Air Quality Prediction with A Meteorology-Guided Modality-Decoupled Spatio-Temporal Network

2025-04-14

AI Summary

Existing air quality forecasting models commonly overlook the critical role of atmospheric conditions, failing to capture the dynamic coupling between meteorology and pollution. To address this, we propose the Meteorology-Guided Modality-Decoupled Spatiotemporal Network (MDSN), the first framework to explicitly decouple and jointly model pollutant observations and multi-pressure-level meteorological data, thereby characterizing atmospheric–pollutant diffusion and chemical transformation dependencies in depth. Methodologically, MDSN integrates multimodal spatiotemporal graph neural networks, dynamic graph structure learning, and weather forecast information embedding. We further introduce ChinaAirNet—the first nationwide, multi-pressure-level meteorology–air quality fused dataset. On ChinaAirNet, MDSN achieves a 17.54% reduction in 48-hour forecasting error over state-of-the-art methods. Both code and dataset will be publicly released.

Technology Category

• Machine Learning: Multimodal Learning
• Computer Vision: Multi-modal Vision
• Data Mining & Knowledge Management: Mining of Spatial, Temporal or Spatio-Temporal Data

Application Category

• Graph Algorithms and Modeling for the Web: Graph neural networks and deep learning approaches for Web-related graphs
• Web Mining and Content Analysis: Web data integration and cleaning
• Systems and Infrastructure for Web, Mobile and WoT: Applied ML and AI for Web-based mobile applications

Abstract

Air quality prediction plays a crucial role in public health and environmental protection. Accurate air quality prediction is a complex multivariate spatiotemporal problem, that involves interactions across temporal patterns, pollutant correlations, spatial station dependencies, and particularly meteorological influences that govern pollutant dispersion and chemical transformations. Existing works underestimate the critical role of atmospheric conditions in air quality prediction and neglect comprehensive meteorological data utilization, thereby impairing the modeling of dynamic interdependencies between air quality and meteorological data. To overcome this, we propose MDSTNet, an encoder-decoder framework that explicitly models air quality observations and atmospheric conditions as distinct modalities, integrating multi-pressure-level meteorological data and weather forecasts to capture atmosphere-pollution dependencies for prediction. Meantime, we construct ChinaAirNet, the first nationwide dataset combining air quality records with multi-pressure-level meteorological observations. Experimental results on ChinaAirNet demonstrate MDSTNet's superiority, substantially reducing 48-hour prediction errors by 17.54% compared to the state-of-the-art model. The source code and dataset will be available on github.

Problem

• Predicting air quality by modeling spatio-temporal and meteorological interactions
• Overcoming underestimation of atmospheric conditions in existing prediction models
• Integrating multi-pressure-level weather data for accurate pollution dependency analysis

Innovation

• Modality-decoupled spatio-temporal network for air quality
• Multi-pressure-level meteorological data integration
• Nationwide dataset combining air and weather data