GFS vs ECMWF Demand-Hub Outlook: 2026-06-07T00:00Z

Executive Summary

The 15-day outlook shows a significant divergence between models, with the GFS forecasting substantially higher cooling demand than the ECMWF. The GFS projects 159.0 weighted CDDs versus the ECMWF's 132.4, a difference driven by warmer forecasts in the key Midwest and East Coast hubs. Despite GFS remaining the hotter model, it trended significantly cooler (-18.2 wCDD) cycle-over-cycle, while the ECMWF remained stable, adding complexity to the demand picture.

Hub-Level Weather

The national-level model disagreement is driven by large differences in the most heavily weighted demand centers:

• Midwest (Chicago, 35% weight): The models show a stark contrast. The GFS projects 110.1 CDDs, while the ECMWF is much cooler at 82.4 CDDs. The GFS shows a period of intense heat around June 10-11 and a warmer finish, whereas the ECMWF is cooler in the back half of the forecast, even generating 10.9 HDDs versus the GFS's 3.2.
• East (New York City, 30% weight): The GFS is notably hotter, accumulating 169.8 CDDs compared to the ECMWF's 140.4 CDDs. The largest divergence appears in the 10-15 day period.
• South Central (Houston, 15% weight): Both models are hot, but the GFS is exceptionally strong, calling for 302.3 CDDs against the ECMWF's 241.6 CDDs. The GFS maintains intense heat deep into the forecast while the ECMWF shows a cooling trend.

Demand Implications (Natural Gas)

The GFS solution implies robust power burn for cooling demand, suggesting a tighter supply/demand balance for natural gas over the next two weeks. The ECMWF outlook points to a more moderate, closer-to-normal demand scenario.

The significant cooler trend in the latest GFS run (-18.2 weighted CDDs) is a key bearish development compared to its prior forecast. In contrast, the ECMWF's stability (+0.3 wCDD run-over-run) lends more confidence to its more moderate outlook. The market may place less weight on the GFS's hotter absolute forecast given its volatility and bearish trend.

Model Agreement & Caveats

Model agreement is low, with a convergence score of 0.779. This low score reflects the material disagreements in key hubs beyond the first week and warrants caution. While convergence saw a minor improvement from the last cycle (+0.01), the overall signal remains uncertain.

This analysis is based on raw, unblended GFS and ECMWF model output from Open-Meteo for specific city hubs. These point forecasts may not perfectly represent demand across the entire weighted region. Degree days are calculated from the daily mean temperature against a 65°F base.