How Supercomputers Are Making Hurricane Forecasting Much Easier

It'll never be perfect, but it's going to get pretty close.

hurricane

With Hurricane Dorian having devastated the Bahamas and beginning to do some damage in Florida, we’re again facing the unfortunate fact that climate change is going to make these kinds of storms more frequent and more deadly. Not only that, climate change researchers believe we’re starting to enter a period where the hurricane season will be significantly longer — meaning Americans in the Southeast and in Puerto Rico will have less time to recover between these potentially catastrophic hurricane seasons.

Don’t curl into a ball and start sucking your thumb, though, because there is some good news: We are getting a lot better at tracking these monsters.

After Hurricane Sandy tore through the East Coast in 2012, Congress approved more than $80 million in supplemental funding for the National Weather Service (NWS) to spend on improving its storm forecasting and tracking capabilities. That money went to investing in supercomputers for the agency and research and development. This investment in supercomputing means the agency can more quickly process information it receives from its storm tracking satellite, the GOES-16 (originally called the GOES-R, and data it receives from buoys, aircrafts, and other sources.

See also: GOES-16 Captures Lightning Strike the Earth From Outer Space

New hurricane models that are being used by the NWS, such as the HMON and HWRF, mean these supercomputers can now predict within about 100 miles where a hurricane is heading three days ahead of time. That’s down from about 300 miles in 1990.

Before these supercomputers, much of the data these agencies were receiving before and during a storm was wasted because the computers and the models they were using couldn’t analyze it all in a useful way.

Michael Mann, a professor of atmospheric science at Penn State University, tells Inverse increased analysis is only the beginning.

“It has improved a huge amount, thanks to more sophisticated numerical simulation models combined with better observations, especially satellite data,” Mann says. “We need to continue to refine the models, obtain the best possible in-site and remote observations, and incorporate those observations into the model predictions themselves.”

This NASA video shows Hurricane Dorian on the morning of Tuesday, September 3, 2019, from the point of view of the ISS.

The National Oceanic and Atmospheric Administration (NOAA) has also been working on using drones to collect data from hurricanes for the past few years.

“The more the climate changes, the more we’ll have to adapt the models we’re using.”

Six drones that were designed by defense contractor Raytheon were actually sent into the eye of Hurricane Maria back in 2017, and the agency has been testing out these kinds of drones for storm tracking ever since. The drones are small and designed to be expendable, and they are launched from large aircrafts like the Lockheed P-3 Orion used by the NOAA. Each drone can cost upwards of $10,000, which is why they’re not utilized in every hurricane, but the agency is hoping to find ways to reduce that cost.

It seems safe to say our ability to predict where these massive storms are heading next and what they’ll do will improve greatly as we utilize things like artificial intelligence and machine learning more in the fight against climate change. A team of researchers at Penn State, AccuWeather, Inc., and the University of Almería in Spain has actually developed “a framework based on machine learning linear classifiers — a kind of artificial intelligence — that detects rotational movements in clouds from satellite images that might have otherwise gone unnoticed.” Within just a few years, this kind of technology could take predictions to a whole other level.

Mann says the fact that the climate is changing is “yet another uncertain variable thrown into the chaotic mathematical equations” that help us forecast and track these hurricanes. The more the climate changes, the more we’ll have to adapt the models we’re using to factor in those changes in water temperature, wind patterns, and more.

Using supercomputers, artificial intelligence, drones, airplanes, buoys, satellites, and more, we’re slowly getting closer to a point where we could be able to predict where a hurricane is heading and what it’ll look like with almost granular detail. Weather is, of course, never something you can predict with 100 percent accuracy, but we might get pretty close.

The only thing that’s going to stop these storms from becoming more frequent and more deadly is actually cutting emissions and removing carbon from the atmosphere, but assuming they’re going to keep getting worse, at least we’ll be able to tell people when they need to get the hell out of town.