GFS vs ECMWF Demand-Hub Outlook: 00z Sep 04, 2026

Executive Summary

Models show significant disagreement in the 15-day outlook, with the GFS forecasting a much hotter, more demand-intensive period than the ECMWF. The GFS projects 204.0 total weighted CDDs versus the ECMWF's 151.0. This divergence is driven almost entirely by the back half of the forecast (days 10-15), where the GFS shows a late-season heatwave across the key Midwest and Eastern demand hubs, a pattern the ECMWF does not see.

Hub-Level Weather

The primary drivers of the national forecast divergence are the highest-weighted demand centers:

• Midwest (Chicago, 35% weight): This hub shows the largest split. The GFS forecasts 211.9 cumulative CDDs, a stark contrast to the ECMWF's 149.1. The difference is extreme late in the period; for September 15th, the GFS forecasts a hot 24.7 CDDs while the ECMWF sees a mild 4.7 CDDs.
• East (New York City, 30% weight): A similar pattern exists here, with the GFS projecting 153.9 CDDs versus the ECMWF's 93.7. The GFS sees a very warm end to the forecast, while the ECMWF trends toward more seasonal temperatures.
• South / South Central (Houston & Atlanta, 25% combined weight): Both models show strong cooling demand, but the GFS is consistently hotter. Houston shows 325.2 CDDs (GFS) vs. 269.5 (ECMWF), and Atlanta shows 309.1 (GFS) vs. 266.4 (ECMWF).

Demand Implications (Natural Gas)

The competing forecasts present two distinct scenarios for Natural Gas demand.

• GFS Scenario: The GFS's hotter outlook, particularly its late-period heat, implies a significant extension of the cooling season. This would support unseasonably strong power burn demand and be bullish for NG prices relative to seasonal norms.
• ECMWF Scenario: The ECMWF suggests a more typical transition into autumn, with cooling demand steadily waning. This outlook is more neutral-to-bearish, aligning with expectations of declining power generation loads.

Heating demand is negligible in both models, with total weighted HDDs below 3.0 in either scenario. The market focus remains entirely on cooling load.

Model Agreement & Caveats

The overall model convergence score is 86.7%, which, while relatively high, masks the critical divergence in the long range. Confidence in the forecast beyond day 8 is low due to this split.

Notably, this GFS run shifted dramatically warmer versus its prior cycle, adding +58.7 weighted CDDs. In contrast, the ECMWF was very stable, shifting only -5.5 CDDs. Such a large single-run change in the GFS often indicates model instability and warrants skepticism. Until subsequent runs confirm this hot trend, the more conservative and consistent ECMWF solution should be weighted more heavily, especially for the days 10-15 period.

This analysis is based on a specific set of demand hubs from Open-Meteo and may not represent the full U.S. consumption picture. Model forecasts are subject to significant revision.