GFS vs ECMWF Demand-Hub Outlook: 2026-06-14 00Z

Executive Summary

Both GFS and ECMWF models project a strong, cooling-dominated demand profile for the next 15 days, with the GFS showing a slightly higher national load (138.8 weighted CDDs vs. 128.7 for ECMWF). The primary source of model disagreement is a significant GFS-led cool shot into the Midwest mid-forecast, which generates unseasonal heating demand not seen in the ECMWF. Confidence is moderate (Convergence Score: 0.67), as both models trended significantly hotter run-over-run, particularly the ECMWF (+28.0 weighted CDDs).

Hub-Level Weather

The forecast divergence is almost entirely driven by the high-weight Midwest (Chicago) hub: * GFS Cold, then Hot: The GFS is notably colder from June 17-22, accumulating 21.4 HDDs where the ECMWF has almost none. It then flips much hotter than the ECMWF in the final four days of the forecast. * ECMWF Steadily Warmer: The ECMWF shows a more consistent, mild-to-warm pattern for Chicago, totaling only 4.4 HDDs for the entire 15-day period.

Elsewhere, the GFS is generally the hotter model: * South Central (Houston): GFS projects 287.0 CDDs, a significant 31.6 CDDs more than the ECMWF's 255.4. * East (New York City): GFS is moderately warmer, projecting 150.2 CDDs versus 142.0 for the ECMWF, with most of the separation occurring late in the forecast.

Demand Implications (Natural Gas)

The outlook implies robust and above-normal natural gas demand for power generation (power burn) to meet air conditioning load. * National Cooling Demand: The GFS projects a stronger cooling demand signal, primarily from the South Central and East hubs. The total demand-weighted outlook is for 138.8 CDDs (GFS) vs. 128.7 CDDs (ECMWF). * Midwest Heating Demand Anomaly: A key risk to the bullish demand case comes from the GFS, which forecasts 7.5 weighted HDDs nationally, nearly 5x the ECMWF's 1.6 weighted HDDs. This GFS-led heating demand, centered in the Midwest, would temper national demand mid-forecast if it verifies. * Run-over-Run Trend: Both models made significant bullish revisions. The GFS added +20.0 weighted CDDs and the ECMWF added +28.0 weighted CDDs compared to their prior runs, signaling a strengthening conviction for widespread heat.

Model Agreement & Caveats

• Confidence: Forecast confidence is moderate. The Convergence Score of 0.67 (out of 1.0) reflects the substantial disagreement over the temperature evolution in the Midwest.
• Divergence: The convergence score decreased by 0.06 from the previous cycle, indicating the models are moving further apart. This lower agreement, combined with large run-over-run changes, warrants caution, especially for the Day 8-15 period.
• Source: This analysis uses Open-Meteo's implementation of the GFS Seamless and ECMWF IFS 0.25° models. The interpretation is specific to this data source.