TopoFlow: Physics-guided Neural Networks for high-resolution air quality prediction
arXiv:2602.16821v1 Announce Type: new Abstract: We propose TopoFlow (Topography-aware pollutant Flow learning), a physics-guided neural network for efficient, high-resolution air quality prediction. To explicitly embed physical processes into the learning framework, we identify two critical factors governing pollutant dynamics: topography and wind direction. Complex terrain can channel, block, and trap pollutants, while wind acts as a primary driver of their transport and dispersion. Building on these insights, TopoFlow leverages a vision transformer architecture with two novel mechanisms: topography-aware attention, which explicitly models terrain-induced flow patterns, and wind-guided patch reordering, which aligns spatial representations with prevailing wind directions. Trained on six years of high-resolution reanalysis data assimilating observations from over 1,400 surface monitoring stations across China, TopoFlow achieves a PM2.5 RMSE of 9.71 ug/m3, representing a 71-80% improve
arXiv:2602.16821v1 Announce Type: new Abstract: We propose TopoFlow (Topography-aware pollutant Flow learning), a physics-guided neural network for efficient, high-resolution air quality prediction. To explicitly embed physical processes into the learning framework, we identify two critical factors governing pollutant dynamics: topography and wind direction. Complex terrain can channel, block, and trap pollutants, while wind acts as a primary driver of their transport and dispersion. Building on these insights, TopoFlow leverages a vision transformer architecture with two novel mechanisms: topography-aware attention, which explicitly models terrain-induced flow patterns, and wind-guided patch reordering, which aligns spatial representations with prevailing wind directions. Trained on six years of high-resolution reanalysis data assimilating observations from over 1,400 surface monitoring stations across China, TopoFlow achieves a PM2.5 RMSE of 9.71 ug/m3, representing a 71-80% improvement over operational forecasting systems and a 13% improvement over state-of-the-art AI baselines. Forecast errors remain well below China's 24-hour air quality threshold of 75 ug/m3 (GB 3095-2012), enabling reliable discrimination between clean and polluted conditions. These performance gains are consistent across all four major pollutants and forecast lead times from 12 to 96 hours, demonstrating that principled integration of physical knowledge into neural networks can fundamentally advance air quality prediction.
Executive Summary
This article proposes TopoFlow, a physics-guided neural network for high-resolution air quality prediction. TopoFlow integrates topography and wind direction into its architecture, achieving a significant improvement in PM2.5 prediction compared to operational forecasting systems and AI baselines. The model demonstrates reliable performance across various pollutants and forecast lead times, indicating the potential of principled integration of physical knowledge into neural networks for air quality prediction. The findings have significant implications for improving air quality forecasting and decision-making in regions with complex terrain, such as China.
Key Points
- ▸ TopoFlow integrates topography and wind direction into its neural network architecture
- ▸ Achieves a 71-80% improvement over operational forecasting systems and a 13% improvement over AI baselines
- ▸ Reliable performance across various pollutants and forecast lead times
Merits
Strength in Physical Knowledge Integration
The article effectively embeds physical processes into the learning framework, leveraging topography and wind direction to improve air quality prediction. This principled integration of physical knowledge into neural networks demonstrates a significant advancement in the field.
Demerits
Limited Generalizability
The study focuses on a specific region (China) and a limited dataset, which may limit the generalizability of the findings to other regions with different terrain and climate characteristics.
Expert Commentary
The article presents a significant advancement in air quality prediction, leveraging the principled integration of physical knowledge into neural networks. The findings demonstrate the potential of TopoFlow to improve air quality forecasting, particularly in regions with complex terrain. However, the study's limited generalizability and focus on a specific region and dataset may limit the scope of the findings. Further research is needed to explore the applicability of TopoFlow in other regions and to evaluate its performance in different environmental contexts.
Recommendations
- ✓ Future studies should investigate the applicability of TopoFlow in regions with different terrain and climate characteristics to evaluate its generalizability and transferability.
- ✓ The development of more accurate and reliable air quality prediction systems, informed by the findings of this study, can have significant implications for policy decisions related to air quality management and pollution control.
