UT Austin researchers help predict local weather during 2024 Paris Olympics (2024)

The 2024 Paris Olympics gives elite athletes from around the world — including dozens with ties to Austin — the chance to compete on a global stage. But researchers are also using the international event as an opportunity to refine and advance weather-forecasting techniques in urban areas.

The "Olympics is not just about competing," UT Austin Professor Dev Niyogi said. "It is, probably to me, also about computing."

Niyogi, who teaches in the Jackson School of Geosciences and co*ckrell School of Engineering, is leading a team using artificial intelligence to develop daily, hyperlocal weather forecasts that are shared with Olympic officials in France.

Niyogi said current weather forecasts are focused on regional, larger-scale weather patterns — not the unique environment of cities. He said the models don’t take into account factors like how the sun heats up an urban area, how the wind might flow through buildings, the impact of shade from trees or pollution.

“These aspects are not represented in a detailed manner in the weather models,” he said.

So, while we might know whether it’s going to rain (see the Games’ opening ceremony), it’s a lot harder to determine what Niyogi calls “thermal comfort.” That's how you, I or an Olympic athlete might feel while walking or competing outside.

“There is a high degree of weather sensitivity in terms of the [Olympic] operations, not just from the athletes and their performance," he said, "but the experience you will have as a visitor there."

Detailed forecasts quick

Niyogi said figuring out thermal comfort requires looking at a bunch of “micro effects” on the local level.

“These effects cannot be put into our large models," he said, "not because we don’t understand them, but because if we were to solve those equations from a micro to a regional scale they would take an enormous amount of computational power and they would also take a lot of time."

That's where artificial intelligence comes in. AI is helping Niyogi and his fellow researchers generate detailed data for weather models more quickly.

“A lot of agencies across the world are now trying to set up these AI-based forecasts to supplement their physics-based forecasts,” said Harsh Kamath, a graduate research assistant and PhD student who works with Niyogi.

Naveen Sudharsan, a postdoctoral fellow who is also working on the project, said another advantage of using AI is that the forecasts are ready in mere minutes.

Manmeet Singh, another postdoctoral fellow on the team, said AI allows them to forecast weather by the hour and by the kilometer.

“So it’s hyperlocal in every sense in space and in time and it is very important for decision-making,” he said.

He said it can help a city determine which areas get the most rain or which communities are the most vulnerable to heat.

Applications in Austin

UT researchers aren’t just applying this approach to the Paris Olympics; they’re putting it to use in Austin as part of the UT-City Climate CoLab, which seeks to localize climate information.

“The tools and techniques that we’re developing and testing for places like the Olympics have a ready application for what we can do in our neighborhoods in Austin,” Niyogi said.

He said hyperlocal data can help inform the city’s decision making about where to plant trees, put up shade structures at playgrounds or install bus shelters.

“You want to go to places which are most vulnerable and where the impact is highest,” Niyogi said.

Allysa Dallmann works with Niyogi and is the climate coordinator for the TExUS Lab, which investigates extreme weather in urban areas.

She said city departments she works with have emphasized the need for more localized climate data. She said Austin has weather stations at the airport and Camp Mabry, but that leaves a lot of areas without coverage.

“They’re trying to understand locally in Austin where are there pockets in the city where potentially there are people who are in need of more resources due to extremes in either heat, air quality, precipitation [and] flooding,” she said.