GFS vs ECMWF Demand-Hub Outlook: 2026-04-20T00:00Z

Executive Summary

The 15-day outlook shows a material disagreement between models, with the ECMWF forecasting significantly higher heating demand and the GFS projecting stronger early-season cooling demand. This divergence grows substantially in the 8-15 day period, driven by conflicting temperature forecasts in the Midwest and South. Run-over-run trends show the GFS shifting significantly warmer, increasing the model spread and lowering overall forecast confidence.

Hub-Level Weather

The primary drivers of the national forecast are the Midwest (Chicago, 35% weight) and the East (New York City, 30% weight).

• Midwest (Chicago): The ECMWF is notably colder, forecasting 188.1 cumulative HDDs versus 172.0 HDDs from the GFS. The primary divergence occurs after April 28th; for example, on April 30th the GFS forecasts only 2.8 HDDs while the ECMWF sees a much colder 16.6 HDDs.
• East (New York City): A similar colder trend exists in the ECMWF, which forecasts 220.5 HDDs compared to the GFS's 200.3 HDDs. The largest disagreements appear in the latter half of the forecast.
• South Central (Houston): Models show a major split on cooling demand. The GFS is far warmer, projecting 168.0 cumulative CDDs, more than double the ECMWF's forecast of 70.7 CDDs. This is caused by the ECMWF bringing a sharp cool-down around April 29th that is absent in the GFS.
• South (Atlanta): The GFS is also significantly warmer here, forecasting 63.5 CDDs to the ECMWF's 33.9. The ECMWF also sees lingering heating demand (48.4 HDDs) that the GFS largely dismisses (7.4 HDDs).

Demand Implications (Natural Gas)

The conflicting model solutions present two distinct natural gas demand scenarios for late April and early May.

• National Heating Demand: The ECMWF presents a more supportive (higher demand) picture with 167.1 weighted cumulative HDDs. The GFS is more bearish (lower demand) at 149.6 weighted HDDs. The GFS trended warmer versus its prior run (-3.8 delta HDDs), while the ECMWF was stable (+0.1 delta HDDs).
• National Cooling Demand: The GFS suggests a much stronger start to cooling season, projecting 31.6 weighted cumulative CDDs, primarily from the South. This is double the ECMWF's forecast of 15.4 weighted CDDs. The GFS also trended significantly higher on cooling demand run-over-run (+4.1 delta CDDs).

Overall, the GFS suggests a faster transition away from heating demand toward shoulder-season lows, whereas the ECMWF maintains more robust late-season heating load.

Model Agreement & Caveats

Confidence in this 15-day forecast is low-to-moderate due to significant model disagreement, particularly beyond the first week.

• Convergence: The national convergence score of 0.2621 reflects the notable spread between models. Confidence is further reduced by the run-over-run trend, as model agreement worsened from the prior cycle (+0.0404 delta on the convergence score).
• Caveats: This analysis is based on deterministic GFS and ECMWF runs for a limited set of demand-weighted hubs. It does not incorporate ensemble guidance, which would provide a better sense of probabilistic outcomes and forecast uncertainty. The underlying data is provided by Open-Meteo.