Figure Captions

Figure 1: Representative patterns in daily 500-hPa geopotential height anomalies [m, shaded] for all months from 1948-2019 calculated using Self-Organizing Maps. The domain covers 30oN-80oN and 180-60oW (midlatitude North America/NE Pacific). Percentages indicate the frequency of occurrence of each node during winter months (JFM). Numbers to left of each node are for reference purposes. Data to generate the SOM were obtained from the NCEP/NCAR Reanalysis (Kalnay et al., 1996).
Figure 2: Distributions of days (y-axis) of node number (1-12, x-axis) corresponding to two days following an LDE in a particular node. Matrices correspond to node placement in master SOM shown in Fig. 1, indicated with bold numbers in upper left corners.
Figure 3: Change in the frequency of occurrence (days) of each node from 1961-1989 to 1991-2019 during (a) all months and (b) winter months (JFM). The small (large) Xs indicate changes that are statistically significant with 90% (95%) confidence.
Figure 4: Winter (JFM) temperature extremes associated with each node of the master SOM. (a) Number of days (shading) that air temperature anomalies at 925 hPa exceed 1.5 σ. (b) Number of days that air temperature anomalies at 925 hPa fall below 1.5 σ. Data are from the NCEP/NCAR reanalysis.
Figure 5: Winter (JFM) precipitation extremes associated with each node of the master SOM. Shading indicates number of days that daily precipitation anomalies exceed 1.5 σ. Data are from the NCEP/NCAR reanalysis.
Figure 6: (a) Time series of weather whiplash events (WWEs) per year during winters (JFM) from 1949 to 2019 (x-axis), derived using data from the NCEP reanalysis. Bold solid (dashed) lines indicate trend significance with 95% (90%) confidence. (b) displays differences in the number of WWEs during two 20-year intervals: 1950-1969 to 2000-2019. The X indicates statistical significance > 95% based on student’s t-test.
Figure 7: Timeseries of Euclidean distances (y-axis, unitless) between the node in which an LDE occurs and the node that contains the daily field two days after the LDE, averaged over winter months (JFM) each year from 1950 to 2019 (x-axis). Bold (thin) trend lines are significant at 95% (90%).
Figure 8: Changes in the number of winter (JFM) WWEs from 1979-1989 to 1995-2005 in (a) NCEP/NCAR reanalysis output and in historical simulations from three global climate models: (b) CCSM4, (c) CanESM2, and (d) GFDL-CM3. The small (large) Xs indicate statistical significance > 90% (> 95%) based on a student’s t-test.
Figure 9: Projected changes in the number of winter (JFM) WWEs from 2006-2030 to 2076-2100 in (a) CCSM4, (b) CanESM2, and (c) GFDL-CM3. The Xs indicate statistical significance > 95% based on a student’s t-test.