NOAA gets a new weather forecasting model

MARY REICHARD, HOST: Up next, changes to your weather forecast.

Now, the Babylonians studied clouds to figure out what the skies would do. Aristotle wrote volumes on what caused bad weather.

NICK EICHER, HOST: But real progress in predicting the weather came with modern science: tools that measured things like temperature, atmospheric pressure, and humidity. 

Those tools along with weather satellites and computer models help meteorologists forecast weather fairly accurately five to 10 days in advance.

REICHARD: It’s those computer weather models that are especially important. 

In the United States, the weather model most used is known as the GFS, or “Global Forecast System.” The National Weather Service put that in place in 1980. And now the weather service is rolling out an update. 

Here’s WORLD Radio’s Sarah Schweinsberg.

SARAH SCHWEINSBERG, REPORTER: In October 2012, meteorologists predicted Hurricane Sandy. The hurricane was actually a series of tropical storms wrapped into one powerful force.

AUDIO: Hurricane Sandy is serious. It has already killed 21 people in Haiti, Jamaica, and Cuba.

After Sandy pummeled the Carribbean, the National Weather Service GFS model showed the storm would move northeast—away from the U.S. coast.

But the European Centre for Medium-Range Weather Forecasts model predicted the storm would hook west toward the Mid-Atlantic coast.

The European model turned out to be right.

METEOROLOGIST: Needless to say, this storm, Sandy, continues to strengthen. It has started to make that northwest turn which is so very unusual to see from a tropical system coming up here into our northern waters. 

The European model predicted the left hook seven days out. The GFS model didn’t predict it until several days later.

After Sandy, U.S. meteorologists began to question why the GFS didn’t correctly foresee the storm’s path when the European model did.

Brian Gross directs the National Oceanic and Atmospheric Administration’s Environmental Modeling Center. NOAA oversees the National Weather Service. Gross says after nearly three years of development, meteorologists have upgraded the GFS model’s dynamical core in time for the 2019 hurricane season. The new model is called the GFSFV3.

GROSS: The dynamical core generally handles the prediction of wind and pressure in the atmosphere. From a technical point of view, the FV3, uh, approaches the way that it solves the equations of motion in a different way than the old spectral core did. 

With its improved mathematical algorithms, Gross says the model should better predict hurricanes and extreme weather.

GROSS: This includes a better forecast of the jet stream and the weather that it drives. The simulation of hurricane tracks. There is better precipitation forecasts. There is improved representation of some of the physical processes, uh, beyond just the wind and the pressure to take place in the atmosphere. For example, the formation of precipitation like rain and snow.

Gross says in order to test the new model’s accuracy, the National Weather Service reforecasted three years of weather patterns. The tests suggest the new model improves five-day forecasts and better predicts hurricane tracks and intensity.

GROSS: We’ve had over a hundred scientists and model modelers and software programmers and technicians, uh, involved in evaluating this model….We’ve seen noticeable improvements with the upgrade to the model.

But not everyone is ready to sing the new model’s praises. Cliff Mass is a Professor of Atmospheric Sciences at the University of Washington. He isn’t convinced by the re-forecasting test data—although he thinks it will improve over time.

MASS: I don’t think there’s any evidence that it’s superior over, over their previous modeling systems.

Other meteorologists share Mass’s skepticism, in part because the new model exhibited a “cold bias” this winter. That meant it often over-predicted snowfall.

NOAA’s Brian Gross says the agency took steps to fix that problem and will continue to make improvements. Cliff Mass says if the new model isn’t accurate when it comes to snowfall or other weather patterns… that could have big ramifications. Many regional weather models in the U.S. will be based on the GFSFV3.

MASS: If the global model’s inferior, that means the other models will be inferior as well.

Mass says the new model will have to earn the trust of meteorologists. In the meantime, forecasters will continue to consult other weather models, like the one in Europe.

Roy Spencer is a meteorologist and research scientist at the University of Alabama. He’s also underwhelmed by the new model’s capabilities. But he admits it’s a step in the right direction.

SPENCER: There aren’t a lot of real immediate improvements we’re going to see with this model except maybe for severe weather forecasting.

But Spencer says the National Weather Service may not be completely to blame. A weather model is only as good as the data it has to work with. Spencer says if we want even better forecasts, we need to invest in better weather satellites.

SPENCER: The trouble is that those are expensive to design and launch. And to maintain and monitor the data from satellites is an expensive endeavor. People in the government seemed to be more interested in funding a new science and not, you know, boring old, the same weather forecast stuff. It’s just not that exciting. 

Not exciting perhaps until a storm like Hurricane Sandy roars ashore.

Reporting for WORLD Radio, I’m Sarah Schweinsberg.

(Photo/NOAA) Hurricane Sandy

WORLD Radio transcripts are created on a rush deadline. This text may not be in its final form and may be updated or revised in the future. Accuracy and availability may vary. The authoritative record of WORLD Radio programming is the audio record.