This article originally appeared in the New York Times in May 2023.
A.I. and innovations like hyperlocal radar stations and a constellation of future satellites may soon improve forecasts and smooth out your journeys.
It may be a tough summer to fly. More passengers than ever will be taking to the skies, according to the Transportation Security Administration. And the weather so far this year hasn’t exactly been cooperating.
A blizzard warning in San Diego, sudden turbulence that injured 36 people on a Hawaiian Airlines flight bound for Honolulu, a 25-inch deluge of rain that swamped an airport in Fort Lauderdale, Fla.: The skies have been confounding forecasters and frustrating travelers.
And it may only get worse as the climate continues to change. “Intense events are happening more often and outside their seasonal norms,” said Sheri Bachstein, chief executive of the Weather Company, part of IBM, which makes weather-forecasting technology.
So, will flights just get bumpier and delays even more common? Not necessarily. New sensors, satellites and data modeling powered by artificial intelligence are giving travelers a fighting chance against more erratic weather.
Better data, smarter software
The travel industry “cares about getting their weather predictions right because weather affects everything,” said Amy McGovern, director of the National Science Foundation’s A.I. Institute for Research on Trustworthy A.I. in Weather, Climate and Coastal Oceanography at the University of Oklahoma.
Those better weather predictions rely on a type of artificial intelligence called machine learning, where in essence, a computer program is able to use data to improve itself. In this case, companies create software that uses historical and current weather data to make predictions. The algorithm then compares its predictions with outcomes and adjusts its calculations from there. By doing this over and over, the software makes more and more accurate forecasts.
The amount of data fed into these types of software is enormous. IBM’s modeling system, for example, integrates information from 100 other models. To that, it adds wind, temperature and humidity data from more than 250,000 weather stations on commercial buildings, cellphone towers and private homes around the globe. In addition, it incorporates satellite and radar reports from sources like the National Weather Service, the National Oceanic and Atmospheric Administration and the Federal Aviation Administration. Some of the world’s most powerful computers then process all this information.
Here’s how all this may improve your future trips:
Safer and calmer flights
The skies are getting bumpier. According to a recent report from the National Aeronautics and Space Administration, “severe turbulence at typical airplane cruising altitudes could become two to three times more common.”
Knowing where those disturbances are and how to avoid them “is mission-critical for airlines,” Ms. Bachstein said.
Tomorrow.io, a weather intelligence company based in Boston, received a $19 million grant from the U.S. Air Force to launch more than 20 weather satellites, beginning with two by the end of this year and scheduled for completion in 2025. The constellation of satellites will provide meteorological reporting over the whole globe, covering some areas that are not currently monitored. The system will report conditions every hour, a vast improvement over the data that is currently available, according to the company.
The new weather information will be used well beyond the travel industry. For their part, though, pilots will have more complete information in the cockpit, said Dan Slagen, the company’s chief marketing officer.
The turbulence that caused dozens of injuries aboard the Hawaiian Airlines flight last December came from “an evolving thunderstorm that didn’t get reported quickly enough,” Dr. McGovern said. That’s the kind of situation that can be seen developing and then avoided when reports come in more frequently, she explained.
Fewer snarls on the ground
The F.A.A. estimates that about three-quarters of all flight delays are weather-related. Heavy precipitation, high winds, low visibility and lightning can all cause a tangle on the tarmac, so airports are finding better ways to track them.
WeatherSTEM, based in Florida, reports weather data and analyzes it using artificial intelligence to make recommendations. It also installs small hyperlocal weather stations, which sell for about $20,000, a fifth of the price of older-generation systems, said Ed Mansouri, the company’s chief executive.
While airports have always received detailed weather information, WeatherSTEM is among a small set of companies that use artificial intelligence to take that data and turn it into advice. It analyzes reports, for example, from a global lightning monitoring network that shows moment-by-moment electromagnetic activity to provide guidance on when planes should avoid landing and taking off, and when ground crews should seek shelter. The software can also help reduce unnecessary airport closures because its analysis of the lightning’s path is more precise than what airports have had in the past.
The company’s weather stations may include mini-Doppler radar systems, which show precipitation and its movement in greater detail than in standard systems; solar-powered devices that monitor factors like wind speed and direction; and digital video cameras. Tampa International, Fort Lauderdale-Hollywood International and Orlando International airports, in Florida, are all using the new mini-weather stations.
The lower price will put the equipment within reach of smaller airports and allow them to improve operations during storms, Mr. Mansouri said, and larger airports might install more than one mini-station. Because airports are often spread out over large areas, conditions, especially wind, can vary, he said, making the devices valuable tools.
More precise data and more advanced analysis are helping airlines fly better in cold weather, too. De-icing a plane is expensive, polluting and time-consuming, so when sudden weather changes mean it has to be done twice, that has an impact on the bottom line, the environment and on-time departures.
At the University of Oklahoma, Dr. McGovern’s team is working on using machine learning to develop software that would provide hailstorm warnings 30 or more minutes in advance, rather than the current 10 to 12 minutes. That could give crews more time to protect planes — especially important in places like Oklahoma, where she works. “We get golf balls falling out of the sky, and they can do real damage,” Dr. McGovern said.
More on-time departures and smoother flights are most likely only the beginning. Advances in weather technology, Dr. McGovern said, are “snowballing.”