GFS vs ECMWF Demand-Hub Outlook: 2026-09-13T00:00Z

Executive Summary

Models show significant disagreement over the 15-day forecast, with the GFS much colder and the ECMWF warmer. The GFS projects a national demand-weighted total of 25.8 HDDs, driven by an early-season cold shot in the Midwest and East, while the ECMWF sees only 5.3 HDDs and stronger cooling demand (101.5 CDDs vs. GFS's 87.3). Confidence in the outlook is low due to this divergence and notable model instability since the prior cycle.

Hub-Level Weather

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

• Midwest (Chicago, 35% wt.): This hub is the epicenter of the model disagreement. The GFS forecasts a sharp cooldown in the latter half of the period (Sep 21-25), accumulating 37.4 HDDs. The ECMWF keeps this region much milder, forecasting only 2.9 HDDs for the entire 15-day period and significantly more cooling demand (67.9 CDDs vs. GFS's 29.8).
• East (New York City, 30% wt.): A similar, though less extreme, pattern is evident here. The GFS projects 18.1 HDDs, concentrated after Sep 22, while the ECMWF sees only 4.3 HDDs and stronger overall cooling load.
• South Central (Houston, 15% wt.): Both models forecast persistent, strong cooling demand with zero HDDs. The GFS is notably hotter, projecting 295.7 CDDs compared to the ECMWF's 260.6 CDDs.

Run-over-run deltas show the models moving in opposite directions. The GFS trended significantly colder (+11.6 weighted HDDs), while the ECMWF trended warmer (-6.4 weighted HDDs, +14.9 weighted CDDs) compared to the prior cycle.

Demand Implications (Natural Gas)

The conflicting forecasts present two very different scenarios for Natural Gas demand.

• The ECMWF solution suggests a continuation of late-summer weather patterns, supporting above-average power burn for cooling, which would be a status-quo to slightly supportive price signal for NG.
• The GFS outlook, if it verifies, would introduce the first significant residential/commercial heating demand of the season into key consuming regions. An earlier-than-normal start to the heating season would be a decidedly bullish catalyst.

The market is faced with significant weather uncertainty, pivoting between lingering cooling demand and a potential early start to heating demand.

Model Agreement & Caveats

Overall model confidence is low. The convergence score is 0.388, indicating poor agreement between the GFS and ECMWF. This score has decreased by 0.16 since the prior run, highlighting growing instability and divergence.

The primary caveat is the major disagreement over the timing and intensity of a potential cool-down across the northern U.S. in the 8-15 day period. When models diverge this significantly, especially in opposite directions run-over-run, neither solution should be trusted with high confidence until subsequent runs show better agreement.

This forecast uses a limited set of demand hubs and may not perfectly represent total U.S. load. Model performance can vary, and this data is provided by Open-Meteo.