Weather forecasters: sentinels against natural disaster

On October 15, 1987, Britain’s TV weather forecasts predicted strong winds, but nothing more. The BBC’s TV weather forecaster, commenting on a viewer’s report of a hurricane on the way, said: `Don’t worry, there isn’t.’

That night, ‘Black Friday’, southern England was struck by the storm of the century. Winds gusting to 115mph (185km/h) tore down 15 million trees and caused 19 deaths and over £1000 million of damage. In the public outcry that followed, a major question recurred: Why was there no accurate warning?

The answer was simple: the evidence was not overwhelming, and the forecasters made the wrong judgment. Despite increasing accuracy, weather forecasting is an uncertain science, and always will be.

Growth of a science

For centuries, even local weather was often baffling in its unpredictability. People could only pray, or devise reassuring proverbs based on observation, like:

Red sky at night, shepherd’s delight:

Red sky in the morning, shepherd’s warning.

Local forecasting took a step forward in 1643 when the Italian physicist Evangelista Torricelli invented the barometer to measure the pressure of the air. It was soon noticed that rises and falls in air pressure corresponded to changes in weather, and that a fall often heralded a storm.

But it was only after the invention of the telegraph in the 1840s that it became possible to gather reports from widespread weather stations, so that imminent changes could be predicted with reasonable accuracy. Radio, in the early 20th century, provided another major step forward. In the 1960s, huge technical advances in gathering information and analysing data in computers suggested that meteorology might eventually become an exact science, capable of predicting weather weeks or months ahead.

The amount of information now available to forecasters is staggering. The World Meteorological Organisation receives reports from 9000 outposts and 7500 ships. Manned stations take measurements several times a day, sometimes hourly, under standard conditions (for instance, wind speed is measured at 33ft (10m) from the ground).

In addition, weather balloons released from 950 stations all around the world monitor the atmosphere up to a height of 18 miles (30km). Some 600 aircraft report every day from high over the oceans. Seven weather satellites scan the Earth from space, monitoring the atmosphere to a height of 50 miles (80km).

From all these points, a huge quantity of data – wind speed, wind direction, temperature, cloud cover, rainfall, humidity, air pressure – is amassed. Every day, observations produce 80 million binary digits of computer information – equivalent to the contents of several thousand books. This is fed into a network of 17 stations around the world that together form the Global Telecommunications System. Two centres – the USA’s National Meteorological Centre in Washington and Britain’s Meteorological Office in Bracknell, Berkshire – are World Area Forecast Centres for civil aviation, duplicating each other’s operations in case of breakdown. Computers able to perform up to 3500 million calculations a second process the measurements to produce predictions.

Knowing tomorrow’s weather today is vital to life in the industrialised West. In air traffic control alone, global forecasts that allow airlines to take tail winds into account or re-schedule landings to avoid poor conditions save an estimated £50 million a year in fuel. Whole industries like construction, shipping and agriculture depend crucially on hourly and daily forecasts.

The most dramatic events to test the forecasters are tropical cyclones – huge circular storms that are born over tropical seas. Those that migrate westwards across the Atlantic are called hurricanes, those in the Pacific typhoons. They swing away from the Equator, and die out as they move over land. Hurricanes usually last about a week, and are powered by the warm, moist air above the tropical ocean. As it rises in the centre of the storm, the moisture in the air condenses into clouds, releasing heat and sucking more moist air into the system. Hurricanes usually die down when they reach land, because they become starved of moisture. During the hurricane season, June-November, more than 100 storms form off the African coast, of which about six become hurricanes.

When the typical swirling clouds of a tropical storm are spotted, usually by satellite, the USA’s National Hurricane Centre in Miami swings into action. Its staff sift a mass of data from satellites, radar systems, automated buoys and aircraft to predict its path – in particular where it will come ashore.

In early September, 1988, a trough of low pressure off the African coast grew steadily until, on Saturday, September 10, when it was over the eastern Caribbean, it was designated a hurricane, and named Gilbert. Two days later, Gilbert struck Jamaica with devastating force. Beneath slate-blue skies winds wrecked the island, making one-fifth of the 2.5 million inhabitants homeless, and destroying almost all the crops on which the economy depended – bananas, coconuts, coffee, sugar and vegetables. Prime Minister Edward Seaga called it the worst natural disaster in our modern history’.

Airborne monitor Weather balloons carry aloft radiosondes – packages of instruments that record humidity, atmospheric pressure and temperature. The balloons are released regularly by 950 stations throughout the world.

Then, as Gilbert spun away from the stricken island, it nearly doubled in force, producing wind speeds of up to 175mph (280km/h) – the most powerful storm to hit the Western Hemisphere this century. Its course carefully predicted, Gilbert smashed into Mexico’s Yucatan peninsula at dawn on Wednesday, leaving 30,000 homeless. It could have been far worse – in 1979, Hurricane David killed 1100 and Flora caused 7200 deaths in 1963. Gilbert’s relatively low death-toll of about 300 people was a tribute to the benefits of good weather forecasting.

But the forecasters still could not say exactly what would happen. As Gilbert swung north, the coasts of Texas, Louisiana and Mississippi were put on the alert. People emptied supermarket shelves in a panic of buying and 100,000 of them clogged roads as they fled inland, leaving be hind them houses shuttered and boarded.

In the event, the warnings proved unnecessary. When Gilbert hit the US mainland, it was already dying. It brought strong winds, high tides and heavy rain, but little destruction. There were no further deaths.

Gilbert’s unexpected demise highlights the major problem with weather forecasting: its lack of absolute certainty. Notwithstanding their expensive computers and worldwide resources, forecasters deal only in probabilities.

Weather systems are unpredictable in detail. The figures used to describe varying factors like wind speed and temperature are at best for one moment only. The next second, the same figure becomes an approximation. However small the deviations are from the true values, prediction and reality soon part company.

Scientists accept that tiny events can have huge consequences. They jokingly refer to this unpleasant fact as the Butterfly Effect  the notion that a butterfly flapping its wings in Peking, for example, can affect storm systems in New York. As a result, the current limit of useful forecasting is no more than a few days.

Often a forecaster’s experience of the real world is a better guide to the immediate future than any computer model. For instance, if air is moving from the cool North Sea to adjacent European countries, it can build cloud in a thin layer that may either bring rain inland the next day, or evaporate in the heat of the sun. The outcome may depend on a temperature difference of only a few tenths of a degree. But the effects can be dramatically different, producing either a cool, cloudy day or a hot, sunny one.

Even with the best computers and the most efficient information gathering, it is unlikely that forecasts will ever be accurate more than two weeks ahead.

Medium-range forecasts have improved with technical innovations. Three-day forecasts for Europe, produced in the European Centre for Medium-range Weather Forecasts in Reading, Berkshire, are now as accurate as one-day forecasts a decade ago. But long-range forecasting (beyond ten days), on the other hand, has not proved reliable.

There is hope, of a sort. Scientists believe there is a relationship between changing sea temperatures and certain weather conditions. For instance, every three to seven years at Christmas time, a warm current known as El Nino, extends down through the chilly waters off the west coast of South America. Besides having serious consequences for the weather, wildlife and industries locally, El Nino also causes either milder or colder winters in the United States. No one yet knows why, but perhaps one day El Niño’s effect will become predictable.

Detecting instruments feed data to a computer that builds up cloud images in colour-code on a TV monitor.