Researchers benchmark air quality forecasting models for Dallas

Researchers tested whether adding physics rules to machine learning models improves air quality predictions in Dallas using EPA data from 2022–2024. The physics-guided models predicted pollutant levels more accurately and consistently with real-world chemistry, especially for short-term forecasts, suggesting cities should use this approach when deploying automated air quality warning systems.

This is a technical benchmarking paper with no new policy, infrastructure deployment, or structural change — it's a comparison of academic forecasting methods on historical data with no evidence of actual adoption or real-world impact.