000 m but was inconsistent from
section to section, suggesting that high-frequency
variability (the estimated ‘eddy error’) was masking
any smaller-amplitude but more extensive decadal
changes. Below 2000 m, on the other hand, though
the temperature change seemed consistent between
sections (cooling), the magnitude of the temperature change was small (order 0.03°C), and small
systematic errors in the salinity data (hence density) from earlier years argued against decomposing this change into components along isopycnals
and those due to isopycnal displacement. The signals between 1000 and 2000 db were however of
consistent sign (warming) from 32°S to 36°N, had
good signal-to-eddy noise ratios, and the decompositions of the signals didn’t change qualitatively
when the possible systematic salt errors were
corrected.
For these sorts of reasons, we can expect the
WOCE snapshot to be blurred in the near-surface
layers and for the intermediate layers to carry the
most reliable and accessible signal of decade-todecade variability in the sparse section data available to us; these intermediate layers will form one
focus of this chapter. Bindoff and McDougall
(1994) have shown that the most thorough
diagnosis of water mass changes results from an
examination of changes on isobaric and neutral
surfaces; both are employed here. A second focus
will concern the long-term variability of the convectively formed mode waters of the North Atlantic;
here, we can expect the signals to be large, vertically homogeneous and relatively well described
over many decades. We place our third emphasis
on the changes in transport of the Gulf Stream,
7.3
The World during WOCE
Bob Dickson, Jim Hurrell, Nathan Bindoff, Annie Wong, Brian Arbic,
Breck Owens, Shiro Imawaki and Igor Yashayaev
557
OCEAN CIRCULATION AND CLIMATE
Copyright © 2001 Academic Press
ISBN 0-12-641351-7
All rights of reproduction in any form reserved
CHAPTER
the WOCE data set
From the start, it has been clear that the primary
WOCE goal of compiling a coherent global ocean
description would also require us to ask whether
the new WOCE dataset is in any sense ‘representative’ of the long-term behaviour of the ocean. This
secondary goal forms the subject of this chapter.
Equally clearly, we can expect to provide only a
partial answer, for two main reasons.
First, to assess how the WOCE period varied
from some pre-existing state, we need pre-existing
data, and that requirement simply can’t be built in
to the experiment design. The very fact that the
WOCE data set is of unprecedented scope, accuracy and quality means that comparable data will
not exist in many ocean areas and depths.
Second, even where such comparisons prove
possible, their value and validity will depend on
whether our past and present sampling was adequate to resolve the amplitudes and time scales of
ocean variability. Our longest time series may well
be adequate for this task, but these are few and
sparse. It will be much less obvious whether and to
what extent the relatively infrequent repeat hydrography of the WOCE Hydrographic Programme (for
example) was able to resolve the important variability of the upper, intermediate and deep ocean.
To illustrate the point, our comparison of section pairs from the mid- and South Atlantic (Section 7.3.4) will show that the variability signal was
large in the top 1000 m but was inconsistent from
section to section, suggesting that high-frequency
variability (the estimated ‘eddy error’) was masking
any smaller-amplitude but more extensive decadal
changes. Below 2000 m, on the other hand, though
the temperature change seemed consistent between
sections (cooling), the magnitude of the temperature change was small (order 0.03°C), and small
systematic errors in the salinity data (hence density) from earlier years argued against decomposing this change into components along isopycnals
and those due to isopycnal displacement. The signals between 1000 and 2000 db were however of
consistent sign (warming) from 32°S to 36°N, had
good signal-to-eddy noise ratios, and the decompositions of the signals didn’t change qualitatively
when the possible systematic salt errors were
corrected.
For these sorts of reasons, we can expect the
WOCE snapshot to be blurred in the near-surface
layers and for the intermediate layers to carry the
most reliable and accessible signal of decade-todecade variability in the sparse section data available to us; these intermediate layers will form one
focus of this chapter. Bindoff and McDougall
(1994) have shown that the most thorough
diagnosis of water mass changes results from an
examination of changes on isobaric and neutral
surfaces; both are employed here. A second focus
will concern the long-term variability of the convectively formed mode waters of the North Atlantic;
here, we can expect the signals to be large, vertically homogeneous and relatively well described
over many decades. We place our third emphasis
on the changes in transport of the Gulf Stream,
7.3
The World during WOCE
Bob Dickson, Jim Hurrell, Nathan Bindoff, Annie Wong, Brian Arbic,
Breck Owens, Shiro Imawaki and Igor Yashayaev
557
OCEAN CIRCULATION AND CLIMATE
Copyright © 2001 Academic Press
ISBN 0-12-641351-7
All rights of reproduction in any form reserved
CHAPTER