tropical Pacific are predictable. Experience is
showing, however, that our levels of skill may
be limited by the imperfect assimilation of information from beyond the tropical Pacific and by
lack of representation of longer-period signals
(Balmaseda et al., 1995; see also Liu and Philander,
Chapter 4.4). Kleeman et al. (1996) and Latif et al.
(1997), among others, have examined the anomalous conditions through the early 1990s and considered mechanisms for the modulation of ENSO,
including extratropical effects.
WOCE data and analyses are likely to play a
key role in resolving these issues since they provide
the first systematic coverage of the mid-latitude
oceans. Studies (e.g. Anderson et al., 2001) are
suggesting that the influence of the other tropical
oceans may be an important factor. WOCE data
will provide the foundations for such research.
WOCE data may also provide a foundation for
identifying other predictable climate signals.
CLIVAR is setting out ambitiously to determine
predictable modes beyond ENSO and WOCE data
sets provide an initial platform for such studies. In
general, one might expect the key information for
longer lead-times (e.g. greater than 6 months) to
be located outside the equatorial waveguide and
beyond the focus of the ENSO observing system. It
is the WOCE observations that are providing some
opportunity to test such ideas.
There is also an inexorable trend toward global
models for ENSO prediction, in part because of
the issues raised above and in part because representation of the impacts of ENSO requires a global
view. It has to be conceded at present that coupled
models for ENSO have relatively little skill beyond
the tropical region (see, for example, Wang et al.,
2001). The WOCE data beyond the Pacific (e.g. in
the Indian Ocean) provide a platform to test and
validate such global models. This trend toward
global domains is being accompanied by a trend
toward more general skill evaluation, typically
global rather than just for the ENSO region (e.g.
Stockdale et al., 1998). Again, the existence of
WOCE (and other) global data sets provides the
opportunity to validate properly outside the region
of the ENSO observing system.
7.4.4.2 Impacting next-generation global
climate prediction models
The impact of WOCE on climate prediction does,
of course, extend beyond the availability of data
provided by WOCE. The inexorable move toward
global coupled models and, implicitly, global
ocean models, means that the climate prediction
community will benefit from ocean modelling and
data assimilation research that has been, and
continues to be, undertaken within WOCE. For
example, Fukumori et al. (1999) discusses useful
measures of consistency in ocean state estimation,
including formal error estimates, the model–data
residuals, and comparisons with independent data,
in an assimilation of TOPEX/POSEIDON data into
a global ocean circulation model. Such results have
direct applications in present climate prediction
systems.
It should also be noted that as we move to operational implementation of climate prediction models, agencies are seeking systems that are more
general. That is, the ocean component of the climate model may also be used to provide ocean
analyses on a range of time scales. This expansion
is being aided by the experience and research
results gained through WOCE. For example, the
model discussed by Wang et al. (2001) also serves
to provide global ocean analyses and will be the
basis for climate change studies. Such synergy has
two-way benefits, allowing the research on global
issues, such as that of WOCE, to be entrained into
operational systems.
7.4.4.3 Data sets for initialization?
Several earlier chapters (for example, Talley et al.,
Chapter 7.1) have demonstrated the great progress
that has been made in the area of data assimilation. While current climate prediction models rely
for the most part on relatively simple data assimilation systems (e.g. Stockdale et al., 1998), the
trend is towards more complex schemes in order
to exploit better new data sets (e.g. altimetry,
Behringer et al., 1998) and to account for other
effects (e.g. salinity, Vossepoel et al., 1999). In
some cases (e.g. Kleeman et al., 1995; Rienecker
et al., 1999) more sophisticated techniques are being
used, both to understand better the impact of the
data and to provide a more sophisticated merging
of observed and modelled information.
The adoption of more sophisticated assimilation
techniques is also greatly improving the ability of
climate prediction models to blend different data
sets (e.g. subsurface temperature and sea level), to
take account of known errors in the data and to
take better advantage of output statistics. An
7.4 Ocean and Climate Prediction – the WOCE Legacy
593
Smith
showing, however, that our levels of skill may
be limited by the imperfect assimilation of information from beyond the tropical Pacific and by
lack of representation of longer-period signals
(Balmaseda et al., 1995; see also Liu and Philander,
Chapter 4.4). Kleeman et al. (1996) and Latif et al.
(1997), among others, have examined the anomalous conditions through the early 1990s and considered mechanisms for the modulation of ENSO,
including extratropical effects.
WOCE data and analyses are likely to play a
key role in resolving these issues since they provide
the first systematic coverage of the mid-latitude
oceans. Studies (e.g. Anderson et al., 2001) are
suggesting that the influence of the other tropical
oceans may be an important factor. WOCE data
will provide the foundations for such research.
WOCE data may also provide a foundation for
identifying other predictable climate signals.
CLIVAR is setting out ambitiously to determine
predictable modes beyond ENSO and WOCE data
sets provide an initial platform for such studies. In
general, one might expect the key information for
longer lead-times (e.g. greater than 6 months) to
be located outside the equatorial waveguide and
beyond the focus of the ENSO observing system. It
is the WOCE observations that are providing some
opportunity to test such ideas.
There is also an inexorable trend toward global
models for ENSO prediction, in part because of
the issues raised above and in part because representation of the impacts of ENSO requires a global
view. It has to be conceded at present that coupled
models for ENSO have relatively little skill beyond
the tropical region (see, for example, Wang et al.,
2001). The WOCE data beyond the Pacific (e.g. in
the Indian Ocean) provide a platform to test and
validate such global models. This trend toward
global domains is being accompanied by a trend
toward more general skill evaluation, typically
global rather than just for the ENSO region (e.g.
Stockdale et al., 1998). Again, the existence of
WOCE (and other) global data sets provides the
opportunity to validate properly outside the region
of the ENSO observing system.
7.4.4.2 Impacting next-generation global
climate prediction models
The impact of WOCE on climate prediction does,
of course, extend beyond the availability of data
provided by WOCE. The inexorable move toward
global coupled models and, implicitly, global
ocean models, means that the climate prediction
community will benefit from ocean modelling and
data assimilation research that has been, and
continues to be, undertaken within WOCE. For
example, Fukumori et al. (1999) discusses useful
measures of consistency in ocean state estimation,
including formal error estimates, the model–data
residuals, and comparisons with independent data,
in an assimilation of TOPEX/POSEIDON data into
a global ocean circulation model. Such results have
direct applications in present climate prediction
systems.
It should also be noted that as we move to operational implementation of climate prediction models, agencies are seeking systems that are more
general. That is, the ocean component of the climate model may also be used to provide ocean
analyses on a range of time scales. This expansion
is being aided by the experience and research
results gained through WOCE. For example, the
model discussed by Wang et al. (2001) also serves
to provide global ocean analyses and will be the
basis for climate change studies. Such synergy has
two-way benefits, allowing the research on global
issues, such as that of WOCE, to be entrained into
operational systems.
7.4.4.3 Data sets for initialization?
Several earlier chapters (for example, Talley et al.,
Chapter 7.1) have demonstrated the great progress
that has been made in the area of data assimilation. While current climate prediction models rely
for the most part on relatively simple data assimilation systems (e.g. Stockdale et al., 1998), the
trend is towards more complex schemes in order
to exploit better new data sets (e.g. altimetry,
Behringer et al., 1998) and to account for other
effects (e.g. salinity, Vossepoel et al., 1999). In
some cases (e.g. Kleeman et al., 1995; Rienecker
et al., 1999) more sophisticated techniques are being
used, both to understand better the impact of the
data and to provide a more sophisticated merging
of observed and modelled information.
The adoption of more sophisticated assimilation
techniques is also greatly improving the ability of
climate prediction models to blend different data
sets (e.g. subsurface temperature and sea level), to
take account of known errors in the data and to
take better advantage of output statistics. An
7.4 Ocean and Climate Prediction – the WOCE Legacy
593
Smith