NCShare Helps NC State University Researchers Analyze Ice-Rock Avalanche in Nepal

October 21, 2026

In September 2026, the North Carolina State University AI Hub for Science used NCShare resources to investigate an ice-rock avalanche on the Nepal-Tibet border that occurred on Aug. 26. Paul Liu, director of the AI Hub for Science, director of international affairs in the college of sciences, faculty fellow at the center for geospatial analytics, and professor in the department of marine, earth, and atmospheric sciences at NC State, led the effort. Liu began using NCShare in 2025 by training language models and conducting severe weather forecasting simulations. 

The avalanche, containing roughly 200 million cubic meters of ice and rock, traveled 170 kilometers. Liu’s team hindcast the event with an open-source three-phase mass-flow model on the NCShare high-performance computing cluster. The researchers collected critical data on how the flow occurred and determined why on-site gauges were destroyed. 

“During the simulation process, we can understand the underlying driving mechanisms by adjusting variables like initial mass volume falling down and the speed of debris flow,” Liu said. “Each simulation required 13 hours of dedicated computing time. Luckily, NCShare provided access to that computing power. I submitted many simulations and received very encouraging results.” 

NCShare resources helped Liu’s team process large amounts of data efficiently, saving them time and money.

“NCShare gives us the freedom to test innovative ideas during emergency events immediately,” Liu said. “Beyond running simulations, being able to run jobs quickly on NCShare allowed us to refine parameters rapidly… Furthermore, the original open-source package relied on serial computing, taking 13 hours per run. We converted the code from serial to parallel computing across multi-core, multi-node setups on NCShare, reducing the run time from 13 hours to 1.5 hours.”

Looking ahead, Liu said he submitted a National Science Foundation proposal to use NCShare to analyze low-oxygen dead zones in the Gulf of Mexico using neural networks. Liu also plans to use NCShare to monitor and predict harmful algal blooms in Falls Lake, North Carolina using AI.