GFS vs ECMWF Demand-Hub Outlook: 2026-08-19T00:00Z

Executive Summary

The 15-day outlook shows a significant divergence in cooling demand, with the GFS model forecasting a substantially warmer and more demand-heavy period than the ECMWF. This split is primarily driven by the GFS projecting much hotter temperatures in the high-demand Midwest. Both models have made large, opposing shifts from their prior runs, with the GFS trending hotter and the ECMWF cooler, indicating low overall forecast confidence despite a high HDD-based convergence score.

Hub-Level Weather

The GFS is materially warmer than the ECMWF on a national demand-weighted basis, projecting 168.8 total CDDs versus the ECMWF's 135.3. The key drivers of this divergence are:

• Midwest (Chicago): The largest source of disagreement occurs in this highest-weighted hub (0.35). The GFS forecasts 119.7 CDDs, nearly double the ECMWF's forecast of 66.0 CDDs.
• South Central (Houston): The GFS is also significantly hotter in the South, calling for 376.2 CDDs compared to 305.3 from the ECMWF.
• East (New York City): The models are in closer agreement in this second-highest weighted hub (0.30), with the GFS at 130.5 CDDs and the ECMWF at 116.4 CDDs.
• Northeast (Boston): In a notable exception, the ECMWF is significantly warmer in Boston, projecting 101.1 CDDs against the GFS's 49.8 CDDs. However, this hub's lower weight (0.1) mutes its national impact.

Demand Implications (Natural Gas)

The GFS model's outlook suggests a significantly more bullish scenario for Natural Gas power burn over the next two weeks. The additional 33.5 weighted CDDs in the GFS relative to the ECMWF imply sustained, above-average cooling demand that would support prices for futures like /tickers/NG.

The run-over-run model changes highlight the forecast volatility: * GFS Delta: Trended sharply hotter, adding +25.7 weighted CDDs from its previous cycle. * ECMWF Delta: Trended sharply cooler, removing -24.3 weighted CDDs from its previous cycle.

This dramatic divergence between models from one run to the next suggests the market should treat either model's specific outcome with caution until a clear trend is established.

Model Agreement & Caveats

The formal convergence_score is high at 0.96, which suggests strong model agreement. However, this metric is primarily driven by agreement on Heating Degree Days (HDDs), where both models correctly forecast minimal heating load for late August.

The true forecast confidence is low. The key takeaways are: 1. There is significant disagreement on CDDs, the primary demand driver for this season. 2. The large, opposing run-over-run deltas indicate model instability and a low-confidence forecast environment.

This analysis is based on a 65°F degree day base from GFS and ECMWF model data provided by Open-Meteo. Hub weights are proprietary and represent their estimated contribution to national weather-driven gas demand.