Storm Week: A Revolution in Forecasting
Without the forecasts and radar technologies developed by the folks at the national weather service, it would be impossible to chase severe weather the way we do today.

Did you know numerical modeling of the atmosphere is considered one of the top five scientific achievements of the 20th century? It ranks right up there with moon landings, medical breakthroughs, and nuclear energy.

National Weather Service forecasters can now predict major weather events a week or more in advance, and the tornado warning lead time is approaching 15 minutes. We recently traveled to the Washington D.C. area to meet the scientists behind this revolution.

The East Coast got slammed this winter with more snow than they'd ever seen. But they also had more than a week to prepare for it. As we know too well in Chicago, that hasn't always been the case. In 1967, 23 inches of snow fell on the city, the largest single snowfall in our history. The difference was back then we had no clue it was coming.

A little more than a decade later, Chicago was hit hard again, but in 1979, we had time to brace for it.

The most prolific tornado in Oklahoma history hit in 1999, yet forecasters warned the public that the atmosphere was ripe for tornadoes days in advance.

And remember how long we watched Hurricane Katrina churn her way through the Atlantic before it hit the Gulf Coast? That was the revolution we're talking about!

If you were to identify one building in this country in which the revolution of weather forecasting has had the greatest single impact, it would be this one, the World Weather building outside Washington DC.

"Over the past 15 years the advances in the forecast enterprise have been absolutely incredible," said Dr. Louis Uccellini, the director of the National Centers for Environmental Protection, or NCEP for short. It's the office that oversees all other government forecast facilities.

"We are always striving to measure how good these forecasts are and using that effort to improve the forecast and I think it's paid off," Dr. Uccellini said. "We now extend our forecast out to day seven with useful skill. The general term that's used is we've added a day of skill per decade."

Three things happened in the past half century to improve forecasts. Remote sensing of the atmosphere by satellites and radars allow us to "see" what's brewing over previously unobserved areas of Earth. That atmospheric data over oceans, land and icepacks is then fed into supercomputers capable of writing mathmatical equations for hurricanes or clouds, what we call numerical models. C

omputers then run the numbers fast enough for forecasters to stay ahead of weather, and better communications are speeding the delivery of severe weather information to the public.

"When I started off at Goddard space flight center," explained Dr. Uccellini, "we had a computer that did a million calculations per second and we thought that this was fabulous. And here we are today at 70 trillion calculations per second."

Seventy trillion mathmatical calculations per second. Can you imagine that? It's mind boggling. Yet that's the speed at which these IBM computers are running the National Weather Service weather forecast models. And we're told in the next year or two... that speed could double or triple again.

Those incredible calculations are creating greater confidence in forecasts.

By using a spaghetti bowl of computer models called ensembles, a range of future weather possibilities emerge: "So instead of just giving you one model run," said Dr. Uccellini, "we're gonna give you 80. And you will see from those 80 models runs that a 67 or 70 or some number of em are all tending towards that storm going North of Chicago, three-to-four days from now. So, you're gonna have increasing confidence."

We are fast approaching the 20th anniversary of the deadly plainfield tornado. Remember, we didn't even have doppler radar back then.