GFS vs ECMWF Demand-Hub Outlook: 2026-05-03T00:00Z

Executive Summary

The 00z model runs offer a conflicting 15-day outlook, with the ECMWF forecasting higher national heating demand while the GFS projects stronger early-season cooling demand. A significant bearish shift in the latest GFS run reduces confidence, with key disagreements centered on the intensity of late-season cold in the East (ECMWF colder) and the strength of heat in the South (GFS hotter).

Hub-Level Weather

The national demand picture is driven by the Midwest and East, which show material model disagreement.

• East (New York City): The largest divergence is in the East, where the ECMWF is significantly colder. Over the 15-day period, it forecasts 119.6 cumulative HDDs for NYC (30% weight) versus just 85.2 HDDs in the GFS. This points to a more prolonged heating season in the ECMWF solution.
• South Central (Houston): Conversely, the GFS is substantially hotter in the key cooling market of Houston (15% weight). The GFS projects 157.9 cumulative CDDs, a notable 46.4 CDDs more than the ECMWF's 111.5 CDDs, signaling a more robust start to cooling demand.
• Midwest (Chicago): In the highest-weighted hub (35%), the cumulative HDD totals are nearly identical (GFS 156.5 vs. ECMWF 156.4). However, the models diverge on timing, particularly in the day 8-15 period, contributing to forecast uncertainty.

Demand Implications (Natural Gas)

The competing forecasts present a mixed signal for national gas demand during this shoulder season.

• Heating Demand: The ECMWF is more supportive of residential/commercial heating demand, projecting 111.2 demand-weighted HDDs nationally, compared to 98.4 weighted HDDs from the GFS.
• Cooling Demand: The GFS implies stronger power burn for cooling, forecasting 37.1 weighted CDDs versus the ECMWF's 30.4. This difference is driven almost entirely by the GFS's hotter outlook for the South.

Overall, the ECMWF supports lingering heating demand, while the GFS suggests a faster transition to summer cooling load.

Model Agreement & Caveats

Forecast confidence is moderate due to material model differences, particularly in the latter half of the outlook.

• Convergence: The forecast has a convergence score of 0.25, indicating moderate disagreement. The weighted mean absolute HDD difference is 2.9 per day.
• Run-over-Run Trends: The latest GFS run showed a significant bearish trend, shedding 34.7 weighted HDDs from its prior forecast. In contrast, the ECMWF remained very stable, with a minor change of -0.7 HDDs. This large corrective shift in the GFS is a key source of uncertainty.
• Source Limitations: This analysis is based on deterministic GFS and ECMWF model runs provided by Open-Meteo. A full analysis should incorporate ensemble guidance to better assess the range of potential outcomes.