and important role. The transient tracers (e.g. Chlorofluorocarbons (CFCs) and tritium/helium-3) provide one of the few methods for following rates of
water mass formation and identifying newly ventilated water masses (e.g. Doney et al., 1998a;
Smethie et al., 2000; Schlosser et al., Chapter 5.8).
Tracers also provide a method for following the
pathways of newly ventilated waters and for determining rates and time scales for water mass spreading. Warner et al. (1996) provide just one example
among many of how measurements and inventories
of CFCs and other tracers can be used to follow the
three-dimensional circulation, knowledge that is
critical for understanding prediction and quantifying predictability, particularly on interannual,
decadal and longer time scales. England and Hirst
(1997) provide one illustration of how such information can be effective in validating and understanding the sensitivity of climate models.
Ultimately, all ocean data are important for
improving our capacity to undertake ocean and
climate prediction. Prediction and predictability
depend fundamentally on knowledge of processes
and WOCE has made a singular contribution to
enhancing understanding, particularly at longer
time scales. Moreover, the methods developed by
WOCE, from profiling floats through to tracer
measurements, provide the technical knowledge
that will be needed as we extend our horizons
beyond ENSO prediction to global prediction, at
longer time scales, and with a higher premium
on detail. The comprehensiveness of the WOCE
data set is also a unique asset that will be of
value as the present generation of climate models
evolve into more complex representations of the
environment.
7.4.3.4 The future
For endeavours such as GOOS and GODAE,
WOCE observations and observational techniques
provide a substantial foundation. Altimetry and
profiling floats (Argo) are arguably the outstanding contributions. However, significant developments are also emerging from other techniques
used by WOCE. Both the VOS and SOOP are
likely to undergo major renovation (Smith et al.,
1999b; Taylor et al., 2001) and we are seeing
enhanced efforts in fixed-point and surface reference data collection (Send et al., 2001). The sealevel network has changed emphases in recognition
of a new capability (for example, the role of precision altimetry in monitoring global change) and
because of improved ability to use sea-level data
for ocean and climate data assimilation (Intergovernmental Oceanographic Commission, 1997;
Ocean Observations Panel for Climate, 1997).
7.4.4 WOCE and climate prediction
One of the fundamental questions posed by
TOGA concerned predictability. Were observed
interannual variations in climate, of which ENSO
is the most prominent, predictable? If so, could it
be exploited for practical purpose, and what were
the limits of this predictability? Based on research
before and during TOGA, it was determined that
not only was ENSO predictable, but it could be
exploited in a practical way. Cane and Zebiak
(1987) provided the first practical demonstration
and many simple and complex coupled models
have followed (Barnston et al., 1999).
But TOGA also left a number of questions
unanswered. For example:
¥ What role do the oceans outside the tropical
Pacific, and deeper than 500 m, play in prediction and predictability?
¥ What are the reasons for temporal variations in
skill?
¥ What are the key data sets for prediction?
The plan here is to explore the role WOCE may
play in answering these questions.
7.4.4.1 ENSO prediction
There are now many models developed for predicting ENSO and, in theory, other seasonal-tointerannual climate variations. It is accepted that
these models have skill (see, for example, Wang
et al., 2001). Trenberth (1998b) discussed the performance of various models during the 1997–98
El Niño event and concluded that the models had
performed well in general. Barnston et al. (1999)
performed an evaluation based on published forecasts and, though agreeing that models displayed
skill, noted that the translation and interpretation
of model predictions to the public lacked precision
and objectivity.
The determination of predictability is normally
done with data analyses or model simulations. In
the case of ENSO, studies dating back to Bjerknes
(1966, 1969) have shown that signals in the
SECTION 7 INSIGHTS FOR THE FUTURE
592
water mass formation and identifying newly ventilated water masses (e.g. Doney et al., 1998a;
Smethie et al., 2000; Schlosser et al., Chapter 5.8).
Tracers also provide a method for following the
pathways of newly ventilated waters and for determining rates and time scales for water mass spreading. Warner et al. (1996) provide just one example
among many of how measurements and inventories
of CFCs and other tracers can be used to follow the
three-dimensional circulation, knowledge that is
critical for understanding prediction and quantifying predictability, particularly on interannual,
decadal and longer time scales. England and Hirst
(1997) provide one illustration of how such information can be effective in validating and understanding the sensitivity of climate models.
Ultimately, all ocean data are important for
improving our capacity to undertake ocean and
climate prediction. Prediction and predictability
depend fundamentally on knowledge of processes
and WOCE has made a singular contribution to
enhancing understanding, particularly at longer
time scales. Moreover, the methods developed by
WOCE, from profiling floats through to tracer
measurements, provide the technical knowledge
that will be needed as we extend our horizons
beyond ENSO prediction to global prediction, at
longer time scales, and with a higher premium
on detail. The comprehensiveness of the WOCE
data set is also a unique asset that will be of
value as the present generation of climate models
evolve into more complex representations of the
environment.
7.4.3.4 The future
For endeavours such as GOOS and GODAE,
WOCE observations and observational techniques
provide a substantial foundation. Altimetry and
profiling floats (Argo) are arguably the outstanding contributions. However, significant developments are also emerging from other techniques
used by WOCE. Both the VOS and SOOP are
likely to undergo major renovation (Smith et al.,
1999b; Taylor et al., 2001) and we are seeing
enhanced efforts in fixed-point and surface reference data collection (Send et al., 2001). The sealevel network has changed emphases in recognition
of a new capability (for example, the role of precision altimetry in monitoring global change) and
because of improved ability to use sea-level data
for ocean and climate data assimilation (Intergovernmental Oceanographic Commission, 1997;
Ocean Observations Panel for Climate, 1997).
7.4.4 WOCE and climate prediction
One of the fundamental questions posed by
TOGA concerned predictability. Were observed
interannual variations in climate, of which ENSO
is the most prominent, predictable? If so, could it
be exploited for practical purpose, and what were
the limits of this predictability? Based on research
before and during TOGA, it was determined that
not only was ENSO predictable, but it could be
exploited in a practical way. Cane and Zebiak
(1987) provided the first practical demonstration
and many simple and complex coupled models
have followed (Barnston et al., 1999).
But TOGA also left a number of questions
unanswered. For example:
¥ What role do the oceans outside the tropical
Pacific, and deeper than 500 m, play in prediction and predictability?
¥ What are the reasons for temporal variations in
skill?
¥ What are the key data sets for prediction?
The plan here is to explore the role WOCE may
play in answering these questions.
7.4.4.1 ENSO prediction
There are now many models developed for predicting ENSO and, in theory, other seasonal-tointerannual climate variations. It is accepted that
these models have skill (see, for example, Wang
et al., 2001). Trenberth (1998b) discussed the performance of various models during the 1997–98
El Niño event and concluded that the models had
performed well in general. Barnston et al. (1999)
performed an evaluation based on published forecasts and, though agreeing that models displayed
skill, noted that the translation and interpretation
of model predictions to the public lacked precision
and objectivity.
The determination of predictability is normally
done with data analyses or model simulations. In
the case of ENSO, studies dating back to Bjerknes
(1966, 1969) have shown that signals in the
SECTION 7 INSIGHTS FOR THE FUTURE
592