GFS vs ECMWF Demand-Hub Outlook: 2026-09-19T00:00Z

Executive Summary

The 15-day outlook shows a seasonal transition from late-season cooling demand to early-season heating load. Models are in broad agreement on the total national demand picture but diverge significantly on the regional composition, with ECMWF forecasting a colder Midwest while GFS is colder in the East. Overall forecast confidence is low, as reflected in the poor convergence score (0.396), and both models trended colder compared to their prior runs.

Hub-Level Weather

The national forecast is primarily driven by the two largest hubs, which show opposing trends:

• Midwest (Chicago, 35% weight): This is the main point of contention. The ECMWF is significantly colder, projecting 67.8 HDDs over the 15-day period, compared to just 48.0 HDDs from the GFS. This difference is most pronounced in the 6-10 day range.
• East (New York City, 30% weight): The models are inverted here, with the GFS showing a colder outlook at 44.4 HDDs versus the ECMWF's 37.3 HDDs. The divergence grows in the back half of the forecast (post Sep 27), where the ECMWF shows a significant warming trend that the GFS lacks.
• South Central (Houston, 15% weight): Both models show strong cooling demand, with GFS at 285.6 CDDs and ECMWF at 290.0 CDDs. The GFS solution turns markedly cooler in the final days of the forecast.
• South (Atlanta, 10% weight): The GFS is much hotter here, forecasting 140.2 CDDs against the ECMWF's 108.4 CDDs.

Demand Implications (Natural Gas)

The forecast implies a transition period for Natural Gas consumption, with power burn for cooling in the South being replaced by emerging residential/commercial heating demand in the North.

• National Demand: Total weighted degree days are similar between models (GFS: 101.7, ECMWF: 105.8), but the composition differs. The ECMWF's colder Midwest view implies stronger early heating demand in a critical consumption region.
• Run-over-Run Shift: Both models trended colder versus the prior cycle. The GFS saw a material shift, adding +6.06 weighted HDDs and shedding -8.36 weighted CDDs. This colder trend exerts modestly bullish pressure on near-term NG futures contracts.

Model Agreement & Caveats

Forecast confidence is low due to significant model disagreement on the evolution of temperatures in key demand hubs.

• Convergence Score: The model agreement score is 0.396, indicating poor convergence. This score deteriorated slightly from the previous run (delta of +0.045), signaling that the models moved further apart.
• Divergence: The primary disagreements occur in the 8-15 day period, particularly regarding the strength and timing of cooler air masses entering the Midwest and East. This reduces the reliability of the long-range signal.
• Source Limitations: This analysis is based on a limited set of demand hubs from Open-Meteo and serves as a proxy for, not a replacement of, a full U.S. population-weighted demand forecast.