GFS vs ECMWF Demand-Hub Outlook: 2026-06-05T00:00Z

Executive Summary

The 15-day outlook shows a material split between models, with the GFS forecasting stronger national cooling demand than the ECMWF (+5.4 weighted CDDs). This divergence is primarily driven by a significantly hotter GFS outlook for the South Central (Houston). Confidence in the forecast is moderate, as the models have trended in opposite directions since the prior cycle, with the GFS adding significant demand (+17.6 CDDs) while the ECMWF shed demand (-14.3 CDDs).

Hub-Level Weather

The national picture is shaped by significant regional disagreements between the GFS and ECMWF models:

• South Central (Houston, 15% weight): This is the largest point of disagreement. The GFS is substantially hotter, forecasting 304.9 CDDs compared to just 227.1 CDDs from the ECMWF over the 15-day period.
• Midwest (Chicago, 35% weight): In the most heavily-weighted hub, the ECMWF is warmer, projecting 137.4 CDDs versus the GFS's 111.5 CDDs.
• East (New York City, 30% weight): The GFS is notably warmer in this key hub, with a cumulative forecast of 172.9 CDDs against the ECMWF's 155.8 CDDs.

The GFS's intense heat in the South Central and its warmer forecast for NYC are the primary drivers of its higher national demand outlook, outweighing the ECMWF's warmer solution for the Midwest.

Demand Implications (Natural Gas)

The early June period is dominated by cooling demand, which drives natural gas consumption for power generation (power burn).

• The GFS solution implies a stronger call on natural gas for power generation, particularly in the South and East.
• The ECMWF outlook, while still showing consistent cooling demand, is less bullish for gas demand due to its more moderate temperatures in Houston and NYC.
• Both models show minimal heating demand, with only minor HDD totals in the Northeast (Boston).

Model Agreement & Caveats

Confidence in the 15-day outlook is moderate and warrants caution due to model divergence and recent trends.

• Convergence: The overall convergence score is 0.78 (out of 1.0), indicating some alignment but highlighting the material regional differences noted above.
• Run-over-Run Shift: Critically, the models are trending in opposite directions. The GFS added +17.6 weighted CDDs versus its prior run, while the ECMWF removed -14.3 weighted CDDs. This opposing trend is a significant red flag for forecast instability, particularly in the 8-15 day period.
• Source: This analysis is based on GFS and ECMWF data provided by Open-Meteo. Model performance can vary, and this two-model comparison does not include ensemble or other operational guidance.