7.6 El Nino and Southern Oscillation
233
prevailing during the cold phase of the El Nino, force an upwelling which, in
the form of a Kelvin wave packet, propagates eastward along the Equator. This
causes cooling at the sea surface in the Eastern Pacific and shallowing of the
thermocline. At the same time, the ocean response in the west to the easterly
winds induces a downwelling Rossby wave packet which propagates westward
(for a description of Rossby waves see Sect. 3.9). The Rossby wave is reflected
at the western boundary and propagates eastward in the form of downwelling
Kelvin waves, which affects the SST. The delay of propagation of the Rossby
waves from the wind stress region to the western boundary, and the return
as a reflected Kelvin waves to the eastern basin, provide the memory for the
oscillation. A positive SST anomaly develops, growing due to unstable air-sea
interactions in the warm phase of El Nino. Thereafter the sequence of events
repeats itself but with a reversed sign, every 3 to 5 years. Thus, in this sense
El Nino is a low-frequency basin-wide mode of oscillation (Neelin et al. 1998).
The hierarchy of El Nino prediction models also includes more sophisticated
coupled ocean-atmosphere models. The simplest coupled models designed for
predictions of an El Nino event are the limited area models. Zebiak and Cane
(1987) developed a coupled atmosphere-ocean model to reproduce certain key
features of the observed EI Nino phenomenon. The model calculates perturbations about the climatological mean state that are specified from observations.
In particular, the model produces recurring warm events that are irregular in
both amplitude and spacing, but favour a 3-4 year period. The predicted warm
events, which include equatorial westerly wind anomalies in the Central Pacific
and large SST anomalies in the Eastern Pacific, are in general agreement with
observations. Zebiak and Cane used their model to successfully forecast the
1986-87 El Nino.
Another class of ocean-atmosphere models consists of physical ocean models
coupled to empirically derived atmosphere models. They are known as hybrid
coupled models. The hybrid coupled models generally yield useful El Nino
predictions for lead times up to about 1 year. Thus, the anomaly correlation
coefficient between observed and predicted SST anomalies, averaged over the
EI Nino 3 region (see above) yields values above 0.5 at lead times of 1 year.
The most complex models applied to EI Nino forecasting are coupled oceanatmosphere general circulation models. Their detailed description is beyond the
scope of this book and the reader should consult Anderson et al. (1998) for an
in-depth review. All these models are still idealized and have large numbers of
adjustable parameters. The efforts to improve the models and collect necessary
experimental evidence are continuing (see for example Godfrey et al. 1998) and
a Coupled General Circulation Model (CGCM) of the ocean and atmosphere
will probably be available before long, capable to anticipate future El Nino
events.
However, at present we are in a better position to forecast El Nino event
some time in advance. This was the case for the 1997 El Nino. In December
1996, normally westward blowing Trade Winds briefly reversed direction in
233
prevailing during the cold phase of the El Nino, force an upwelling which, in
the form of a Kelvin wave packet, propagates eastward along the Equator. This
causes cooling at the sea surface in the Eastern Pacific and shallowing of the
thermocline. At the same time, the ocean response in the west to the easterly
winds induces a downwelling Rossby wave packet which propagates westward
(for a description of Rossby waves see Sect. 3.9). The Rossby wave is reflected
at the western boundary and propagates eastward in the form of downwelling
Kelvin waves, which affects the SST. The delay of propagation of the Rossby
waves from the wind stress region to the western boundary, and the return
as a reflected Kelvin waves to the eastern basin, provide the memory for the
oscillation. A positive SST anomaly develops, growing due to unstable air-sea
interactions in the warm phase of El Nino. Thereafter the sequence of events
repeats itself but with a reversed sign, every 3 to 5 years. Thus, in this sense
El Nino is a low-frequency basin-wide mode of oscillation (Neelin et al. 1998).
The hierarchy of El Nino prediction models also includes more sophisticated
coupled ocean-atmosphere models. The simplest coupled models designed for
predictions of an El Nino event are the limited area models. Zebiak and Cane
(1987) developed a coupled atmosphere-ocean model to reproduce certain key
features of the observed EI Nino phenomenon. The model calculates perturbations about the climatological mean state that are specified from observations.
In particular, the model produces recurring warm events that are irregular in
both amplitude and spacing, but favour a 3-4 year period. The predicted warm
events, which include equatorial westerly wind anomalies in the Central Pacific
and large SST anomalies in the Eastern Pacific, are in general agreement with
observations. Zebiak and Cane used their model to successfully forecast the
1986-87 El Nino.
Another class of ocean-atmosphere models consists of physical ocean models
coupled to empirically derived atmosphere models. They are known as hybrid
coupled models. The hybrid coupled models generally yield useful El Nino
predictions for lead times up to about 1 year. Thus, the anomaly correlation
coefficient between observed and predicted SST anomalies, averaged over the
EI Nino 3 region (see above) yields values above 0.5 at lead times of 1 year.
The most complex models applied to EI Nino forecasting are coupled oceanatmosphere general circulation models. Their detailed description is beyond the
scope of this book and the reader should consult Anderson et al. (1998) for an
in-depth review. All these models are still idealized and have large numbers of
adjustable parameters. The efforts to improve the models and collect necessary
experimental evidence are continuing (see for example Godfrey et al. 1998) and
a Coupled General Circulation Model (CGCM) of the ocean and atmosphere
will probably be available before long, capable to anticipate future El Nino
events.
However, at present we are in a better position to forecast El Nino event
some time in advance. This was the case for the 1997 El Nino. In December
1996, normally westward blowing Trade Winds briefly reversed direction in