GFS vs ECMWF Demand-Hub Outlook: 14 Sep 2026 00z

Executive Summary

The 15-day outlook points to continued, above-normal cooling demand, but with significant disagreement between models on the timing and location of a late-period cooldown. Nationally, the GFS is the warmer of the two models (+10.7 weighted CDDs), primarily driven by stronger heat in the South. Confidence in the forecast is low, as the models show starkly different outcomes for the high-weight Midwest and East hubs in the 8-15 day range.

Hub-Level Weather

The national forecast is a composite of conflicting regional outlooks, reducing overall clarity:

• Midwest (Chicago, 35% weight): This is a key area of disagreement. The GFS is significantly colder in the long-range, forecasting 25.6 HDDs versus only 8.0 HDDs from the ECMWF. Conversely, the ECMWF is warmer in the mid-range, projecting 76.7 CDDs to the GFS's 62.6.
• East (New York City, 30% weight): The model solutions are inverted here compared to the Midwest. The ECMWF is substantially colder, calling for 24.2 HDDs versus just 7.7 HDDs from the GFS. Consequently, the GFS shows much stronger cooling demand with 80.7 CDDs compared to the ECMWF's 41.0.
• South Central (Houston, 15% weight): Both models show persistent cooling demand. The GFS is consistently hotter, forecasting 272.8 CDDs over the 15-day period, a notable +24.3 CDDs above the ECMWF.

Demand Implications (Natural Gas)

The forecast remains supportive for Natural Gas consumption via power burn, driven by cooling demand that is strong for mid-September.

• Cooling Demand: The GFS outlook is more bullish for near-term NG contracts, with its forecast for 108.7 weighted CDDs exceeding the ECMWF's 98.0. The GFS also trended significantly warmer in this cycle, adding +5.4 weighted CDDs compared to its previous run.
• Heating Demand: While heating load is minimal, the extreme model divergence in the Midwest and East creates significant uncertainty for demand as the forecast period progresses. The market is unlikely to price in either a cold GFS Midwest or a cold ECMWF East until there is better model agreement.
• Run-over-Run Shift: The GFS model made a large shift warmer compared to its prior run, shedding 9.8 weighted HDDs. The ECMWF was comparatively stable. This GFS volatility reduces confidence in its long-range cold solution for the Midwest.

Model Agreement & Caveats

Forecast confidence is low due to poor model agreement on key regional patterns.

• Convergence: The national convergence score is a low 0.447, reflecting the major disagreements in the high-demand Eastern and Midwestern hubs.
• Divergence Drivers: The primary source of uncertainty is the placement and intensity of a trough expected in the 8-15 day period. The GFS brings it into the Midwest, while the ECMWF brings it into the East, leading to opposite thermal outcomes.
• Source Limitations: This analysis is based on a select group of demand-hubs provided by Open-Meteo and does not represent a full CONUS population-weighted forecast. Weather patterns between these hubs can influence the national total.