GFS vs ECMWF Demand-Hub Outlook: 2026-05-10 00Z

Executive Summary

The 00Z models show a notable divergence in the 15-day outlook, with the GFS forecasting significantly higher overall weather-driven demand (+15.9 weighted TDDs) than the ECMWF. This difference is driven by the GFS being both colder in the Midwest and materially hotter in the South Central region. Confidence in the forecast is low (convergence score of 0.44), compounded by a significant colder trend in the GFS versus its prior run, which adds to the uncertainty.

Hub-Level Weather

The national divergence is primarily driven by disagreements in two key, high-weight regions:

• Midwest (Chicago, 35% weight): The GFS is substantially colder, forecasting 104.1 cumulative HDDs compared to 82.6 HDDs from the ECMWF. The largest split occurs late in the forecast period (May 24), where the GFS sees a cold shot (13.0 HDDs) while the ECMWF forecasts warm conditions (7.8 CDDs).
• South Central (Houston, 15% weight): The GFS is consistently and significantly hotter than the ECMWF, resulting in a forecast of 248.5 cumulative CDDs versus just 167.7 CDDs from the ECMWF. This +80.8 CDD difference in Houston alone accounts for the bulk of the national cooling demand discrepancy.
• East (New York City, 30% weight): The models are in better agreement, with largely offsetting differences. The GFS forecasts slightly more cooling demand (45.5 CDDs vs. 40.1) while the ECMWF sees slightly more heating demand (59.4 HDDs vs. 51.3).

Demand Implications (Natural Gas)

The conflicting model outlooks present different implications for natural gas demand during this shoulder season month.

• GFS Outlook: The GFS solution is more supportive of natural gas prices. It forecasts a total of 62.8 weighted HDDs and 71.7 weighted CDDs. This implies a combination of lingering heating demand in the Midwest alongside robust, early-season power burn for cooling in the South.
• ECMWF Outlook: The ECMWF presents a more bearish demand picture, with only 56.1 weighted HDDs and 62.5 weighted CDDs. This suggests weaker demand on both the heating and cooling fronts compared to the GFS.
• Run-over-Run Shift: The GFS trended significantly colder versus its prior 12Z cycle, adding +10.1 weighted HDDs. In contrast, the ECMWF was very stable (+1.3 weighted HDDs). This large GFS shift reduces confidence in its latest solution.

Model Agreement & Caveats

Confidence in the 15-day weather outlook is currently low.

• Convergence: The model convergence score is 0.44, indicating poor agreement. The weighted_mean_abs_hdd_diff_by_day of 1.25 further highlights the day-to-day inconsistencies.
• Caveats: The low model agreement, particularly in the Day 8-15 period, suggests the forecast is unstable. The significant cycle-over-cycle change in the GFS is a primary concern and warrants treating its outlook with additional caution until subsequent runs show more consistency.
• Source Limitations: This analysis is based on a specific set of demand hubs and model versions (gfs_seamless, ecmwf_ifs025) from Open-Meteo and does not represent a full, population-weighted national forecast.