University of Manchester Uses NVIDIA Earth-2 to Forecast Air Pollution Across the UK

University of Manchester researchers have used NVIDIA's Earth-2 family of open AI models to forecast air pollution across the United Kingdom. Air pollution contributes to an estimated 30,000 deaths in the U.K. annually, making it a serious public health risk. Traditional chemistry-based models to study air quality are computationally expensive, limiting their detail and frequency.

David Topping, a professor at the University of Manchester's department of Earth and environmental science, found that NVIDIA's Earth-2 models could be adapted for pollution forecasting. Collaborating with the NVIDIA Earth-2 team, Topping and colleagues generated training data from existing chemistry-climate simulations and trained Earth-2 CorrDiff, a generative downscaling model, on Isambard-AI, the U.K.'s national AI supercomputer in Bristol.

The model worked successfully on its first attempt. Topping emphasized the importance of understanding the impact of environmental stressors in the air we breathe to improve human health. With the U.K.-wide pollution model, researchers can simulate potential future scenarios, such as predicting the effects of different pollution-related government policy changes.

This model can also be integrated with data from edge AI devices to provide real-time air quality insights, such as proactive alerts for healthcare organizations. The team plans to release open source training data and workflows, allowing researchers worldwide to create detailed pollution models using their local data. In the future, Topping envisions an agentic interface where a clinician or government agency can ask the model about air pollution in a specific neighborhood, and the system will provide an answer based on the underlying scientific frameworks.