Xi XiaoXi Xiao, a computer science doctoral student at the University of Alabama at Birmingham College of Arts and Sciences, developed a next-generation artificial intelligence model for climate science. The research paper, on which he is the only student author, won the Best Paper Award, the highest honor given at the world’s top-tier conference for high-performance computing, networking, storage and data analysis — 2025 Supercomputing Conference.
The award recognizes research that stands out for originality, innovation, impact and technical quality. Acceptance to the conference is highly competitive, with submissions from top national laboratories, major research institutions and computing teams from around the world.
Xiao’s climate model is called ORBIT-2, which he developed during an internship at Oak Ridge National Laboratory, the largest science and energy laboratory for the United States Department of Energy.
“This award reflects the strength of the collaboration and the growing role UAB researchers are playing in national computing efforts,” said Yuliang Zheng, Ph.D., chair of the Department of Computer Science. “It is exciting to see student contributions recognized at this level.”
At UAB, Xiao works under the supervision of Tianyang Wang, Ph.D., assistant professor, who focuses on applying AI and machine learning on interdisciplinary research.
Advancing climate science
Many traditional climate models struggle to show local weather patterns because their data is too broad. ORBIT-2 is an artificial intelligence model designed to turn low-resolution global climate and weather data into much more detailed, high-resolution predictions. It uses a process called downscaling, which sharpens large-scale climate data into more precise local information.
The research team built the model to run on tens of thousands of graphics processing units simultaneously.
“Exascale computing is among the fastest and most powerful computing available today,” Xiao said. “Because of this speed and scale, ORBIT-2 can reveal local climate patterns that were previously difficult or impossible to see.”
Real-world impact
High-resolution climate predictions will improve flood risk mapping, wildfire and heatwave forecasting, and disaster preparedness, especially for communities at risk.
“This level of detail is essential for planning how cities and regions prepare for climate-related challenges,” Xiao said.
The model can help governments, utility companies and city planners make better decisions about power grids, renewable energy, water resources, transportation systems and long-term infrastructure investments.
“The project closely aligns with U.S. Department of Energy priorities, which identify high-resolution climate prediction as a national research goal,” Xiao said. “ORBIT-2 shows that artificial intelligence models can run at extreme scales while remaining accurate, fast and energy efficient.”