Colorado, Wyoming researchers use AI to predict destructive hailstorms earlier

FORT COLLINS, Colo. Hailstorms can cause significant damage to homes, vehicles and crops across Colorado. But what if people had hours, or even days, of warning before a storm hits?

For Dave Petrocco, a farmer in LaSalle, better forecasting could make a big difference.

Watch the story from Denver7's Northern Colorado and Eastern Plains reporter Peter Choi in the video below:

Petrocco has spent generations working the land, but he says severe weather remains one of the biggest challenges farmers face.

Big dark clouds come by and drop that. I say it's stones of the devil from the sky, Petrocco said.

Hailstorms passed through his crops last June, damaging some of his fields. Fortunately, the hail came early enough in the growing season for the crops to recover.

But unpredictable storms can put farmers, equipment and an entire harvest at risk.

It really is hard to predict. Sometimes, we have people in the field, and we've got tractors and equipment, and we're harvesting and planting, Petrocco said.

That's the problem CSU atmospheric science researchers are trying to address.

"I represent a group of five different professors here: Three at CSU and two at the University of Wyoming," said Kristen Rasmussen, Associate Professor in the Department of Atmospheric Science at Colorado State University. "We also have four research scientists on this project. What we're doing is we're taking two existing technologies. These are both AI-based prediction models."

They are developing artificial intelligence technology they hope will help predict destructive hailstorms earlier and more accurately.

Rasmussen and her team are developing forecasting technology that uses artificial intelligence and historical weather observations to identify the atmospheric conditions that lead to hail formation.

The research combines longer-range machine learning models with AI radar forecasting technology. Researchers hope the system can eventually provide hail forecasts several hours, or even days, ahead of a storm.

What we're doing is trying to build the atmospheric ingredients that lead to hail production. We feed those ingredients into a hail prediction model, and then we try to predict into the future, out to eight days, Rasmussen said.

The research is particularly important in Colorado, where hailstorms can cause billions of dollars in damage. Hail is also a major factor driving homeowners insurance costs across the state.

Rasmussen said earlier warnings could benefit not only farmers, but also homeowners and drivers.

The new technology could give them more time to move vehicles, protect their property, and prepare for approaching storms.

The CSU team is continuing to develop and test the technology, with additional field research planned for spring 2027. Researchers eventually hope to work with the National Weather Service to help put the forecasting system into practical use.

For Petrocco, having more time to prepare could mean fewer losses and less stress.

That would be a real gift and beneficial. And boy, if we had that kind of warning, it would save us a lot of money. Yeah, and risk, and it would save us a lot of stress too, Petrocco said.