GFS vs ECMWF Demand-Hub Outlook: 2026-06-19T00:00Z

Executive Summary

The 15-day outlook shows a significant cooling demand forecast, with the GFS model notably hotter than the ECMWF, primarily driven by warmth in the South and Midwest. The latest model cycle shows a bearish divergence, with the GFS adding +9.9 weighted CDDs while the ECMWF lost -4.0 CDDs, lowering overall forecast confidence. The key drivers are the high-demand Midwest and East hubs, where models disagree on the timing and intensity of late-period heat.

Hub-Level Weather

The national demand picture is primarily shaped by the two largest hubs, which show considerable model disagreement:

• Midwest (Chicago, 35% weight): Models diverge significantly. The GFS forecasts a stronger cooling load (79.0 CDDs) compared to the ECMWF (64.2 CDDs). The ECMWF is also colder overall, projecting 9.1 HDDs versus the GFS's 5.3 HDDs. The main split occurs in the final days (July 2-3), where the GFS shows intense heat (mean temps 80-84°F) while the ECMWF remains much cooler (70-72°F).
• East (New York City, 30% weight): The models are in closer agreement here, though the GFS remains slightly hotter with 178.8 cumulative CDDs versus the ECMWF's 171.3.
• South Central (Houston, 15% weight): The GFS is materially hotter throughout the period, forecasting 319.0 CDDs against the ECMWF's 268.0 CDDs, a difference of 51 CDDs.
• Northeast (Boston, 10% weight): In a notable reversal of the general trend, the ECMWF is significantly warmer here, projecting 120.4 CDDs compared to just 54.8 CDDs from the GFS.

Demand Implications (Natural Gas)

The forecast implies a period of above-normal natural gas demand for power generation (power burn), driven by air conditioning load.

• National Demand: The GFS projects a stronger call on gas, with 159.5 demand-weighted CDDs over the 15-day period. This is approximately 14.6 CDDs higher than the ECMWF's forecast of 144.9 CDDs.
• Run-over-Run Trend: The models have diverged since the prior cycle. The GFS trended significantly hotter, adding +9.9 weighted CDDs. In contrast, the ECMWF trended cooler, removing -4.0 weighted CDDs. This conflicting trend reduces confidence in the overall forecast magnitude.

Model Agreement & Caveats

Forecast confidence is moderate and has decreased since the last model run.

• Convergence Score: The current score is 0.79, indicating moderate agreement but with notable regional and temporal differences. This score has decreased by -0.08 since the previous cycle, highlighting the growing model disagreement.
• Divergence: The primary disagreement is in the Day 8-15 period, where the GFS model shows significantly more heat building across the central and southern U.S. than the ECMWF. The opposite is true for the Northeast.
• Source Limitations: This analysis is based on a limited set of demand hubs from the free Open-Meteo API. It may not perfectly represent a comprehensive, population-weighted national total. The degree day values are calculated from mean daily temperatures and do not capture intraday temperature extremes that can also drive demand.