East U.S. Cold Days 11-15: GFS/ECM Vs. AI Forecasts

Fig. 1: The U.S. gas population weight HDD forecast utilizing all models, their consensus, and comparing with 48 hours ago/climatology.

Discussion: Consensus of overnight operational forecast models yields a very cold forecast for Nov. 29-Dec. 5 although not as cold as the market peak issued on Sunday (Fig. 1). The outlook for Dec. 6-12 is close to the 30-year normal.

The 11-15-day forecast is very important this week due to potential pattern change and brief closing/re-opening markets due to the Thanksgiving Holiday. The best estimate is combining GFS and ECM which produces lingering East U.S. chill while a major warm-up occurs in the West (Fig. 2). Conversely, the AI models offer two opposing extreme scenarios. AIFS maintains the eastern half of the U.S. chill with intensity (Fig. 3) while EXCARTA vigorously warms the entire continent (Fig. 4). GFS and ECM project a polar vortex over Northern Quebec in the 11-15-day period enhanced by snow cover expansion from southern portions of Ontario to Quebec and into the Northeast U.S. The AIFS is too strong with the upper trough on the U.S. East Coast due to lack of farther south snow cover. EXCARTA projects return of a Gulf of Alaska (GOA) trough and to compensate, the warming ridge pattern over the Western States forecast by most models, shifts farther east from the Southwest U.S. to Quebec to produce a warmer solution. The primary climate dynamics driving the weather pattern Dec. 6-11 are a moderate-strength phase_4/phase_5 of the Madden Julian oscillation (MJO), weakening and then returning stratospheric warming across eastern North America, and presence of snow cover. MJO P4/P5 is a warming force on the U.S. in December supporting the warmer forecasts. However, stratospheric warming across eastern North America, present during late NOV/early DEC fueling the 6-10-day East U.S. cold, eases briefly before returning later in the 11-15-day period. The likely scenario is a pulse of warmth into the East early in the 11-15-day period followed by reloading cold on the East Coast in 13-15 days. Expanding snow cover across Eastern Canada into the Northeast U.S. helps the colder solution. In summary, the dynamic reasoning best supports the GFS +ECM projection while unexplained machine learning solutions are not market-worthy forecasts this morning.

Fig. 2-4: The GFS + ECM 11-15-day temperature anomaly forecast compared to AIFS and EXCARTA.