Return to Transcripts main page

American Morning

NASA's New Hurricane Predictor

Aired July 31, 2001 - 09:18   ET

THIS IS A RUSH TRANSCRIPT. THIS COPY MAY NOT BE IN ITS FINAL FORM AND MAY BE UPDATED.

THIS IS A RUSH TRANSCRIPT. THIS COPY MAY NOT BE IN ITS FINAL FORM AND MAY BE UPDATED.
DONNA KELLEY, CNN ANCHOR: In 1989, Hurricane Hugo slammed into Charleston, South Carolina and caused $7 billion worth of damage. Twelve years later, it is still the second costliest hurricane ever in the United States. Today, half a world away Taiwan is beginning to clean up in the wake of the most devastating of eight typhoons to hit that island this year. The storm has killed at least 61 people and at least 152 others are missing.

We are, of course, moving deeper into hurricane season and this year NASA has a new supercomputer that could help improve the accuracy and hurricane predictions of forecasts. To give you an idea of just how important these predictions can be, 19 of the 20 costliest hurricanes on record have come in the past 30 years as more and more of us have moved to coastal areas. But while damages have gone up, death tolls have gone done. Only one of the 20 deadliest hurricanes has struck in the past 30 years thanks, in part, to improving forecasting.

And joining us now to talk about this new NASA supercomputer is William Feiereisen. He's at NASA's Ames Research Center in Moffett Field, California. Mr. Feiereisen, nice to have you join us.

WILLIAM FEIEREISEN, NASA: Good morning, Daryn.

KELLEY: And I'm going to chat with you as well as Chad Myers, our meteorologist. So we'll bring Chad in in just a quick minute or two. What makes this one so great now? What's so special about this supercomputer?

FEIEREISEN: Well, it's really a combination not only of the computer itself, but some accomplishments that we made recently in being able to figure out how to extract more horsepower out of these computers to be able to apply their climate modeling and weather modeling. So it's a combination of several things.

KELLEY: So if you do that, if you can apply the more horsepower, what's that going to do for you? Can you tell us quicker if a hurricane is going to hit or strength or how does this help you?

FEIEREISEN: Well, as time goes on what it will allow us to do is it will allow us to make better predictions of if there are going to be more hurricanes and farther into the future it'll allow us to make predictions about where those hurricanes are going to go and a little bit more accurately when and where they're going to hit.

CHAD MYERS, CNN CORRESPONDENT: Mr. Feiereisen, Chad Myers here.

FEIEREISEN: Hi, Chad.

MYERS: This is actually going to be a climate computer, correct? It's not going to be a weather modeling computer. It's not going to give us a better estimate of how much snow Detroit's going to get compared to Toledo, right?

FEIEREISEN: Yes. In the short-term that's very true. Interestingly enough, there is a difference between global climate modeling, which predicts the long-term trends and what's going to happen with our climate in comparison to weather modeling, which predicts what's going to happen to you where you live in the next few days or few weeks.

MYERS: You've been trying to do this now...

FEIEREISEN: And so this particular computer is really designed to help out in answering questions about what we've done to the climate of the earth, changes in the average temperature of the earth or changes that we have done, let's say, in deforestation and what implication that will have for our climate in the future.

MYERS: You've been trying to do this now for many years, but this new computer is going to allow you to put many more things into that model to help a, to I guess get a more forecast model, a better forecast. What kind of things and what potential are you trying to pull out of this computer that you didn't have with the old Cray or with the old supercomputers?

FEIEREISEN: Well, the climate itself is a very, very complicated combination of physical processes. There are modelings of the ocean. There's modeling of the atmosphere. There's modeling of land usage, vegetation, geological data. And what we find as time goes on is that all of these physical processes interact with each other in much, much more complex ways than we've ever understood previously. And it points out that we need ever more computer horsepower to be able to model these interactions accurately.

KELLEY: Let's go ahead and take a look at some pictures, Mr. Feiereisen. We've got some pictures of a simulation of Hurricane Floyd that happened back in 1999. Can you talk us through this and how the supercomputer now with the new animation, how that helps you, what differences you see and how that could help you now?

FEIEREISEN: OK. Let me talk just a little bit about this particular animation. What this is is it's running one piece of the entire system and so what it does is it starts out with the weather at a particular point in time just before Hurricane Floyd actually starts and then it follows the equations of physics and shows the development of Hurricane Floyd as it comes across Panama and then heads up towards Florida.

But for those who were actually in Florida at the time, you'll notice that it went up the wrong side of Florida. What this points out is that our abilities to be able to simulate the directions of hurricanes and their intensity still needs improvement. And so what we are doing is combining these simulations with data that comes from satellites and it comes from ground based weather stations to be able to correct these models as time goes on.

KELLEY: All right...

FEIEREISEN: And so our ability to be able to combine this experimental data and then do the simulations is something that's going to improve our models and will allow us to more accurately predict these things in the future.

KELLEY: OK, well...

FEIEREISEN: And the key is to be able to have enough computer horsepower to be able to support the earth scientists in the development of their models.

KELLEY: William Feiereisen, who is the chief of NASA advanced supercomputing facility at Ames Research Center in California, thank you very much.

FEIEREISEN: Thank you very much.

KELLEY: It's very interesting and it probably makes great sense to you, Chad, since you're a meteorologist and you can follow along on that so easily.

MYERS: It does.

KELLEY: And it's just get it there faster and better and so it'll make predicting good?

MYERS: You know, I talk to kids a lot about this thing we call weather models. What is a model? What's a model car? It's a bunch of pieces of plastic that you all put together and hopefully when you're done it looks like a car. Well, right now we're trying to model the atmosphere in nine layers. There's a lot more layers than nine. You fly through the atmosphere, there's an infinite number of layers.

Try to build that same model car with nine pieces. It's not going to look so much like a car. You may get three wheels, a fender and a hood. But then you get 50 pieces and the car begins to look a little bit better. You get 100 pieces and then the car really starts -- you get a windshield on it, you get some mirrors on it, it starts to look like a car. So the models get better and better the more pieces we can put in and, you know, the models we were using before can now be run on computers you can buy at Circuit City.

LEON HARRIS, CNN ANCHOR: That's amazing.

MYERS: That is unbelievable. They will send you the model if you want to run the old computer models and run them on your home computers. KELLEY: Why don't you run the forecast on that?

HARRIS: Well, why don't you run over there and give us the forecast?

MYERS: All right, let's get right to it.

OK, temperatures outside across the country are going to be cool and across the northern plains, hot across the south central region and some rain showers now down here across the south. This is a weather computer model. All the lines you see here, all the rain dots you see here, the blobs all put on by the model that we ran this morning here at CNN. Everything was put on the map. The fronts were put on. The lows and the highs were manually drawn in. But everything else you see here, all modeled.

If the model is right, we're right. If the model is wrong, we're wrong. If we only get nine pieces of that car and it's supposed to be a Volkswagen and we make it look like a Buick, this is not going to be a very accurate forecast. Today it probably is. But that's the deal with the models and that's how the models get better and that's how the models are going to make our forecast much better in the future.

TO ORDER A VIDEO OF THIS TRANSCRIPT, PLEASE CALL 800-CNN-NEWS OR USE OUR SECURE ONLINE ORDER FORM LOCATED AT www.fdch.com