GFS vs ECMWF Demand-Hub Outlook: 2026-05-30 00Z

Executive Summary

The 15-day outlook shows significant model disagreement, primarily centered on the Midwest and Northeast. The GFS projects a considerably colder and higher heating-demand scenario, forecasting more than double the national weighted HDDs of the ECMWF (15.9 vs 7.8). Conversely, the models are in closer alignment on cooling demand, with the ECMWF showing slightly more national CDDs (112.8 vs 108.2). Overall forecast confidence is low due to these regional divergences.

Hub-Level Weather

The primary drivers of the national forecast are the high-weight hubs in the Midwest and East, which is also where the models disagree most.

• Midwest (Chicago, 35% weight): This hub is the main point of contention. The GFS is dramatically colder, forecasting 16.0 HDDs and 64.6 CDDs. The ECMWF is much warmer, seeing only 1.9 HDDs and a robust 90.5 CDDs. The models show opposing solutions for the June 7-9 and June 12-13 periods.
• East (New York City, 30% weight): While closer on heating demand (GFS 9.8 HDDs vs ECMWF 7.7 HDDs), the GFS projects significantly more late-period warmth, resulting in 102.4 CDDs versus the ECMWF's 84.4 CDDs.
• Northeast (Boston, 10% weight): Similar to the Midwest, the GFS is much colder here, forecasting 73.6 HDDs compared to 48.5 HDDs from the ECMWF.
• South Central (Houston, 15% weight): Both models are in solid cooling season territory. The GFS is slightly hotter, projecting 270.7 CDDs against the ECMWF's 240.6 CDDs over the 15-day period.

Demand Implications (Natural Gas)

The conflicting model solutions present two distinct paths for natural gas demand in early June.

• Heating Demand: The GFS outlook implies lingering residential/commercial heating demand in the key consuming regions of the Midwest and Northeast. This would be supportive of prices compared to the ECMWF, which largely ends heating demand for the season. The national difference of ~8.1 weighted HDDs is material.
• Cooling Demand: Both models indicate a steady build in cooling demand and associated power burn, which is typical for this time of year. The total 15-day load is comparable between models, suggesting agreement on the general trend of summer demand kicking in, especially in the South.

The GFS has trended significantly warmer since its prior run (+10.9 weighted CDDs), while the ECMWF showed a minor cooler trend (-0.6 weighted CDDs), indicating the models are moving in opposite directions and increasing forecast uncertainty.

Model Agreement & Caveats

Forecast confidence is low. The convergence score of 0.5865 reflects the major disagreements in the Midwest and Northeast. The large divergence often occurs in the 8-15 day range, where model accuracy naturally degrades.

• Run-over-Run Shifts: The GFS trended materially hotter in its 00Z cycle (+10.9 weighted CDDs) compared to its prior run, while also adding a small amount of heating demand (+0.9 weighted HDDs). The ECMWF was more stable, with minor cooler trends. This model divergence run-over-run warrants caution.
• Source Data: This analysis is based on raw, unprocessed model output from Open-Meteo for specific city grid points. These points are proxies for wider demand regions and may not perfectly capture the integrated regional load.