GFS vs ECMWF Demand-Hub Outlook: 2026-09-02T00:00Z

Executive Summary

The 15-day outlook shows a net bullish bias for cooling demand, with the GFS model projecting a higher national load (+191.4 CDDs) than the ECMWF (+178.9 CDDs). This difference is driven by a significant GFS heat forecast for the South, which is offset by the ECMWF showing stronger cooling demand in the high-weight Midwest hub. A notable bearish shift in the latest GFS run (-17.7 CDDs) reduces confidence in its hotter solution and suggests the overall forecast may be softening.

Hub-Level Weather

The primary model disagreement is regional. The GFS is substantially hotter in the southern demand centers, while the ECMWF is hotter across the Midwest and parts of the Northeast.

• Midwest (Chicago): The largest demand hub (weight: 0.35) shows a key disagreement. The ECMWF is notably hotter, projecting 205.3 cumulative CDDs versus the GFS's 183.2.
• South Central (Houston): The GFS projects significant heat in this region (weight: 0.15), forecasting 319.6 CDDs compared to just 237.8 CDDs from the ECMWF—a difference of over 81 CDDs.
• South (Atlanta): A similar pattern exists in the South (weight: 0.10), with the GFS forecasting 321.6 CDDs versus the ECMWF's 243.2.
• East (New York City): In the second-largest hub (weight: 0.30), models are in closer agreement, with the GFS slightly warmer at 143.7 CDDs vs the ECMWF's 132.8.

Demand Implications (Natural Gas)

The forecast for early to mid-September indicates persistent cooling demand, which will support power burn for Natural Gas. The GFS model implies stronger-than-normal demand, particularly from the South, which would be bullish for the NG contract. However, the ECMWF's cooler southern forecast presents a more moderate demand picture.

The most significant development is the run-over-run trend. The GFS model shed a substantial 17.7 weighted CDDs from its prior cycle, while the ECMWF remained stable (-1.0 CDD). This bearish GFS shift suggests a potential correction towards the ECMWF's less extreme heat, reducing the upside risk for gas-fired power generation.

Model Agreement & Caveats

The overall model convergence score is high at 0.9563, suggesting good agreement on the broad pattern of above-normal temperatures. However, this high-level score masks the significant regional divergences noted above.

• Primary Caveat: The GFS model's large bearish shift (-17.7 CDDs) vs the prior run indicates instability in its forecast. This reduces conviction in its outlier heat for the South and suggests the consensus forecast is trending cooler.
• Confidence: Confidence in the near-term (1-5 days) is high, but the significant model divergence in the 6-15 day period, particularly in the South, warrants caution. The stable ECMWF run currently appears to be the more reliable signal.
• Source: This analysis is based on raw GFS and ECMWF model output from Open-Meteo and does not incorporate proprietary blending or analyst adjustments.