Forecasting extremes is not easy to do!
Fig. 1: Most recent 7-day U.S. temperature anomaly verification from NOAA.
Discussion: The historic arctic outbreak is starting to lose its punch but remains quite intense over the Great Plains. The most recent 7-day temperature anomaly observation analysis identifies the extreme cold peaking at 30F to 45F below normal from northern Nebraska to Montana (Fig. 1).
As a review, the arctic air is the result of a perfect set of circumstances to generate the historic chill. In order of sequence, an extreme arctic air reservoir set-up in central to northeast Eurasia early-to-middle winter. Stratospheric warming events were the primary trigger. Remember the arctic blast into China in late December? This air mass was from the source region described.
High-latitude, high-pressure blocking became permanent from early December up until a few days ago. During that time the negative arctic oscillation was one of the 4 strongest in the 1950-2020 climatology. There was probably some inevitability that North America would be struck by arctic air despite almost none during December and January.
During early February a pile of heavy convection gathered near the Dateline in the tropical Pacific Ocean. The heat release from this convection motivated a “ridge-bridge” of high-pressure cross-polar across northwestern North America toward Siberia enabling the low-level air mass trajectory to create a river of arctic air into North America. The tropical Madden Julian oscillation amplified a northeast Pacific upper ridge compensated for by a central North America polar vortex which was entraining the cross-polar arctic air.
Fig. 2-4: Day 11-15 forecasts for Feb. 10-16 and the temperature forecast error by the GFS, GFS ENS and ECM ENS models.
So…how did the extended-range models perform? The 11-15 day forecast showing a pattern change toward much colder or much warmer temperatures is generally a market-mover for natural gas. Forecast skill performance by the 11-15 day outlooks centered on Feb. 10-16 by the GFS, GFS ENS and ECM ENS (Fig. 2-4) indicate the models were much too warm (or not nearly cold enough).