GFS vs ECMWF Demand-Hub Outlook: 2026-05-12T00:00Z

Executive Summary

The GFS and ECMWF models present a strongly divergent 15-day outlook, reducing overall forecast confidence. The GFS projects a much warmer national picture, forecasting significantly less heating demand (35.1 weighted HDDs) and substantially more cooling demand (95.3 weighted CDDs) than the ECMWF (58.6 weighted HDDs, 59.1 weighted CDDs). This divergence is driven primarily by the GFS forecasting a major warm-up in the East from May 20th onward, a solution that represents a massive warmer shift from its previous run.

Hub-Level Weather

The national forecast divergence is almost entirely driven by conflicting outlooks for the high-demand Eastern and Midwestern hubs beyond the first week.

• East (New York City, 30% weight): This hub shows the most extreme disagreement. The GFS forecasts a cumulative 27.2 HDDs and 88.8 CDDs, while the ECMWF sees a much colder 69.1 HDDs and only 20.0 CDDs. The split occurs after May 19th, with the GFS calling for daily mean temperatures in the upper 70s while the ECMWF brings in a cool down with means in the upper 50s.
• Midwest (Chicago, 35% weight): A significant, though less extreme, disagreement is also present here. The ECMWF is colder with 76.3 cumulative HDDs versus the GFS's 54.6 HDDs. This difference is concentrated in the May 19-22 period, where the ECMWF forecasts a notable late-season cold shot that the GFS misses.
• South Central (Houston, 15% weight): Both models show consistent cooling demand, but the GFS is materially hotter, accumulating 240.7 CDDs compared to the ECMWF's 182.6 CDDs.

Demand Implications (Natural Gas)

The conflicting model solutions imply very different near-term natural gas demand scenarios.

• GFS Scenario: The GFS outlook suggests a rapid transition from heating to cooling demand. The minimal heating load (35.1 wHDDs) would be quickly replaced by significant, above-normal cooling demand (95.3 wCDDs), driven by strong power burn in the East and South. This is a bullish weather-demand forecast.
• ECMWF Scenario: The ECMWF suggests a more tepid demand environment. It maintains a notable block of late-season heating demand (58.6 wHDDs) in key population centers while showing a much more modest and near-normal start to cooling season (59.1 wCDDs). This is a more neutral to bearish weather-demand forecast compared to the GFS.

Model Agreement & Caveats

Confidence in the 15-day forecast is low due to poor model agreement and significant GFS instability.

• Convergence: The convergence score of 0.31 indicates a high degree of disagreement between the models, particularly in the 8-15 day period.
• Model Instability: The run_deltas show a very large run-over-run shift in the GFS model, which added +30.7 weighted CDDs and removed -32.2 weighted HDDs versus its prior cycle. The ECMWF, by contrast, was almost unchanged. Such a dramatic single-run shift in the GFS reduces confidence in its new, hotter solution and suggests it may be volatile in subsequent runs.
• Source Limitations: This analysis is based on deterministic model runs from Open-Meteo. A review of ensemble guidance would be necessary to assess the full range of potential outcomes and the probability of either solution verifying.