232
7 Ocean Currents
that the observing system met many of the data requirements in the equatorial
Pacific Ocean between 8°N and 8
0
S. A detailed description of all instrumentation used during TOGA was given by McPhaden et al. (1998). Fortunately,
TOGA spanned a decade in which there was both a large swing from EI Nino
to La Nina conditions (1991-1995). Thus, the observing system contributed to
scientific progress in modelling studies of short-term climate variability during
these TOGA phases.
An excellent overview of the process of forecasting EI Nino, not only from
the scientific perspective, can be found in Glantz (1996) and Anderson et al.
(1998). Here, we only concentrate on the possibility of predicting EI Nino
from the geophysical point of view. Studies with coupled atmosphere-ocean
models indicate that the interactions between atmosphere and ocean manifest themselves in terms of modes which are inter-annual fluctuations between
warm EI Nino and cold La Nina, and hence correspond to possible Southern
Oscillations. These periodic oscillations are predictable, provided there are no
high-frequency ('weather' type) disturbances which are not correlated with sea
surface temperature changes. The impact of the high frequency components
results in irregular, inter-annual Southern Oscillations.
As the Southern Oscillation is a low frequency event, it will therefore be possible to make predictions by extrapolating this low-frequency trend. The success
of such extrapolation depends on the correlation between various Southern Oscillation indices. For example, it was discovered by Walker that sea surface
temperature in the central equatorial Pacific during the northern summer, and
rainfall variations over India during those months, are highly correlated and can
be used as predictors for subsequent developments (Philander, 1990). Another
example of the correlation of various events is the seasonal pressure anomaly at
Darwin and rainfall in India. If during March, April and May, the anomaly is
above normal and is increasing, then heavy rains over India in June, July and
August are unlikely. Also, when the pressure is below normal and is falling,
droughts are unlikely. Advanced statistical techniques have recently been used
to identify the principal modes of climate variability on different space and
time scales (Latif et al. 1998).
Barnett et al. (1988), applying canonical correlation analysis developed a
method to predict the EI Nino event of 1986 at lead times of 3 to 9 months.
Long-term predictions for periods of a few years indicate that if an EI Nino
occurs in a certain year, then it is highly likely that it will appear again 3 to
4 years later. However, only the TOGA experiment has provided more reliable
experimental data on the atmosphere and ocean circulations, and inspired the
development of a hierarchy of various EI Nino prediction schemes. Besides the
different explanation of nature of the EI Nino, offered by those schemes, it
is now commonly believed that EI Nino is a fundamental oscillatory mode of
the coupled atmosphere-ocean system and the memory of the coupled system
resides in the ocean thermocline state (Latif et al., 1998). According to the
'delayed action oscillator' scenario, the easterly winds over the Western Pacific,
7 Ocean Currents
that the observing system met many of the data requirements in the equatorial
Pacific Ocean between 8°N and 8
0
S. A detailed description of all instrumentation used during TOGA was given by McPhaden et al. (1998). Fortunately,
TOGA spanned a decade in which there was both a large swing from EI Nino
to La Nina conditions (1991-1995). Thus, the observing system contributed to
scientific progress in modelling studies of short-term climate variability during
these TOGA phases.
An excellent overview of the process of forecasting EI Nino, not only from
the scientific perspective, can be found in Glantz (1996) and Anderson et al.
(1998). Here, we only concentrate on the possibility of predicting EI Nino
from the geophysical point of view. Studies with coupled atmosphere-ocean
models indicate that the interactions between atmosphere and ocean manifest themselves in terms of modes which are inter-annual fluctuations between
warm EI Nino and cold La Nina, and hence correspond to possible Southern
Oscillations. These periodic oscillations are predictable, provided there are no
high-frequency ('weather' type) disturbances which are not correlated with sea
surface temperature changes. The impact of the high frequency components
results in irregular, inter-annual Southern Oscillations.
As the Southern Oscillation is a low frequency event, it will therefore be possible to make predictions by extrapolating this low-frequency trend. The success
of such extrapolation depends on the correlation between various Southern Oscillation indices. For example, it was discovered by Walker that sea surface
temperature in the central equatorial Pacific during the northern summer, and
rainfall variations over India during those months, are highly correlated and can
be used as predictors for subsequent developments (Philander, 1990). Another
example of the correlation of various events is the seasonal pressure anomaly at
Darwin and rainfall in India. If during March, April and May, the anomaly is
above normal and is increasing, then heavy rains over India in June, July and
August are unlikely. Also, when the pressure is below normal and is falling,
droughts are unlikely. Advanced statistical techniques have recently been used
to identify the principal modes of climate variability on different space and
time scales (Latif et al. 1998).
Barnett et al. (1988), applying canonical correlation analysis developed a
method to predict the EI Nino event of 1986 at lead times of 3 to 9 months.
Long-term predictions for periods of a few years indicate that if an EI Nino
occurs in a certain year, then it is highly likely that it will appear again 3 to
4 years later. However, only the TOGA experiment has provided more reliable
experimental data on the atmosphere and ocean circulations, and inspired the
development of a hierarchy of various EI Nino prediction schemes. Besides the
different explanation of nature of the EI Nino, offered by those schemes, it
is now commonly believed that EI Nino is a fundamental oscillatory mode of
the coupled atmosphere-ocean system and the memory of the coupled system
resides in the ocean thermocline state (Latif et al., 1998). According to the
'delayed action oscillator' scenario, the easterly winds over the Western Pacific,