The National Oceanic and Atmospheric Administration (NOAA) has announced a significant leap forward in weather prediction with the deployment of a new generation of artificial intelligence (AI)-driven global weather models. This groundbreaking initiative promises faster, more efficient, and remarkably accurate forecasts, utilising a fraction of the computational resources previously required.
Key Takeaways
- NOAA has launched three new AI-driven global weather models: AIGFS, AIGEFS, and HGEFS.
- These models offer improved forecast accuracy, particularly for large-scale weather patterns and tropical tracks.
- Significant reductions in computational resource usage and faster delivery of forecast products are key benefits.
- The development is a result of Project EAGLE, a collaborative effort involving NOAA and industry partners.
A New Era in Weather Forecasting
NOAA Administrator Dr. Neil Jacobs hailed the deployment as a "significant leap forward in American weather model innovation." He emphasised that these AI models represent a "new paradigm" for NOAA, enabling enhanced accuracy for large-scale weather events and tropical storm tracks. Furthermore, the models will deliver forecast products to meteorologists and the public more rapidly and at a reduced cost due to drastically lower computational expenses.
The Three Pillars of AI Forecasting
The new suite comprises three distinct AI applications:
- AIGFS (Artificial Intelligence Global Forecast System): This model leverages AI to deliver weather forecasts more quickly and efficiently than traditional systems, using up to 99.7% fewer computing resources. A single 16-day forecast can be completed in approximately 40 minutes, a substantial reduction in latency.
- AIGEFS (Artificial Intelligence Global Ensemble Forecast System): An AI-based ensemble system, AIGEFS provides a range of probable forecast outcomes, offering meteorologists and decision-makers a broader perspective. Early results indicate improved performance over the traditional GEFS, extending forecast skill by an additional 18 to 24 hours.
- HGEFS (Hybrid-GEFS): This pioneering "grand ensemble" model is a first-of-its-kind for an operational weather centre. It uniquely combines the AI-based AIGEFS with NOAA's flagship Global Ensemble Forecast System (GEFS). Initial testing shows that HGEFS consistently outperforms both AI-only and physics-only ensemble systems.
Enhanced Accuracy and Efficiency
The AIGFS model demonstrates improved forecast skill for many large-scale features and notably reduces tropical cyclone track errors at longer lead times. While v1.0 shows a degradation in tropical cyclone intensity forecasts, future versions are expected to address this. The AIGEFS offers comparable forecast skill to the operational GEFS but requires only 9% of its computing resources. The HGEFS, by blending physics-based and AI-based systems, creates a more robust ensemble that better represents forecast uncertainty, leading to superior performance across most verification metrics.
Collaborative Innovation
This advancement is an outgrowth of Project EAGLE, a joint initiative involving NOAA's National Weather Service, Oceanic and Atmospheric Research labs, the Environmental Modeling Center, and the Earth Prediction Innovation Center. NOAA scientists collaborated with academia and private industry, leveraging Google DeepMind's GraphCast model as an initial foundation and fine-tuning it with NOAA's own Global Data Assimilation System analyses. This collaborative approach underscores NOAA's commitment to continuous innovation in forecasting technology.