Environmental science and technological advancement may not be something people always think of as being related, but at Colorado State University, they go hand in hand.
Researchers at CSU are partnering with the University of Oklahoma on a five-year program to work with artificial intelligence and study how AI can be utilized in environmental research, according to a CSU SOURCE article.
Matthew Rogers, a research associate and education and outreach coordinator for the Cooperative Institute for Research in the Atmosphere, wrote in an email to The Collegian that the program, which has received $20 million from the National Science Foundation, was “rigorously reviewed” before being chosen.
“Research (professor) Imme Ebert-Uphoff has worked with the program lead (professor) Amy McGovern at OU on other projects and (has) co-written a paper with McGovern on the blending of academia, government and private sector interests for AI research,” Rogers wrote in the email. “When the NSF solicitation came along in September of last year, it was a natural match to the work they had already been doing.”
Ebert-Uphoff, who works at CIRA and in the department of electrical and computer engineering, will work on the program alongside multiple CSU researchers, including associate professor of atmospheric science Elizabeth Barnes and computer science professor Chuck Anderson.
The SOURCE article states that CSU will lead two “use-case” studies to “better understand tropical storms and prediction of severe weather events” on multiple time scales.
“Trustworthy AI tools can lead to more trustworthy forecasts and warnings, and that’s a major benefit to society.” -Matthew Rogers, CIRA education and outreach coordinator
Rogers explained that use-case studies are examples of how new AI techniques improve a specific research task.
“We already have some AI tools that look at certain observations — infrared satellite imagery related to atmospheric water content, for example — that tells us about storm structures, or can predict a zone of severe storm formation,” Rogers wrote.
According to an article by Renee Cho of State of the Planet, the news blog for The Earth Institute at Columbia University, AI improves weather and extreme event forecasting because in its calculations it incorporates real-world elements, such as atmospheric and ocean dynamics and chemistry.
“Trustworthy AI tools can lead to more trustworthy forecasts and warnings, and that’s a major benefit to society,” Rogers wrote.
Rogers wrote that AI applications exist that allow researchers to find patterns in signals they could not otherwise see, such as trends in temperature or humidity datasets. Additionally, AI can generate data from outside sources, giving researchers more to work with for modeling and forecasting, Rogers wrote.
“Just like statistics and experimentation, AI is a tool that lets us do more with the observations we have,” Rogers wrote.
At the institute, Rogers wrote, researchers will work on a very comprehensive timeline, reporting progress to the NSF as they go.
“One of the key components of the institute is broadening AI participation and workforce development,” Rogers wrote. “I’m on a team that is working to bring education and training for AI to the workforce and the classroom; teaching about when and how AI can be applied, … so the next generation workforce is getting what they’ll need for their future careers.”
Understanding and using AI can be a steep learning curve at first, Rogers wrote. With the AI process, the answers a researcher gets might have more to do with how they ran the algorithm than the underlying science. Rogers wrote that Ebert-Uphoff and Barnes, along with Anderson, have extensive experience helping scientists understand AI.
“Properly-used, trustworthy AI could be a game changer in terms of collecting, analyzing and communicating our research results — that’s true for CIRA, for CSU and for the field in general,” Rogers wrote.
Serena Bettis can be reached at email@example.com or on Twitter @serenaroseb.