Enhancing Graph Neural Networks Using Proximity Graphs for Dust Source Emission Forecasting

2026-06-18

AI Summary

This study addresses the challenge of modeling the complex spatiotemporal dynamics of dust emission sources, which traditional methods struggle to capture effectively. To this end, it systematically introduces multiple geometric proximity graphs—including Delaunay triangulation, Gabriel graphs, k-nearest neighbor graphs, and Yao graphs—into graph neural networks (GNNs) as topological structures for message passing, thereby explicitly encoding spatiotemporal dependencies among dust sources. Experimental results demonstrate that proximity-based GNNs (e.g., GraphSAGE, GCN, and GAT) significantly outperform both GNNs using random graphs and conventional time-series models such as LSTM on the task of dust emission prediction, confirming the effectiveness and novelty of the proposed approach.

Abstract

Accurate prediction of dust source emissions is critical for mitigating the significant environmental and health hazards posed by dust storms. Traditional forecasting methods often struggle to capture the complex spatiotemporal dynamics of these phenomena. In this paper, we demonstrate that proximity graphs enable Graph Neural Networks (GNNs) to effectively model the intricate spatial and temporal relationships between data points. Specifically, we use proximity graphs--such as Delaunay triangulation, Gabriel graph, k-Nearest Neighbor graph, and Yao graph--as the input for GNNs (including GraphSAGE, Graph Convolutional Networks, and Graph Attention Networks) to perform message passing. Our approach highlights the effectiveness of integrating proximity graphs with GNNs for robust and accurate dust source forecasting. To emphasize the importance of proximity graph representations, we compare our method against GNNs using random graphs for message passing. The results show that GNNs with proximity graphs significantly outperform those with random graphs and are also far superior to Long Short-Term Memory (LSTM) model in dust source emission forecasting.

Problem

• dust source emission forecasting
• spatiotemporal dynamics
• Graph Neural Networks
• proximity graphs
• environmental hazards

Innovation

• proximity graphs
• Graph Neural Networks
• dust source emission forecasting
• spatiotemporal modeling
• message passing