GFS vs ECMWF Demand-Hub Outlook: 2026-06-12 00Z

Executive Summary

The 15-day outlook shows national cooling demand forecasts from GFS and ECMWF are in close alignment, with weighted cumulative CDDs of 127.2 and 125.3, respectively. However, this national agreement masks significant regional divergences, with the GFS forecasting much stronger heat in the South Central region (Houston) while the ECMWF is materially hotter in the East (NYC, Boston). Both models have trended significantly cooler versus their prior runs, a bearish signal for near-term gas demand.

Hub-Level Weather

The primary drivers of the national demand forecast are the Midwest, East, and South Central hubs, which exhibit the most significant model disagreement:

• Midwest (Chicago, 35% weight): Models diverge notably mid-forecast. The ECMWF is significantly cooler, forecasting 7.6 cumulative HDDs and 54.9 CDDs, compared to the GFS's 0.3 HDDs and 59.4 CDDs.
• East (New York City, 30% weight): The ECMWF is substantially hotter in this key demand center, projecting 132.2 cumulative CDDs versus the GFS's 109.7 CDDs. This pattern is mirrored in Boston.
• South Central (Houston, 15% weight): The GFS is far more bullish on heat in Texas, forecasting 315.2 cumulative CDDs. This is a substantial 58 CDDs higher than the ECMWF's forecast of 257.2 CDDs.

Demand Implications (Natural Gas)

The close aggregate national CDD forecast suggests a similar level of national power burn for natural gas between the two models. However, the regional differences imply significant uncertainty for basis pricing and regional grid load. The GFS scenario would stress ERCOT more heavily, while the ECMWF points to higher demand in PJM and ISO-NE.

The most significant signal from this cycle is the run-over-run trend. Both models shed a meaningful amount of demand compared to 6 hours prior: * GFS: Lost 7.7 weighted CDDs (delta_weighted_cumulative_cdd_gfs: -7.6968) * ECMWF: Lost 4.0 weighted CDDs (delta_weighted_cumulative_cdd_ecmwf: -3.976)

This coordinated bearish trend increases confidence that near-term demand expectations should be revised lower.

Model Agreement & Caveats

The overall model convergence score is moderate-to-good at 0.806, which is a slight improvement (+0.0358) from the prior cycle. While this national score indicates reasonable top-level agreement, traders should be cautious given the stark regional disagreements outlined above. The forecast confidence is highest in the bearish run-over-run trend and lowest in the regional distribution of cooling demand.

This analysis is based on deterministic GFS and ECMWF runs from Open-Meteo. It does not account for the range of possible outcomes captured by ensemble forecasting systems, which should be consulted for a more complete view of forecast uncertainty.