GFS vs ECMWF Demand-Hub Outlook: 2026-06-17 00Z

Executive Summary

The 15-day outlook is dominated by cooling demand, with the GFS model forecasting a significantly hotter and more demand-intensive period than the ECMWF. The GFS projects 165.6 demand-weighted CDDs, a notable 29 CDDs above the ECMWF's 136.6. Both models trended significantly hotter compared to their prior runs, adding support to the overall cooling demand picture, though confidence is tempered by a key divergence in the 8-15 day period.

Hub-Level Weather

The national-level disagreement is primarily driven by the GFS forecasting much stronger heat in key eastern and central demand hubs during the second half of the forecast period.

• Midwest (Chicago): This is the largest point of contention. The GFS is substantially hotter, projecting 98.8 cumulative CDDs versus just 65.1 for the ECMWF. The divergence begins around June 25th and becomes extreme by June 27-28, where GFS mean temperatures reach the low 80s while the ECMWF remains in the high 60s to low 70s.
• East (New York City): The GFS is also materially hotter, forecasting 176.2 CDDs compared to the ECMWF's 147.9. This difference is most pronounced from June 26th onward.
• South Central (Houston): Both models show strong, persistent heat, but the GFS is consistently hotter, accumulating 312.1 CDDs against the ECMWF's 267.1.

Demand Implications (Natural Gas)

The forecast points to a robust period for power burn driven by air conditioning load, with negligible heating demand.

• The GFS solution implies a significantly stronger call on natural gas for power generation compared to the ECMWF. The ~29 national weighted CDD difference between the models represents a material variance in potential daily power burn.
• The run-over-run trend is bullish for gas demand. The GFS added 9.9 weighted CDDs and the ECMWF added a substantial 14.9 weighted CDDs from the prior cycle, signaling a strengthening trend toward above-normal cooling demand.

Model Agreement & Caveats

Confidence in the forecast, particularly beyond day 7, should be moderate due to significant model disagreement.

• The overall convergence score is 0.79, indicating a moderate but not strong level of agreement between the models. This score has decreased slightly (-0.01) from the previous run, showing the models have moved further apart.
• The primary caveat is the large divergence in the 8-15 day period, especially in the heavily-weighted Chicago hub. Until the models come into better agreement on the late-period pattern, the ultimate demand impact remains uncertain.
• This analysis is based on a limited set of demand hubs and weather model grid points from Open-Meteo. It may not capture all regional nuances relevant to the U.S. natural gas market.