GFS vs ECMWF Demand-Hub Outlook: 2026-05-13T00:00Z

Executive Summary

The 15-day outlook presents a conflicting demand picture, with the ECMWF forecasting significantly more heating demand and the GFS projecting stronger cooling demand. The ECMWF is colder overall, calling for 48.7 weighted HDDs versus the GFS's 38.3. This difference is driven by a late-period cool shot in the Midwest, a region where model disagreement is highest. Conversely, the GFS is warmer in the South, forecasting 70.4 weighted CDDs against the ECMWF's 64.6.

Hub-Level Weather

The primary drivers of the national forecast and model disagreement are:

• Midwest (Chicago, 35% weight): This is the key point of divergence. The ECMWF is materially colder, forecasting 82.2 cumulative HDDs compared to 68.8 HDDs from the GFS. The disagreement is most pronounced in the May 20-24 period, where the ECMWF shows a significant late-season heating event that the GFS largely misses.
• South Central (Houston, 15% weight) & South (Atlanta, 10% weight): The GFS is consistently warmer across the southern hubs, which drives its higher national cooling demand forecast. The GFS projects 225.4 CDDs for Houston and 146.0 CDDs for Atlanta, compared to the ECMWF's forecast of 198.9 and 118.5, respectively.
• East (New York City, 30% weight): The ECMWF projects more net heating demand (36.6 HDDs vs. GFS's 20.0), but also shows a stronger, earlier spike in cooling demand around May 19-20.

Demand Implications (Natural Gas)

The conflicting model signals suggest different risks for natural gas demand during this shoulder season month.

• ECMWF Outlook: The higher HDD forecast implies a stronger-than-expected call on natural gas for residential and commercial heating, particularly if the Midwest cold shot materializes in the back half of the forecast. This represents a more supportive (bullish) scenario for gas prices.
• GFS Outlook: The higher CDD forecast points to an earlier-than-normal start to significant cooling demand, which would increase natural gas consumption for power generation, primarily in the South.
• Run-over-Run Trends: Both models added a small amount of heating demand versus their prior runs (delta_weighted_cumulative_hdd_gfs: +1.7, delta_weighted_cumulative_hdd_ecmwf: +1.7). Critically, the GFS significantly reduced its cooling demand forecast (delta_weighted_cumulative_cdd_gfs: -16.3), a major cooler shift that brings it closer to the ECMWF's view.

Model Agreement & Caveats

Confidence in this forecast is moderate, with notable regional caveats.

• Convergence: The national convergence_score of 0.29 indicates a moderate level of disagreement. While this score has improved since the last cycle (delta_convergence_score: -0.11), significant regional differences remain, especially in the Midwest where the daily HDD difference averages over 5.0.
• Caveats: The primary uncertainty lies in the Day 8-15 period. The conflicting solutions for the Midwest—a key heating demand center—highlight the low confidence in temperature patterns that far out. The market should treat the potential for late-season heating demand as a risk factor rather than a certainty.
• Source: This analysis is based on GFS Seamless and ECMWF IFS 0.25° data from Open-Meteo. Interpretations may vary from other data providers.