GFS vs ECMWF Demand-Hub Outlook: 14-Jul 00Z

Executive Summary

The 15-day outlook points to strong cooling demand, with the ECMWF model forecasting a slightly higher national load (215.5 weighted CDDs) compared to the GFS (208.1 wCDDs). While high-level agreement is strong, this masks significant regional disagreements in the Midwest and Southeast. Both models have trended materially cooler compared to their prior runs, reducing the total forecast demand.

Hub-Level Weather

The national demand picture is driven by offsetting regional forecasts between the two models, primarily in the Midwest and South.

• Midwest (Chicago, 35% weight): This is the area of greatest disagreement. ECMWF is substantially hotter, forecasting 207.1 cumulative CDDs versus just 163.2 from the GFS. The divergence grows significantly in the back half of the forecast (post July 24th), where GFS calls for a sharp cool-down while ECMWF maintains summer warmth.
• East (New York City, 30% weight): Models are in close agreement, with ECMWF forecasting 221.9 CDDs and GFS at 214.4 CDDs. Both show a period of intense heat through July 17th before moderating.
• South (Atlanta, 10% weight): In a direct contrast to the Midwest, the GFS is significantly hotter in the Southeast, projecting 280.5 CDDs against the ECMWF's much cooler 194.8 CDDs.
• South Central (Houston, 15% weight): The GFS is moderately warmer, with 298.2 CDDs versus 280.7 CDDs from the ECMWF.

Demand Implications (Natural Gas)

The forecast implies robust natural gas demand for power generation (power burn) throughout the 15-day period, consistent with mid-summer heat.

• The first 7 days show strong model agreement on above-normal temperatures across the key eastern and midwestern demand centers, supporting a firm demand outlook.
• The primary uncertainty for gas demand arises in week two. The ECMWF's hotter Midwest solution would sustain higher power burn in that key region. Conversely, the GFS's hotter solution in the South and South Central would concentrate cooling demand there.
• While the net national demand difference between models is modest (~7.4 weighted CDDs), the significant regional splits have strong implications for regional power pricing and pipeline basis differentials.

Model Agreement & Caveats

Overall model confidence is high, but caution is warranted due to the significant regional splits and recent model trends.

• Convergence: The high-level convergence score is strong at 0.97, indicating good agreement on the national pattern. However, as noted, this high score obscures major regional differences that are critical for assessing risk.
• Run-over-Run Deltas: Both models trended notably cooler compared to the prior cycle. The GFS lost -14.3 weighted CDDs and the ECMWF lost -10.9 wCDDs, indicating a less extreme heat forecast than 6 hours prior.
• Source Limitations: This analysis is based on GFS Seamless and ECMWF IFS 0.25° model data from Open-Meteo. The demand hub weightings are illustrative for a national aggregate view.