GFS vs ECMWF Demand-Hub Outlook: 2026-05-28T00:00Z

Executive Summary

Models show significant disagreement over the 15-day outlook, with the GFS forecasting a much hotter and higher-demand scenario than the ECMWF. The GFS projects a total of 118.7 weighted Cooling Degree Days (CDDs), a notable +35 CDDs above the ECMWF's 83.6. This divergence is driven by a massive +20 CDD bullish shift in the GFS from its prior run, suggesting low confidence in the forecast until models converge.

Hub-Level Weather

The primary drivers of the national forecast, the Midwest and East, show the largest model disagreements.

• East (New York City, 30% weight): The GFS is materially hotter, projecting 103.0 CDDs versus just 44.9 CDDs from the ECMWF. The GFS sees a significant heat event building after June 7th, while the ECMWF maintains a much more moderate temperature profile.
• Midwest (Chicago, 35% weight): Models disagree on the timing of temperature swings. The ECMWF is warmer in the first week, generating early-period CDDs. The GFS is cooler initially (15.2 HDDs vs. 2.7 HDDs) but becomes extremely hot late in the forecast, particularly from June 8-11.
• South (Houston & Atlanta, 25% combined weight): Both hubs show a consistently hotter GFS solution. The GFS projects 47.5 more CDDs in Houston and 70.3 more CDDs in Atlanta over the 15-day period.

Demand Implications (Natural Gas)

The conflicting outlooks present two distinct scenarios for natural gas power burn over the next two weeks.

• GFS Scenario: The GFS outlook implies a strong, above-normal start to the summer cooling season. The projected heat, especially across the high-demand Eastern and Southern hubs, would drive robust power generation demand and be bullish for natural gas prices. The +20.5 weighted CDD addition versus the prior cycle is a significant tightening signal.
• ECMWF Scenario: The ECMWF suggests a more moderate, near-normal demand profile. Its cooler solution for the East and Midwest would temper national power burn, while its forecast for some lingering heating demand in the Northeast (Boston: 72.2 HDDs) highlights a more transitional, shoulder-season pattern.

Model Agreement & Caveats

Confidence in this forecast is low due to poor model agreement and significant run-over-run volatility in the GFS.

• Convergence: The model agreement score is low at 0.55 (where 1.0 is perfect agreement), reflecting the major divergence in temperature outlooks, particularly in the 8-15 day period.
• Run-over-Run Shift: The GFS added +20.5 weighted CDDs and shed -7.1 weighted HDDs from its previous cycle. The ECMWF, in contrast, was almost completely stable. This type of large, unilateral shift by one model warrants extreme caution. The market should wait for subsequent runs to see if the ECMWF trends toward the GFS's hotter solution or if the GFS reverts.
• Source Limitations: This analysis is based on a specific set of demand-weighted hubs provided by Open-Meteo and may not perfectly align with proprietary, comprehensive national demand models. Degree days are calculated from a daily mean temperature.