A Graph Neural Network Approach for Localized and High-Resolution Temperature Forecasting

2025-11-29

AI Summary

Contemporary numerical weather prediction models fail to resolve urban microscale extreme heat variations, resulting in delayed heatwave warnings for vulnerable populations. To address this, we propose a high-resolution spatiotemporal temperature forecasting framework based on graph neural networks (GNNs). The method constructs a dynamic geographical graph structure to capture localized spatial dependencies and integrates multi-source spatiotemporal features within a 24-hour sliding window to enable fine-grained 1–48 hour forecasts. It supports transfer learning for data-scarce regions and enables multi-step prediction. Evaluated over southwestern Ontario, Canada, the model achieves an overall mean absolute error (MAE) of 1.93°C and a 48-hour single-step MAE of 2.93°C—substantially outperforming conventional meteorological models. This work provides a scalable, deployable technical pathway for early urban heat-risk warning systems.

Technology Category

• Planning, Routing, and Scheduling: Optimization of Spatio-temporal Systems
• Machine Learning: Graph-based Machine Learning
• Knowledge Representation and Reasoning: Geometric, Spatial, and Temporal Reasoning

Application Category

• Graph Algorithms and Modeling for the Web: Graph neural networks and deep learning approaches for Web-related graphs
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Web Mining and Content Analysis: Large pretrained models with web data

Abstract

Heatwaves are intensifying worldwide and are among the deadliest weather disasters. The burden falls disproportionately on marginalized populations and the Global South, where under-resourced health systems, exposure to urban heat islands, and the lack of adaptive infrastructure amplify risks. Yet current numerical weather prediction models often fail to capture micro-scale extremes, leaving the most vulnerable excluded from timely early warnings. We present a Graph Neural Network framework for localized, high-resolution temperature forecasting. By leveraging spatial learning and efficient computation, our approach generates forecasts at multiple horizons, up to 48 hours. For Southwestern Ontario, Canada, the model captures temperature patterns with a mean MAE of 1.93$^{circ}$C across 1-48h forecasts and MAE@48h of 2.93$^{circ}$C, evaluated using 24h input windows on the largest region. While demonstrated here in a data-rich context, this work lays the foundation for transfer learning approaches that could enable localized, equitable forecasts in data-limited regions of the Global South.

Problem

• Develops a Graph Neural Network for localized, high-resolution temperature forecasting.
• Addresses the failure of current models to capture micro-scale temperature extremes.
• Aims to enable equitable early warnings for vulnerable populations in data-limited regions.

Innovation

• Graph Neural Network for localized temperature forecasting
• Spatial learning enables high-resolution multi-horizon predictions
• Transfer learning foundation for data-limited region applications