GFS vs ECMWF Demand-Hub Outlook: 2026-06-02 00Z

Executive Summary

The 15-day outlook shows significant model disagreement, resulting in a low confidence forecast. While the national demand-weighted cooling degree day (CDD) totals are superficially similar (GFS: 126.1, ECMWF: 131.5), this masks major, offsetting regional differences. The primary conflict is in the high-demand Midwest, where the ECMWF is substantially hotter, and the East, where the GFS projects significantly more heat, particularly late in the period.

Hub-Level Weather

The national forecast is being driven by sharp disagreements in the two largest demand hubs:

• Midwest (Chicago, 35% weight): This is the key battleground. The ECMWF is significantly more bullish, forecasting 155.6 CDDs versus just 84.4 CDDs from the GFS. The divergence is most acute in the 8-15 day period, where the ECMWF maintains summer-like warmth while the GFS brings in a sharp cool-down (e.g., a 14°F temperature difference on June 9th).
• East (New York City, 30% weight): The models are inverted here, with the GFS being much hotter. It projects 124.5 CDDs compared to the ECMWF's 84.2. The GFS shows a late-period heat event (June 13-14) that the ECMWF completely misses, instead forecasting mild temperatures in the mid-60s.
• Northeast (Boston, 10% weight): The GFS is notably colder, forecasting 52.2 heating degree days (HDDs) versus only 16.4 HDDs from the ECMWF. This GFS forecast for late-season heating demand is a significant outlier.
• South Central (Houston, 15% weight): Both models show consistent cooling demand, though the GFS is warmer, accumulating 259.5 CDDs to the ECMWF's 229.0.

Demand Implications (Natural Gas)

The conflicting model guidance presents two very different scenarios for natural gas power burn over the next two weeks.

• The ECMWF solution, with strong and sustained heat in the Midwest, implies robust cooling demand and higher-than-normal power burn for this time of year.
• The GFS solution is considerably more bearish. Its cooler forecast for the Midwest, combined with notable heating demand in the Northeast, points to significantly weaker power sector gas demand.

Given the offsetting nature of the regional forecasts, the net national demand signal is weak and unreliable.

Model Agreement & Caveats

Confidence in this forecast is low. The overall model convergence score is 0.62, indicating significant disagreement between the GFS and ECMWF.

This uncertainty is compounded by the latest run-over-run model trends. Compared to the previous cycle, the ECMWF trended warmer (+6.6 weighted CDDs), while the GFS trended cooler (-6.4 weighted CDDs). This growing divergence between models signals instability in the medium-range pattern, warranting caution before committing to either temperature scenario.

This analysis is based on gridded model data from Open-Meteo for specific hub locations and may not perfectly capture all regional demand nuances.