GFS vs ECMWF Demand-Hub Outlook: 2026-09-01T00:00Z

Executive Summary

The 15-day outlook shows a significant divergence between models, with the ECMWF forecasting a materially larger cooling load than the GFS. The ECMWF projects 189.9 demand-weighted CDDs compared to the GFS's 165.8. This difference is driven by a much cooler GFS solution in the high-demand Midwest and East hubs, a view that represents a sharp colder trend from its prior run (-28.6 CDDs).

Hub-Level Weather

The primary model disagreement is centered on the Midwest and East, with the South showing an opposing pattern.

• Midwest (Chicago, 35% weight): The models diverge sharply in the Day 8-12 period. The GFS turns significantly cooler, resulting in a 15-day cumulative 130.8 CDDs. The ECMWF remains much warmer through this period, accumulating 215.4 CDDs, a difference of over 84 CDDs in this key region.
• East (New York City, 30% weight): A similar story unfolds in the East. The GFS brings in a cool shot around Day 9, registering 6.15 HDDs and finishing with 116.5 CDDs. The ECMWF is consistently warmer, forecasting 147.5 CDDs and zero HDDs.
• South Central (Houston, 15% weight): In contrast to the northern hubs, the GFS is substantially hotter in the South, projecting 339.6 CDDs versus the ECMWF's 246.4 CDDs.

Demand Implications (Natural Gas)

The conflicting model views present a mixed signal for Natural Gas demand, though the national picture leans warmer on the ECMWF.

• ECMWF's Warmer National View: The ECMWF's higher demand-weighted CDD total (189.9) implies stronger power burn for air conditioning and is more supportive for NG futures than the GFS outlook. The ECMWF also trended hotter versus its prior run, adding 14.8 weighted CDDs.
• GFS's Cooler Northern Trend: The GFS's sharp cooler trend (-28.6 weighted CDDs run-over-run) is a bearish development. Its forecast for 5.1 weighted HDDs, concentrated in the Northeast, hints at the end of strong cooling demand in key population centers.

Overall, the ECMWF presents a more bullish demand scenario, while the GFS has aggressively shifted toward a cooler, less supportive pattern.

Model Agreement & Caveats

Confidence in this forecast is moderate and has decreased since the last cycle.

• Convergence Score: The current model convergence score is 0.75, indicating notable disagreement. This has deteriorated from the previous cycle, falling by 0.14. The models are actively diverging, reducing forecast reliability.
• Run-over-Run Shift: The primary reason for the lower confidence is the GFS's significant cooling trend against a modest warming trend in the ECMWF. This widening gap suggests instability in the medium-range pattern.
• Source: This analysis is based on GFS and ECMWF model data from Open-Meteo, using a 65°F base for all degree-day calculations. Forecast accuracy typically decreases beyond 7-10 days.