Spatiotemporal PM2.5 forecasting via dynamic geographical Graph Neural Network

With the growing interest in data-driven methods, Graph Neural Networks (GNNs) have demonstrated strong performance in ${\mathrm{PM}}_{2.5}$ forecasting as a deep learning architecture. However, GNN-based methods typically construct the graph based solely on the distance between stations, and few methods introduce geographical factors that significantly affect the spatial dispersion of ${\mathrm{PM}}_{2.5}$, leading to performance bottlenecks. Additionally, these methods often fail to process the dynamic wind–field data comprehensively, resulting in inaccurate ${\mathrm{PM}}_{2.5}$ dispersion graph construction. These shortcomings greatly limit the interpretability of GNN models in forecasting air pollution. To address these issues, we propose a deep learning method that combines Graph Convolution Network (GCN) with Long Short-Term Memory (LSTM), leveraging geographical information within a dynamic graph. The model captures spatial dependencies between ${\mathrm{PM}}_{2.5}$ monitoring stations using a dynamic directional graph derived from the wind–field data and a static graph to represent inherent geographical relationships. The combination of GCN and LSTM enables the extraction of both spatial and temporal correlations. The results of experiments suggest that our proposed model, which offers great interpretability, outperforms state-of-the-art methods, especially in 24, 30, and 36 hours forecasts.