height. They showed that eddy variability was the
dominant mechanism for interannual change in
equatorward transport (over the period 1991–99),
a potentially important mechanism for coupled
model prediction on interannual time scales. Such
results, together with other studies on sampling
and representativeness (Smith et al., 1999b), are
able to guide future measurement strategies and
provide direction on the minimal requirements for
in-situ data. There are also many examples of XBT
data being used to monitor boundary currents.
Meyers (1996) has monitored the upper portion of
the Indonesian Throughflow using data from an
XBT line established in 1984, initially as part of
the TOGA network, but more recently as a
WOCE contribution. This line (designated as IX1)
is the basis for constructing the longest continuous
estimate of the variability of upper part of the
Indonesian Throughflow.
The utility of sea-level measurements for monitoring tropical ocean variability has been known
for some time (e.g. Wyrtki, 1979), though by the
end of TOGA the consensus was that sea-level
measurements were not the highest priority for
ENSO prediction (OOSDP, 1993; International
Toga Project Office, 1994). The tropical island
data were a shared interest with WOCE and
arrangements were made for fast delivery of data,
including key data sets beyond the tropics, to aid
in the calibration of altimetric data. More recently
a study and review were conducted to determine
the appropriate network for all applications,
including climate change (OOSDP, 1997). The
study concluded that the global tide gauge network had multiple benefits, including for initialization of climate models.
One of the observational methods that are now
revolutionizing the way we do oceanography has
its origins firmly in WOCE. Floats were originally
developed as a way of taking Lagrangian measurements of ocean currents at depth. In order to solve
the problem of providing regular fixes on the position of the float, a depth-cycling version was
developed that would regularly rise from the preassigned depth to the surface where satellites could
be used for fixing and communication. It was
quickly realized that this also provided a convenient platform for gathering information on temperature profiles. The design was adapted to
include temperature sensors and the profiling float
was born. Through a recent initiative called Argo
(Argo Science Team, 1998), a global programme
of regularly profiling floats is being developed
with the promise of around 80 000–100 000 profiles of temperature and salinity over the upper
2000 m being returned annually, from around
3000 floats.
As noted above, there was an understandable
emphasis on deep measurements in WOCE since it
was believed that the mean circulation was fundamentally determined by the slow, deep circulation
of heat and fresh water. Given the limited
resources and limited lifetime, it was not feasible
to resolve fully the upper ocean variability; this
places some limitation on the utility of WOCE
observations for studying seasonal-to-interannual
predictability, at least on the shorter time scales
(see Section 7.4.5 and Talley et al., Chapter 7.1).
There are, however, several aspects of the deep
measurement programme of WOCE that are of
direct relevance to ocean and climate prediction.
For example, the Japan Meteorological Agency
lines at 137°E provide a unique view of interannual variability as a function of depth in the western and equatorial Pacific (e.g. Qiu and Joyce,
1992) and, together with other North Pacific data
sets, provide a context for developing theories on
predictability on interannual-to-decadal time
scales. Johnson et al. (2000) used data collected
during regular cruises to service the tropical mooring array to provide a fuller context for upperocean variability in the equatorial Pacific during
1996–98. Dickson et al. (Chapter 7.3) provide
several other examples of enhanced knowledge of
variability resulting from deep ocean measurements taken through WOCE.
Ocean and climate prediction can be thought of
as a challenge in two parts. First, through observation and modelling, one must try to identify and
delineate ocean modes (signals) and processes that
are important for the future evolution of (coupled)
climate modes. Then, again through the skilful
application of models, one attempts to follow the
pathways and progress of these modes into the
future, all the while accepting the fact that natural
and artificial error growth will erode skill along the
paths and with time (see also the discussion in
Section 7.4.6). As the lead times are pushed further
into the future we expect errors associated with
intermediate and deep water circulation to play an
increasingly important role. In this respect, tracer
measurements taken during WOCE have a unique
7.4 Ocean and Climate Prediction – the WOCE Legacy
591
Smith
dominant mechanism for interannual change in
equatorward transport (over the period 1991–99),
a potentially important mechanism for coupled
model prediction on interannual time scales. Such
results, together with other studies on sampling
and representativeness (Smith et al., 1999b), are
able to guide future measurement strategies and
provide direction on the minimal requirements for
in-situ data. There are also many examples of XBT
data being used to monitor boundary currents.
Meyers (1996) has monitored the upper portion of
the Indonesian Throughflow using data from an
XBT line established in 1984, initially as part of
the TOGA network, but more recently as a
WOCE contribution. This line (designated as IX1)
is the basis for constructing the longest continuous
estimate of the variability of upper part of the
Indonesian Throughflow.
The utility of sea-level measurements for monitoring tropical ocean variability has been known
for some time (e.g. Wyrtki, 1979), though by the
end of TOGA the consensus was that sea-level
measurements were not the highest priority for
ENSO prediction (OOSDP, 1993; International
Toga Project Office, 1994). The tropical island
data were a shared interest with WOCE and
arrangements were made for fast delivery of data,
including key data sets beyond the tropics, to aid
in the calibration of altimetric data. More recently
a study and review were conducted to determine
the appropriate network for all applications,
including climate change (OOSDP, 1997). The
study concluded that the global tide gauge network had multiple benefits, including for initialization of climate models.
One of the observational methods that are now
revolutionizing the way we do oceanography has
its origins firmly in WOCE. Floats were originally
developed as a way of taking Lagrangian measurements of ocean currents at depth. In order to solve
the problem of providing regular fixes on the position of the float, a depth-cycling version was
developed that would regularly rise from the preassigned depth to the surface where satellites could
be used for fixing and communication. It was
quickly realized that this also provided a convenient platform for gathering information on temperature profiles. The design was adapted to
include temperature sensors and the profiling float
was born. Through a recent initiative called Argo
(Argo Science Team, 1998), a global programme
of regularly profiling floats is being developed
with the promise of around 80 000–100 000 profiles of temperature and salinity over the upper
2000 m being returned annually, from around
3000 floats.
As noted above, there was an understandable
emphasis on deep measurements in WOCE since it
was believed that the mean circulation was fundamentally determined by the slow, deep circulation
of heat and fresh water. Given the limited
resources and limited lifetime, it was not feasible
to resolve fully the upper ocean variability; this
places some limitation on the utility of WOCE
observations for studying seasonal-to-interannual
predictability, at least on the shorter time scales
(see Section 7.4.5 and Talley et al., Chapter 7.1).
There are, however, several aspects of the deep
measurement programme of WOCE that are of
direct relevance to ocean and climate prediction.
For example, the Japan Meteorological Agency
lines at 137°E provide a unique view of interannual variability as a function of depth in the western and equatorial Pacific (e.g. Qiu and Joyce,
1992) and, together with other North Pacific data
sets, provide a context for developing theories on
predictability on interannual-to-decadal time
scales. Johnson et al. (2000) used data collected
during regular cruises to service the tropical mooring array to provide a fuller context for upperocean variability in the equatorial Pacific during
1996–98. Dickson et al. (Chapter 7.3) provide
several other examples of enhanced knowledge of
variability resulting from deep ocean measurements taken through WOCE.
Ocean and climate prediction can be thought of
as a challenge in two parts. First, through observation and modelling, one must try to identify and
delineate ocean modes (signals) and processes that
are important for the future evolution of (coupled)
climate modes. Then, again through the skilful
application of models, one attempts to follow the
pathways and progress of these modes into the
future, all the while accepting the fact that natural
and artificial error growth will erode skill along the
paths and with time (see also the discussion in
Section 7.4.6). As the lead times are pushed further
into the future we expect errors associated with
intermediate and deep water circulation to play an
increasingly important role. In this respect, tracer
measurements taken during WOCE have a unique
7.4 Ocean and Climate Prediction – the WOCE Legacy
591
Smith