Summary and Conclusions

Abrupt shifts from one anomalous weather condition to another substantially different one are disruptive to agriculture (e.g., thaw/freeze episodes that can cause early budding on fruit trees), winter recreation enterprises, management of municipal utilities, animal behavior, and a variety of human activities. The term “weather whiplash” recently entered the public discourse to describe these types of shifts, but a commonly accepted definition has yet to emerge. In this study we propose a characterization that differentiates normal frontal passages, which often usher in marked but short-lived weather changes, from more disruptive events in which a persistent and anomalous weather regime is abruptly replaced by another strikingly different one. Whiplash events typically involve a persistent winter cold spell that is supplanted by a winter heatwave, a prolonged drought followed by days of storminess, or the reverse order of these transitions.
In this study, we demonstrate a new method to identify and trace weather whiplash events (WWEs) based on characteristic large-scale patterns of daily upper-level (500 hPa) geopotential height anomalies over the domain encompassing the northeast North Pacific Ocean and North America. Patterns are objectively determined from 72 years of daily height fields (from NCEP/NCAR reanalysis), and each day is classified into a matrix of representative clusters or nodes by a neural-network-based tool called self-organizing maps (Figs. 1 and S9 ). Persistent long-duration events (LDEs) are identified when a string of four or more consecutive days occurs in a single cluster. A WWE is then diagnosed when the pattern two days following a LDE transitions to a cluster that is sufficiently different from the one where the LDE occurred, as measured by the Euclidean distance between clusters in the matrix. Extreme temperatures and heavy precipitation associated with each pattern in the matrix are also characterized. This information is used to describe the expected shift in weather conditions before and after a WWE.
We assess the number of WWEs that originate from LDEs occurring in each node during all months (Fig. 6a ), as well as changes in WWE frequency during winter (JFM; Fig. 6b ) and summer (JAS;Fig. 6c ). The clusters dominated by high or low height anomalies spanning high latitudes tend to spawn the most WWEs (nodes #1 and #12). Patterns featuring a distinct meridionally oriented dipole of height anomalies (primarily nodes #3/4 and #9/10) also produce a relatively large number of WWEs.
Changes in WWEs over time are more robust in future model projections assuming minimal abatement of carbon emissions (Fig. 7 ) than during past decades. WWEs originating in nodes #1 and #3 increase significantly from 2006-2030 to 2076-2100 during both winter and summer, while those originating in node #12 decrease significantly. Increasing WWEs are projected for nodes with positive height anomalies in high latitudes, while WWEs will decrease when the Arctic is anomalously cold.
Using extreme temperature and precipitation patterns associated with each node, we analyzed the sensible weather shifts that are likely to occur in nodes where WWEs become more or less likely. In addition to illuminating specific weather transitions in particular regions, we find more generally that a WWE originating in node #1 (large positive height anomalies in high latitudes) is most likely to shift to nodes #3 or #9 (Fig. 2 ), both of which are substantially more meridional in character. This finding suggests that a persistent episode of strong AAW is likely to be followed by a more amplified upper-level circulation pattern. Interestingly, WWEs originating in nodes #1 and #3 are likely to jump between these two nodes, resulting most notably in repetitious cold spells in midlatitudes of North America and fluctuating wet/dry periods in west-central regions. During summer months, WWEs originating in nodes #1 and #3 will produce similar shifts in temperatures and precipitation, but with somewhat less distinct contrasts (Fig. 5 ). Shifts in temperature and precipitation extremes associated with WWEs initiated in each node for any season can be assessed using this methodology.
Our objective in this paper was to demonstrate a new approach to measure the occurrence of WWEs, which large-scale atmospheric patterns are more or less likely to initiate them, the extreme weather conditions associated with the patterns before and after a WWE, and how their frequency may change in the future as greenhouse gas concentrations continue to rise. Future efforts will examine WWEs in other sectors of the globe and investigate relationships between WWE frequency and changes in the climate system, such as rapid Arctic warming, disruptions of the stratospheric polar vortex, and fluctuations in natural climate oscillations. Because WWEs tend to cause abrupt and disruptive shifts in weather conditions, a better understanding of these events – and the extremes associated with them – will better enable preparations by decision-makers that will lessen their impacts.