Hybridizing deep learning models and a chemical transport model for medium-term PM2.5 forecasts in the Yangtze River Delta, China

Key Points

• Enhanced forecasting accuracy was achieved for PM2.5 levels using a hybrid model approach.
• The model effectively combines deep learning with traditional chemical transport methods for better predictions.
• Observation period included multiple seasonal variations to assess model robustness across different environments.
• Implications suggest that hybrid modeling could be crucial for improving air quality management strategies.