308 Martin Fischer
the initial state from which the coupled model prediction is started. The quality of
the wind stress field over the whole initialization period of about one to two years
is crucial for the quality of the initial state, and thus for the forecast.
Assimilating oceanic observations during the initialization phase yields an additional source of information, and may thus compensate for errors and/or uncertainties in the forcing data or even for model errors. Over the last few years several
groups started to develop integrated data assimilation/coupled general circulation
model forecast systems and investigated the impact of different types of ocean
measurements (Ji et al. 1997, Rosati et al. 1996, Fischer et al. 1997).
The paper is organized as follows. In section 16.2 a description of ENSO and the
impact of ENSO on climate and economy is given, and the possible predictability
of ENSO is discussed. In section 16.3 different types of ENSO prediction models
are presented and in section 16.4 we discuss several initialization methods and
present results from different groups. Especially the assimilation of oceanic surf ace
observations like sea surface height and sea surface temperature is discussed in
more detail. This is an important issue, since these data can be measured with satellite based instruments which provide a complete and regular coverage in space and
time. In section 16.5 the interpretation of forecast results is addressed and examples for skill measurements and corrections of systematic errors are given. The
paper concludes with a summary in section 16.6.
16.2 The El Niiio / Southern Oscillation Phenomenon
16.2.1 Description
The El Nifio/Southem Oscillation phenomenon is the strongest climate signal on
time scales of a few months up to several years. It is characterized by a weakening
of the trade winds in the equatorial Pacific and a huge redistribution of heat from
the westem to the eastem equatorial Pacific.
In Fig. 16.1 the Pacific ocean is presented schematically in two different states.
During normal conditions, relatively warm surface water is driven from the east
Pacific to the west Pacific by the trade winds. Thus an east-west pressure gradient
is built up, which is in the depth balanced by a west-east tilt ofthe thermocline. The
depth of the thermocline is about 150 to 200 meters in the west Pacific and only
about 30 to 50 meters in the east Pacific. Due to the westerly flow of surf ace water
cold subsurface water upwells at the equator, and thus leads to the so called cold
tongue region in the eastem equatorial Pacific. The temperature difference between
the warm pool and the cold tongue region can reach values up to about six degrees
centigrade. Under these conditions the coupled air-sea system is in balance (Fig.
16.1a.
If, for some reason, the trade winds get weaker, or the temperature maximum of
the warm pool is shifted to the east, the whole system may get out ofbalance. Due
to the shift of warm water further to the east, the center of the convective cell is
also shifted to the east and thus decreases the strength ofthe trade winds. This leads
the initial state from which the coupled model prediction is started. The quality of
the wind stress field over the whole initialization period of about one to two years
is crucial for the quality of the initial state, and thus for the forecast.
Assimilating oceanic observations during the initialization phase yields an additional source of information, and may thus compensate for errors and/or uncertainties in the forcing data or even for model errors. Over the last few years several
groups started to develop integrated data assimilation/coupled general circulation
model forecast systems and investigated the impact of different types of ocean
measurements (Ji et al. 1997, Rosati et al. 1996, Fischer et al. 1997).
The paper is organized as follows. In section 16.2 a description of ENSO and the
impact of ENSO on climate and economy is given, and the possible predictability
of ENSO is discussed. In section 16.3 different types of ENSO prediction models
are presented and in section 16.4 we discuss several initialization methods and
present results from different groups. Especially the assimilation of oceanic surf ace
observations like sea surface height and sea surface temperature is discussed in
more detail. This is an important issue, since these data can be measured with satellite based instruments which provide a complete and regular coverage in space and
time. In section 16.5 the interpretation of forecast results is addressed and examples for skill measurements and corrections of systematic errors are given. The
paper concludes with a summary in section 16.6.
16.2 The El Niiio / Southern Oscillation Phenomenon
16.2.1 Description
The El Nifio/Southem Oscillation phenomenon is the strongest climate signal on
time scales of a few months up to several years. It is characterized by a weakening
of the trade winds in the equatorial Pacific and a huge redistribution of heat from
the westem to the eastem equatorial Pacific.
In Fig. 16.1 the Pacific ocean is presented schematically in two different states.
During normal conditions, relatively warm surface water is driven from the east
Pacific to the west Pacific by the trade winds. Thus an east-west pressure gradient
is built up, which is in the depth balanced by a west-east tilt ofthe thermocline. The
depth of the thermocline is about 150 to 200 meters in the west Pacific and only
about 30 to 50 meters in the east Pacific. Due to the westerly flow of surf ace water
cold subsurface water upwells at the equator, and thus leads to the so called cold
tongue region in the eastem equatorial Pacific. The temperature difference between
the warm pool and the cold tongue region can reach values up to about six degrees
centigrade. Under these conditions the coupled air-sea system is in balance (Fig.
16.1a.
If, for some reason, the trade winds get weaker, or the temperature maximum of
the warm pool is shifted to the east, the whole system may get out ofbalance. Due
to the shift of warm water further to the east, the center of the convective cell is
also shifted to the east and thus decreases the strength ofthe trade winds. This leads