January 2000 had already been an especially snowy month for the residents of Raleigh, North Carolina. Accustomed to an average yearly snowfall of 2 to 4 inches, the city had already seen an accumulation of more than 3 inches of snow by the twenty-third of January. That afternoon, the forecast called for an additional 1 to 2 inches that evening.
By midnight, with snow falling steadily, the forecast was 4 to 6 inches by morning, but after dawn residents of Raleigh awoke to find a startling 20 inches of powdery frozen precipitation on the ground. With thousands of people trapped in their homes and even interstate highways impassable, the governor declared a state of emergency. The North Carolina Department of Transportation’s small fleet of snowplows was woefully inadequate to remove the record accumulation, which shattered a 107-year-old record.
How could a forecast go so spectacularly wrong, especially with the powerful computers and technology now available? There are some very good reasons, but they are complex. Fully understanding them requires an appreciation of how the art and science of weather forecasting has evolved and why it’s still an inexact science.
Figuring out what the weather is going to do next has always been a challenge, and before there were reliable instruments to predict future conditions, weather watchers relied on nature for clues. Changes in animals and plants, as well as signs in the sky, were often used to predict coming weather, and very often this folksy weather wisdom had some basis in fact.
When early American settlers noticed that birds were going to roost early, they knew that rain or snow was approaching. They also watched flowers such as dandelions, which fold their petals before a storm. If night brought a halo around the Moon, rain was expected. (Now we know that a lunar halo is caused by light shining through the ice crystals of cirrus or cirrostratus clouds that often precede thunderstorms.)
For generations, country folk have relied on the woolly bear caterpillar (sometimes called the woolly worm) for their winter weather forecasts. According to legend, the little black-and-brown caterpillar can predict the severity of the coming winter by the width of its bands: the wider the brown segment, the milder the winter.
Just as modern forecasters can’t boast of a 100 percent success rate in their forecasts, folk wisdom often got it wrong. Have you ever heard that “lightning never strikes twice in the same place”? Tell that to the workers in the Empire State Building, which gets hit about twenty-three times a year, on average. In fact, during one especially bad thunderstorm it was struck eight times in twenty-four minutes.
After the development of weather instruments, forecasting became much more accurate. But as late as the mid-1950s, incoming weather data were assembled and plotted on charts by hand. In many cases forecasts were based on historical records that were compared with current conditions for similarities. Often the results were surprisingly accurate, but human brains were just not up to the task of quickly analyzing the huge quantities of data that were needed to consistently generate accurate forecasts.
In 1943, with most scientific resources being directed toward the war effort, scientists at the Moore School of Electrical Engineering at the University of Pennsylvania began construction on a machine that would, quite literally, change the world.
Dubbed ENIAC (Electronic Numerical Integrator and Computer), the groundbreaking device was designed to compute the trajectories of ballistic artillery shells. ENIAC was a behemoth by today’s standards, containing 18,000 vacuum tubes (which broke down at the average rate of one every seven minutes) and 1,500 relay switches. It weighed in at a hefty 30 tons, and could compute fourteen 10-digit multiplications per second. Operators used 6,000 switches and a host of jumper cables to program the beast.
In 1946, while using ENIAC to simulate nuclear explosions, Princeton mathematician John von Neumann realized that the computer might also be used for weather prediction. In 1950 his team produced the first computer-based numerical weather forecast, and while it wasn’t an unqualified success, it did show that numerical forecasting was feasible.
Although the house-sized ENIAC was much faster than manual calculations, those early electronic pioneers learned that atmospheric conditions change more rapidly than the early machines could calculate them. The sheer volume of data meant that there was still much room, and need, for improvement.
In 1954, elements of the Navy, Air Force, and Weather Bureau formed the Joint Numerical Weather Prediction Unit in Suitland, Maryland, to further refine numerical forecasting, and in 1955 the unit began issuing regular real-time forecasts. Still, computerized forecasts were not as accurate as the older subjective methods.
While computer makers toiled to create faster processors, the science of weather observation and forecasting was about to take its next giant leap. In 1946, with Hitler’s V2 rocket attacks on London still a fresh wound, the Army Air Forces and the RAND Corporation cosponsored a paper entitled “Preliminary Design of an Experimental World-Circling Spaceship.” Though mostly concerned with military surveillance, the paper mentioned that such a satellite might also make a good weather reconnaissance platform.
In 1954 the first pictures of a tropical storm were taken from space using a US Navy Aerobee rocket. The spectacular photos galvanized the meteorological community: for the first time a giant weather system could be seen in its entirety. But rockets took time to prepare for launch, were expensive and difficult to recover, and spent very little time over their targets. What was really needed was an observation platform that could stay in space for long periods of time, and the World-Circling Spaceship was it.
NASA had its hands full just getting TIROS 1 off the ground. While still in the planning stages, the satellite went through a succession of proposed launch vehicles until the Thor-Able rocket (a modified intercontinental ballistic missile) was chosen. The first Thor-Able blew up 146 seconds after launch, and the first prototype TIROS spun out of control after reaching orbit.
On April 1, 1960, the recently formed National Aeronautics and Space Administration (NASA) launched the world’s first weather satellite atop an Air Force Thor-Able rocket. Weighing only 263 pounds, TIROS 1 (Television and Infrared Observation Satellite) began to return dozens of pictures of the earth and its cloud cover on its very first day in orbit. Though grainy, those first crude pictures were a snapshot of the future for weather forecasters, who for the first time had an eye in the sky for tracking weather systems like fronts and hurricanes. Orbiting the planet every ninety-nine minutes, TIROS 1 spotted a tropical cyclone in the waters of the South Pacific north of New Zealand nine days after launch, the first storm to be detected by satellite. Until that time, tropical storm prediction relied on ship and aircraft reports that were often spotty and unreliable.
Today NOAA operates two satellites called GOES (Geostationary Operational Environmental Satellites); one keeps an eye on weather conditions in North and South America and most of the Atlantic Ocean, and the other monitors part of North America and the Pacific Ocean. Geosynchronous satellites provide an overview of a whole hemisphere’s weather conditions since they are capable of imaging the full disk of Earth in one snapshot.
GOES satellites sport two main instruments called an imager and a sounder. The imager measures the amount of energy being radiated from the earth and how much solar energy is being reflected from the surface and atmosphere. The sounder takes the earth’s temperature and determines the atmospheric moisture level, as well as surface and cloud top temperatures and ozone levels.
GOES is also outfitted with a search-and-rescue transponder and a space environment monitor consisting of a magnetometer, an X-ray sensor, a high-energy proton and alpha detector, and an energetic particles sensor. These instruments allow GOES to report on the state of the solar wind and warn of approaching solar storms caused by CMEs (coronal mass ejections) from the Sun. That’s a lot of bang for the buck considering GOES started out strictly as a weather satellite.