¿ìè¶ÌÊÓƵ

AT Weston-super-Mare on Monday, 26 February, council officials were
expecting the sea to surge 24 centimetres on top of its normal tide of 12.9
metres. In fact, the tide was nearly 2 metres above normal, resulting in
widespread floods. The town’s experience was matched in nearby Avonmouth,
where the Meteorological Office’s tidal gauges measured a surge of 1.5 metres,
double the level forecast.

But not everyone was caught by surprise. Towyn, near Colwyn Bay in North
Wales, suffered the worst flooding. There, however, the Met Office says
its predictions were ‘spot on’. It also successfully predicted the highest-ever
tide recorded at Liverpool, at lunch time on 26 February, one which meteorologists
estimate should occur only once in every 250 years. The following day Liverpool
had a once in 200 years tide.

The Met Office has just taken delivery of a Pounds sterling 5-million
Cray computer that should improve Britain’s weather forecasts. But as the
recent floods and storms emphasise, weather prediction is still a long way
from being perfect. Even with the refinements offered by faster computing,
uncertainty will remain; storms may still slip through the forecasters’
nets, and tidal surges will again surprise those who live by the sea.

The discrepancies experienced last month were partly the result of the
forecasters’ computer model of the tides, since this was primarily designed
to predict floods on the east coast of Britain. This emphasis reflects a
desire to prevent a repetition of the disastrous floods of 1953 in which
more than 300 people died in Britain alone.

Relatively simple principles underpin the model. It relates variation
in tides to factors such as barometric pressures and winds. The forecasters
calibrate the model with figures from tide gauges out in the ocean, and
then use it to turn any weather forecast into a forecast of tides.

The chief problem is that the computer model works by dividing the ocean
into a grid of ‘boxes’ 35 kilometres across. This is adequate for predicting
tides in the North Sea. But it is far less successful in places such as
the English Channel and the Bristol Channel, where tides vary on a scale
much smaller than anything such 35-kilometre boxes could measure. The coastline
is more irregular, and consequently the currents vary locally; the Bristol
Channel, for example, is not even 35 kilometres wide.

The Met Office in Britain is ultimately responsible for all flood alerts
(although the National Rivers Authority uses its own computer models to
produce the local figures issued to councils such as that at Weston-super-Mare).
The Met Office hopes to improve its tidal forecasts after its Cray supercomputer
comes into service later this year. The forecasters say that the computing
power of the Cray will enable them to reduce the size of the grids to around
8 kilometres. Another option would reduce the grid to around 12 km, with
grids of 4 km or even 1 km in particularly difficult locations. This extra
detail could mean more precise forecasts of the complex tides around Britain’s
intricate coastline.

The Met Office’s first task is to convert its software for the general
weather forecast to work on the Cray. Trials forecasts will begin on 1 April.
‘Our initial task is to see if it gives good forecasts,’ says Mike Cullen,
head of the design team working on the transfer of software to the Cray.
Medium-term forecasts are planned to cover six days; the greater computer
power can cope with 60 000 points spanning the world on a 90 km grid. The
grid for the short-term forecast will similarly be assembled with 30 000
points over a 50 km grid.

The great advantage of the Cray is its ability to improve the analysis
of data. On a typical day the Met Office receives some 12 000 observations.
The computer works in two stages. First, it analyses the data as it arrives,
day and night, to build up a model weather map. Secondly, it uses this data
map to run the model forward in time to make the forecast. The superior
computing power of the Cray will enable the forecasters to narrow the model’s
resolution as well as improving the analysis, for example by checking for
errors.

The new computer should be producing its first forecasts later this
year. But the Cray will still depend heavily on the observations, and not
all the data are of good quality. Satellites, for example, are unable to
provide good vertical resolution of weather patterns.

Cullen says that the new machine will help forecasters with specific
types of weather: ‘Rain is so intermittent, but cloud will be better represented.
We hope for a noticeable improvement in forecasts of rain or snow.’ Even
so, the system will still be vulnerable to storms that develop suddenly,
such as that of October 1987, or to relatively small weather patterns which
slip through the grid.

The storm that hit London and the Southeast in October 1987 highlighted
the Met Office’s biggest prediction failure in recent years. The storm broke
before anyone in Britain was prepared for it, although other countries in
Europe were warned. Forecasters missed the early signs of this storm out
at sea, partly because of cuts in monitoring capacity. In the early 1970s
there were eight weather ships in the North Atlantic, financed by a consortium
of European countries; these had been cut back to two the year before the
storm to save money.

Another reason for the failure was that satellites had largely replaced
the kind of observations formerly carried out by passenger liners. Satellites
measure wind speed by looking at clouds, but the current generation of instruments
cannot measure wind speeds under cloud cover – a significant drawback since
most violent storms have plenty of cloud cover.

The forecasters need a three-dimensional view of the weather. This comes
from releasing weather balloons twice a day into the atmosphere. Since the
1987 storm, the Met Office has arranged for 13 merchant ships to release
weather balloons. Although this information is limited, depending on the
ship’s position, it still helps to improve the information being fed into
the models.

The success that the Met Office had in forecasting the weather on 25
January – if not all last month’s floods – was the result of three factors.
First, the merchant ships improved the quality of information fed into the
computer. Secondly, the storm did not develop suddenly, and forecasters
were able to track its development for several days. Thirdly, the storm
was much more widespread than that of 1987, in which damage was limited
to southeast England.

The next improvement in data collection will not come until the launch
in 1992 of the new European Remote-Sensing satellite, known as ERS I. The
satellite will be a test-bed for new instruments that can beam microwaves
through clouds to measure the roughness of the sea’s surface, and hence
deduce wind speed. The instruments will not, however, help forecasters to
measure the wind direction.

Meanwhile in the US, equipment that the National Weather Service plans
to use in the next few years should produce data on wind velocity and other
factors routinely. The service is working towards better and more detailed
forecasting through a substantial investment in new technology. In future,
it will use a battery of instruments to make observations automatically,
feeding the information into computers which will allow meteorologists to
monitor the reams of constantly changing data as the weather evolves.

The two countries have quite different weather patterns. In Britain,
the weather pattern is usually determined by a series of depressions sweeping
in from over the Atlantic.

Typically, these depressions are large, more than 1000 kilometres wide,
and about 100 of them cross Britain every year. The US suffers from much
more violent winds and storms, such as the hurricanes that sweep in from
the east.

Frank Singleton, of the Met Office, says that the American approach
to forecasting is not really appropriate for Britain’s weather, because
‘we don’t have the same sort of severe storms in this country’. It would
not be cost effective, he says. But if Europe’s climate changes, our weather
could become more extreme.

¿ìè¶ÌÊÓƵs are divided about whether the recent spate of storms in Britain
are a manifestation of the greenhouse effect. Mick Kelly, of the Climatic
Research Unit at East Anglia, says the storms may reflect global warming.
Other scientists say that Britain always had spells of windy weather. According
to the Met Office, there is no reason to think these storms fall outside
the normal run of weather. But if Britain does suffer further spells of
violent weather, we may need to look more closely at the techniques employed
in the US to collect information, particularly data relating to severe storms.

After the modernisation of the US’s National Weather Service, there
will be 115 weather forecast offices across the country, each receiving
data from a satellite in geostationary orbit, together with conventional
data collection. There will also be more than 1000 automated observing systems
and a network of 160 stations emitting radar pulses which will detect weather
patterns using the doppler principle.

New technologies

The limited nature of the information that satellites can deliver has
helped to spur forecasters on in seeking better ways to observe the weather.
Hugo, the hurricane that devastated the Caribbean and the southeastern seaboard
of the US last autumn, was an example of the surprises in store even with
satellite monitoring, according to Joe Golden, of the National Oceanic and
Atmospheric Administration in Washington DC. When one of NOAA’s planes flew
into this hurricane, the winds in the wall of cloud around the eye of the
storm were stronger than the satellite images indicated – so strong that
one of the plane’s engines stopped. The pilot had to circle in the eye for
45 minutes until the engine could be restarted.

The crew escaped, with a record of wind speed that was all the more
valuable for its rarity. Elsewhere, recording machinery suffered heavily.
‘There are no anemometer records left from the Caribbean,’ says Golden,
‘and only two in Puerto Rico. The only information we have is from a man
who stayed on his boat during the storm and saw his anemometer reading 150
°ì²Ô´Ç³Ù²õ.’

Golden emphasises that forecasting the effects of storms such as Hugo
will remain difficult while the grid of observations remains as sparse as
it is at present. He is emphatic that automatic measurements of the weather
will benefit the forecasters.

As an example, he cites the wind profiler, an instrument being developed
by NOAA to measure wind velocity at different levels throughout the lower
20 kilometres of the atmosphere. There will be 30 profilers, each consisting
of three upward-pointing doppler radar aerials. Golden, who says the profilers
will take over from two weather balloons released each day at sites about
500 kilometres apart, believes they ‘can only increase our ability to anticipate
²õ³Ù´Ç°ù³¾²õ’.

NOAA has been developing weather technology for more than 10 years.
The new network of pulsed doppler radar instruments, known as NEXRAD (which
stands for ‘next generation radar’), will be set up at civil and military
airports and some weather stations across the country at a cost of over
a billion dollars.

The primary aim of NEXRAD is to make flying safer. To do this, the equipment
measures wind speed and direction, using the doppler shift of a burst of
radiation reflected back from moving air at a particular point. In addition,
the strength of the signal reflected shows whether or not it is raining,
snowing or whether hail is falling at each point. In addition, turbulent
air gives back a stronger signal.

A further advance over standard weather radars is the resolution that
the new system can achieve, using a beam of radiation only 1 degree across.
Each instrument has a range of about 200 kilometres, and a resolution of
a few kilometres.

Although England and Wales are covered by a network of radar stations,
the instruments are not equipped to use the doppler effect to find the wind
direction. This is deliberate. A study carried out between 1982 and 1983
by the Met Office advised that Britain’s weather pattern did not merit such
detailed analysis; concentration on problems such as flood warnings would
bring most benefits. New radar equipment to be set up in Scotland next year
will be able to work as a doppler system, if required, although its primary
purpose will be to complete the radar network across Britain.

The Met Office has a doppler system on test at Cowes on the Isle of
Wight. According to Singleton: ‘The thinking at the moment is that doppler
radar in the UK won’t have the same applications as in the US. It will give
more signal, and more sensitivity, and identify and get rid of misleading
echoes within the system.’

In the American system, each radar station is automatic, controlled
by computer, and works in real time, allowing meteorologists to monitor
storms as they evolve, as Golden describes. ‘We can see on our screens the
mesocyclones (intermediate-size cyclones) in storms that lead to tornadoes
about 20 minutes or half-an-hour before they touch down.’ As many towns
in the Midwest have tornado alert systems, such as sirens, 20 minutes’ warning
could be enough.

The American system is focused on fore casting the severe storms and
hurricanes that have done so much damage each year. Britain has considered
that such capability is not needed here. If the greenhouse effect does mean
further unusual weather patterns in Europe, things may have to change.