Subseasonal-to-Seasonal Predictability of Cold Air Outbreaks in the Central United States

Abstract

Wintertime cold air outbreaks (CAOs) in the central United States (US) are high-impact events characterized by the southward intrusion of polar air into the midlatitudes. These events, such as the prolific February 2021 CAO, pose serious socioeconomic, environmental, and infrastructural risks. Despite their significance, the characteristics, evolution, and predictability of CAOs in this region remain underexplored, particularly on the subseasonal-to-seasonal (S2S) timescale - i.e., two weeks to two months, where forecast skill is low. This dissertation addresses these gaps through three main objectives: (1) developing a comprehensive climatology of central US CAOs; (2) identifying and assessing key S2S sources of predictability of central US CAOs; and (3) linking extreme central US cold to electricity demand through weather regimes. The development of a central US CAO climatology and the subsequent analyses reveal that central US CAOs are strongly associated with anomalous ridging over Alaska and across the Arctic, as well as with stratospheric wave reflection and downward propagation of circulation anomalies associated with a weak stratospheric polar vortex (SPV), respectively. Furthermore, CAOs that are associated with anomalous Alaskan ridging also demonstrate a North Pacific wave train emanating from the tropical Pacific. To test the importance of the stratospheric and tropical circulations in central US CAO development, targeted nudging experiments are employed in an S2S forecast model to reveal whether improving the representation of the stratosphere and tropical variability increases S2S skill for two CAO events. For the December 2017/18 CAO event, the nudged forecast runs show significant improvement in the long-lead (Weeks 3-5) forecast for the CAO and the tropospheric circulation pattern. However, for the January 2004 CAO event, limited forecast improvement occurs in the nudging experiments, underscoring event-to-event variability in predictability and evolution of central US CAO events. We then build on our nudging experiments by evaluating S2S model skill in forecasting CAOs in general across multiple operational S2S forecast models. Results show that a persistent weak SPV and the presence of high-latitude blocking improve forecast reliability, while false alarm forecasts can be tied to regime transitions that diverge from the observed evolution. As an example of a CAO with substantial impacts, particularly on the electrical grid, that occurred following weak SPV conditions, we also assess the predictability of the February 2021 CAO using S2S forecasts from two operational models, demonstrating that the accurate simulation of two high-latitude Rossby wave breaking events, and associated blocking patterns, contributed to improved predictability at lead times of 2-3 weeks. Finally, this work establishes a strong link between winter weather regimes, extreme cold episodes, and electricity demand across the central US: regimes with anomalous ridging over Alaska, the Arctic, and the US West Coast (i.e., those most frequently associated with extreme cold) increase the probability of extreme electricity demand by 20-100%. Together, these findings advance our understanding of the dynamical pathways that govern central US CAOs and demonstrate how precursor signals can extend the horizon of skillful forecasts. By clarifying the conditions under which predictability arises and fails, this work lays a foundation for improving S2S prediction of high-impact cold extremes.