UT-GraphCast Hindcast Dataset: A Global AI Forecast Archive from UT Austin for Weather and Climate Applications

2025-06-20

AI Summary

To address the scarcity of long-term, high-resolution global weather and climate forecast data, this study develops an AI-driven global reforecast dataset spanning 1979–2024. Leveraging the physics-informed graph neural network GraphCast, the model is trained on ERA5 reanalysis data to produce deterministic 15-day forecasts initialized daily at 00 UTC, resolving the globe at ~25 km horizontal resolution across 37 vertical levels. This constitutes the first publicly available, 45-year medium-range AI reforecast archive, with full-atmosphere inference completed in under one minute per forecast. The dataset includes over ten essential meteorological variables—including temperature, wind components, and geopotential height—with substantially enhanced spatiotemporal resolution. It enables climate attribution, numerical model verification, predictability studies, and the establishment of AI-weather benchmarks.

Technology Category

• Computer Vision: Remote Sensing / Geospatial AI
• Knowledge Representation and Reasoning: Geometric, Spatial, and Temporal Reasoning
• Humans and AI: Human-Aware Planning and Behavior Prediction

Application Category

• Graph Algorithms and Modeling for the Web: Foundation models and LLMs for Web-related graphs
• Search and Retrieval-Augmented AI: Vertical and domain-specific search
• Web Mining and Content Analysis: Large pretrained models with web data

Abstract

The UT GraphCast Hindcast Dataset from 1979 to 2024 is a comprehensive global weather forecast archive generated using the Google DeepMind GraphCast Operational model. Developed by researchers at The University of Texas at Austin under the WCRP umbrella, this dataset provides daily 15 day deterministic forecasts at 00UTC on an approximately 25 km global grid for a 45 year period. GraphCast is a physics informed graph neural network that was trained on ECMWF ERA5 reanalysis. It predicts more than a dozen key atmospheric and surface variables on 37 vertical levels, delivering a full medium range forecast in under one minute on modern hardware.

Problem

• Global weather forecast archive for 45 years
• Physics-informed AI model predicts atmospheric variables
• Fast medium-range forecasts on modern hardware

Innovation

• Physics-informed graph neural network
• Trained on ECMWF ERA5 reanalysis
• Fast medium-range forecast generation