A graduate student at Algoma University in Sault Ste. Marie has developed a new computer program that uses artificial intelligence to predict how Canada’s forests might die off as the climate changes. James Gu, a student in the school’s Master of Computer Science (Thesis-Based) program, presented his work on May 7, 2026, at the Science Carnival hosted at the Canadian Bushplane Heritage Centre.
Gu worked under the supervision of Dr. Rashid Khokhar, an assistant professor in the university’s Faculty of Computer Science and Technology. The project aims to provide better tools for understanding forest health by using a neural network-based forecasting system enhanced with graph neural network structures to identify complex relationships between various environmental factors that lead to tree loss.
To train the model, the team combined forestry data from the Ontario Forest Research Institute and the provincial Ministry of Natural Resources and Forestry with public climate datasets covering the years 1952 to 2024. Researchers point to historical events, such as the 2001-2002 drought in western Canada that resulted in the widespread death of aspen trees, as a primary reason why such predictive tools are needed to help prepare for future environmental shifts.
The system is currently undergoing stability testing to ensure the results remain accurate even if the data it receives is altered. Initial tests suggest the model shows a degree of stability, indicating potential robustness, though further validation is ongoing — a key step toward making it useful for those who manage and protect Canada’s forests.
