Can AI Replace Weather Balloons? Climate Modeling Adapts As Risks Rise

The climate risk to commercial and residential properties across the U.S. is quickly growing, with trillions of dollars worth of real estate exposed to weather and climate events. At the same time, President Donald Trump is hoping to slash the budget of federal forecasters. 

If he is successful, the weather models that property insurance companies rely on would likely still hold up, but property owners stuck in the path of a dangerous storm may not be so lucky, said Howard Botts, chief scientist at Cotality.

Damage from disasters costing at least $1B exceeded $600B in 2023 alone, according to federal data. More is at risk every day, and more people are putting themselves in the path of disaster. 

“There’s a pretty close correlation between high-risk areas and the areas where people are choosing to move,” said Howard Botts, chief scientist at Cotality.

Under new federal priorities, real-time data could become less robust, making it harder to make short-term predictions. Some startups are already exploring artificial intelligence solutions, Botts said. 

The National Oceanic and Atmospheric Administration, which conducts research on active storms and provides the forecasts that Americans rely on, has been targeted for budget cuts as part of Republicans’ efforts to shrink government.  

The One Big Beautiful Bill moving through Congress would cut NOAA’s budget 26.2% to $4.5B, including downsizing the office responsible for conducting weather and climate studies.  

Cotality, a property data analytics firm that was known as CoreLogic before a March rebrand, is unaffected by the cuts because it relies on after-the-fact data as opposed to forecasts, Botts said. Its model combines past radar data, global climate patterns and hundreds of data points for every piece of property in the United States to give parcel-level forecasts. 

But the quality of the weather forecasts that average people use could be diminished, at least in the near term. NOAA launches weather balloons twice a day that provide key climate data that is universally used to predict the strength and direction of storms, Botts said.

“It looks like they're going to cut back on those, and that could have a real-time impact on forecasting models,” Botts said during a presentation Wednesday at the National Association of Real Estate Editors’ 59th annual conference in New Orleans. 

Private entities like universities may have to step in to close the gap, which would take time. Artificial intelligence also offers promise, Botts said, as the traditional physics-based way of predicting weather is challenged by a crop of startups that use specially trained AI models that generate forecasts.

 “A lot of these [companies] already have contracts with major retailers, amusement parks, transportation — and they're finding they can create weather forecasting models, which at least they claim are more accurate at a much faster rate,” Botts said. “The jury is still out on that part.”

The reliability of forecasting becomes more important by the day, with Cotality’s models predicting wide swaths of the country that aren't accustomed to climate disasters and storms will become increasingly vulnerable to severe weather. Storms themselves are also becoming more damaging as global temperatures tick up, causing them to drop even more drenching rains. 

Flash floods have supplanted riverside floods as the major source of loss today, he said. Winds used to be the most damaging aspect of hurricanes, but today, it’s flooding from drenching rains that reach further inland than ever before. 

“For every one degree centigrade the temperature goes up, the air can hold 7% more water,” Botts said. “As we get warmer and warmer, there's more water in the air, and that's led to a fundamental shift in the kinds of flooding we're getting.”