In the winter of 1997/1998 the world experienced another El
Niņo event, a disruption to global weather that occurs every few years.
This El Niņo was notable for two reasons. First, it was the strongest
this century and second, scientists accurately predicted its effects
many months in advance. This was possible because of a highly sophisticated
computer model developed with the help of EU-funding by a team involved
in the PROVOST project. The model represents a marriage between a
10-day weather forecasting model developed by the European Centre
for Medium-Range Weather Forecasts (ECMWF) and an ocean model developed
by a German team of oceanographers.
The 97/98 El Niņo caused not only serious
disastrous flooding in America, Africa and Asia, but also
led to severe droughts and bush fires in Indonesia (our picture)
and Brazil. All these climatic regional perturbations were
in fact well predicted by the medium-term weather forecasting
models of the European meteorologists involved in the PROVOST
Usually, water in the western Pacific Ocean
is warm at the equator. As you move east towards the American continent,
the ocean gets much colder because it is continually suffused by upwellings
of deep, cool water. However, every two to seven years, the build-up
of heat energy in the western ocean causes warm water to drift eastwards
and strong westward-blowing trade winds to subside. This significant
change in climate, which has been given the odd nickname of "El Niņo"
(the "Little Boy") has serious consequences for many parts of the
world. The 97/98 El Niņo caused disastrous flooding in Africa, in
China, and all along the Pacific coast of the American continent.
It also led to severe droughts and bush fires in Indonesia and Brazil.
Europe escaped the more violent consequences of El Niņo, but we still
saw some pretty unusual weather.
All of this was predicted by a team of scientists at ECMWF, the European
Centre for Medium-Range Weather Forecasts based at Reading in the
UK. The Centre is funded by the national weather services of 18 European
countries. Meteorologists in Europe decided long ago to pool resources
to improve short- and medium-term weather forecasting, sharing the
very high cost of the necessary computer hardware. The EU-funded PROVOST
project has been contributing to the work of the ECMWF since 1996.
Making forecasts possible
The task of developing a computer model capable of long-term forecasts
began in earnest with TOGA. This major research programme was co-ordinated
by the United Nations and ran from the mid-1980s until the mid-1990s.
It was important for the further development that has taken place
as part of PROVOST for three reasons.
- It demonstrated that there was a scientific basis for seasonal
forecasts that could predict the effects of an El Niņo event some
months before they were felt.
- It promoted acceptance that El Niņo has a global effect on the
- It set in place the TOGA-TAO array - a network of moored buoys
in the ocean that constantly feed back information about the physical
conditions in the Pacific Ocean. The temperature of the ocean
is measured every day to a depth of 500m at each ATLAS buoy mooring
in the TOGA-TAO observing array in the Pacific and transmitted
via satellite for distribution on the Global Telecommunications
System. In addition, wind velocity and humidity are measured and
transmitted three times per day. Buoys remain in position for
up to a year before being recovered and refurbished.
Scientists at ECMWF became convinced that better forecasts of the
effects of El Niņo would be possible if they were to couple a weather
prediction model to an ocean model. The Centre already had a fully
operational atmospheric model that could make accurate 10-day global
forecasts, but they had no ocean model for the Pacific. They knew
that a team at the Max-Planck Institute for Meteorology in Hamburg
had built the Hamburg Ocean Primitive Equation model (HOPE). This
system had everything they needed and the two groups quickly established
collaboration. Very shortly afterwards, a call was made for projects
by the Environment and Climate programme (within the EU's Fourth
Framework Programme). A consortium coordinated by the ECMWF submitted
a proposal for a research project based on this cooperation and
including other European research groups working on the same subject,
which was accepted for EU support. PROVOST (PRediction Of climate
Variations On Seasonal Timescales) began in 1996.
Predicting the effects of the 1997/98
The blue areas indicate areas of the world for which wetter
weather than usual was forecast; the yellow/orange areas indicate
forecasts of drier weather than usual. The deeper the colour,
the more severe the effect. The floods in China and along
the Pacific Coast from the US to South America, the very dry
conditions in Indonesia and Brasil, as well as the warm winter
over Europe were all clearly predicted.
A fully operational model
A version of the HOPE model was coupled to the ECMWF atmospheric
model and, since then, much of the PROVOST project has been dedicated
to testing the atmospheric component of the model using data available
from previous El Niņo years. The raw data was fed into the model
and the predictions that it made were compared to the actual global
weather patterns that are a matter of record. "We call these tests
hindcasts, rather than forecasts, and they are tremendously revealing,"
comments Tim Palmer. "We have had to make hundreds of runs to fine
tune the model and to find out the limits of its accuracy but we
are now running it in real time."
Two years into the project, things are going well. The model is
fully developed and the recent strong El Niņo gave it the chance
to impress. The predictions of world weather made last autumn have
turned out to be very close to reality (see map). But, as Tim Palmer
points out, "This is a good result, but it could be a mixed blessing.
The ocean signals that allow us to predict the effects of are much
easier to distinguish from the background 'noise' in the atmosphere
when the El Niņo event is a strong one. In the future, the model
will face the tougher test of weaker El Niņo events, and forecasts
will not be as easy to make. We still have a lot of work to do."