16 ENSO Predictions with Coupled Ocean
Atmosphere Models
MARTIN FISCHER
Consiglio Nazionale delle Ricerche, Istituto per 10 studio delle Metodologie
Geofisiche Ambientali
16.1 Introduction
The strongest signal on the short range climatic time scale, ranging from a few
months to several years, is the El Nifio/Southern Oscillation Phenomenon (ENSO).
It is characterized by a weakening of the trade winds along the equator and a huge
redistribution of heat from the western to the eastern tropical Pacific. The impacts
ofENSO are felt worldwide through a disruption ofthe atmospheric general circulation pattern (Ropelewski and Halpert 1987), which leads, for instance, to severe
droughts in Northwest Australia and South East Asia during an El Nifio event. The
Northeast region of BraziI and Zimbabwe are other regions where rainfall variations are highly correlated with ENSO indices and there are many other examples.
Off the Peruvian coast, the marine ecosystem is directly affected by the oceanic
variations in upwelling, which has a large impact on the fishery industry. For alI
these regions, reliable ENSO forecasts would offer decision-makers an opportunity
to take account of anticipated climate variations, in order to reduce impacts of
ENSO on the economy. Thus, it is a logical consequence that the study of ENSO
predictability has become a field of major research.
ENSO can be regarded as a slow oscillation of the coupled ocean atmosphere
system. According to the theory of the delayed aetion oscillator (Schopf and
Suarez 1988) the evolution ofwarm and cold events is determined by the propagation of equatorial waves and their reflection at the eastern and western boundaries,
which implies that ENSO is inherently predictable on time scales up to a few years.
Since ENSO is a phenomenon ofthe coupled system, coupled ocean-atmosphere
models are required for its prediction. Such models have been developed over the
last 10 years by different groups and applied to ENSO forecasts (e.g. Cane et al.
1986, Barnett et al. 1988, Leetmaa and Ji 1989, Latif et al. 1994). ENSO forecasting is an initial value problem, i.e. the further evolution of the system depends
highly on the initial state from which it started. Although ENSO is a phenomenon
of the coupled system, its evolution is determined mainly by the ocean. Therefore
the initialization can be considered as an initial value problem for the ocean. Thus,
a very important task in carrying out an ENSO forecast is to determine the oceanic
initial state as accurately as possible.
The usual way to initialize an ENSO forecast with a coupled ocean-atmosphere
model is as follows. The oceanic model is integrated in an uncoupled mode and
forced by observed wind stresses, heat fluxes, and fresh water fluxes which yields
Atmosphere Models
MARTIN FISCHER
Consiglio Nazionale delle Ricerche, Istituto per 10 studio delle Metodologie
Geofisiche Ambientali
16.1 Introduction
The strongest signal on the short range climatic time scale, ranging from a few
months to several years, is the El Nifio/Southern Oscillation Phenomenon (ENSO).
It is characterized by a weakening of the trade winds along the equator and a huge
redistribution of heat from the western to the eastern tropical Pacific. The impacts
ofENSO are felt worldwide through a disruption ofthe atmospheric general circulation pattern (Ropelewski and Halpert 1987), which leads, for instance, to severe
droughts in Northwest Australia and South East Asia during an El Nifio event. The
Northeast region of BraziI and Zimbabwe are other regions where rainfall variations are highly correlated with ENSO indices and there are many other examples.
Off the Peruvian coast, the marine ecosystem is directly affected by the oceanic
variations in upwelling, which has a large impact on the fishery industry. For alI
these regions, reliable ENSO forecasts would offer decision-makers an opportunity
to take account of anticipated climate variations, in order to reduce impacts of
ENSO on the economy. Thus, it is a logical consequence that the study of ENSO
predictability has become a field of major research.
ENSO can be regarded as a slow oscillation of the coupled ocean atmosphere
system. According to the theory of the delayed aetion oscillator (Schopf and
Suarez 1988) the evolution ofwarm and cold events is determined by the propagation of equatorial waves and their reflection at the eastern and western boundaries,
which implies that ENSO is inherently predictable on time scales up to a few years.
Since ENSO is a phenomenon ofthe coupled system, coupled ocean-atmosphere
models are required for its prediction. Such models have been developed over the
last 10 years by different groups and applied to ENSO forecasts (e.g. Cane et al.
1986, Barnett et al. 1988, Leetmaa and Ji 1989, Latif et al. 1994). ENSO forecasting is an initial value problem, i.e. the further evolution of the system depends
highly on the initial state from which it started. Although ENSO is a phenomenon
of the coupled system, its evolution is determined mainly by the ocean. Therefore
the initialization can be considered as an initial value problem for the ocean. Thus,
a very important task in carrying out an ENSO forecast is to determine the oceanic
initial state as accurately as possible.
The usual way to initialize an ENSO forecast with a coupled ocean-atmosphere
model is as follows. The oceanic model is integrated in an uncoupled mode and
forced by observed wind stresses, heat fluxes, and fresh water fluxes which yields