GFS vs ECMWF Demand-Hub Outlook: 2026-05-16 00Z

Executive Summary

The 15-day outlook shows a significant divergence between models, with the GFS forecasting a much stronger start to the cooling season. Nationally, the GFS projects 99.0 weighted CDDs, considerably higher than the ECMWF's 87.5. This bullish GFS signal is driven by substantial warmth in the East and South Central hubs. Confidence in the forecast is low, reflected by a poor convergence score of 0.47, despite both models trending warmer compared to the prior cycle.

Hub-Level Weather

The primary drivers of the national forecast are the high-weighted Midwest and East Coast hubs, where the models show considerable disagreement, particularly in the 8-15 day period.

• East (New York City, 30% weight): This hub is the main source of divergence. The GFS is substantially hotter, forecasting 92.2 cumulative CDDs versus only 67.1 from the ECMWF. The GFS sees a late-period heat event that the ECMWF does not.
• Midwest (Chicago, 35% weight): In contrast to the East, the ECMWF is warmer here, projecting 69.4 CDDs against the GFS's 54.7. The ECMWF maintains mild conditions late in the forecast period while the GFS brings in a significant cool down around May 26-27.
• South Central (Houston, 15% weight): Both models show strong cooling demand, but the GFS is notably hotter, with 239.8 CDDs compared to the ECMWF's 194.2.

Demand Implications (Natural Gas)

The forecast reflects a transition from heating to cooling demand, typical for mid-May. The key takeaway is the difference in the intensity of early summer power burn for natural gas.

• GFS Scenario: The 99.0 weighted CDDs suggest a stronger-than-normal start to cooling demand, which would be bullish for gas prices by increasing power sector consumption.
• ECMWF Scenario: The 87.5 weighted CDDs and slightly higher weighted HDDs (23.1 vs 18.6 for GFS) imply a more moderate, less impactful weather pattern for gas demand.
• Overall, heating demand is minimal in both models, with the focus squarely on the magnitude of emerging cooling load.

Model Agreement & Caveats

Confidence in this 15-day outlook should be low due to significant model disagreement.

• Convergence: The model convergence score is 0.47, indicating poor agreement. This is primarily driven by starkly different solutions for the weather pattern in the eastern U.S. after day 8.
• Run-over-Run Deltas: Both models trended warmer versus the prior 18Z cycle. The GFS saw a major bullish shift, adding 12.6 weighted CDDs and shedding 17.1 weighted HDDs. The ECMWF's shift was more modest (+7.8 CDDs, -4.2 HDDs). This large GFS jump may indicate model instability.
• Source Limitations: This analysis uses Open-Meteo city-point forecasts as a proxy. These are not official, population-weighted NOAA degree day products and should be interpreted as directional indicators for demand.