Winter 2019 Recap

April 25, 2019

Special recap of this past winter

Summary

Winter 2018/19 was characterized by warmth in Alaska, the Southeastern United States (US), Europe, the Middle East and East Asia but cold in much of Canada, the Northern and Western US, North Africa and Central Asia while Siberia was mixed.
El Niño was anticipated for the upcoming winter. There was some uncertainty whether it would be a Central or Eastern Pacific based El Niño. The thinking was a Central based El Niño would favor an overall colder winter in the Eastern US. However, sea surface temperatures (SSTs) in the equatorial Pacific did not support El Niño until late in the winter and I don’t believe that El Niño was much of a factor on the Northern Hemisphere winter weather.
October Siberian snow cover was slow to advance most of the month but had a rapid spurt to end the month. October Siberian snow cover extent (SCE) was above normal but less than recent years. Still I believe that it was influential in modulating the behavior of the stratospheric polar vortex (PV) in early- to mid-winter.
Low sea ice in the Barents-Kara seas helped anchor high pressure in the region, especially Northern Europe, for much of November and December. Prolonged Scandinavian blocking resulted in a stratospheric polar vortex split in early January with the minor daughter vortex over Europe and the major daughter vortex over North America.
The PV disruption was the dominant weather event of the North American winter resulting in a relatively cold mid-latitude winter and mild Arctic across the North American continent.
In contrast the impact from the PV disruption was more limited across Eurasia. The greatest negative departures for the whole Northern Hemisphere (NH) were observed in Central Asia with more regional cold temperatures in parts of the Mediterranean and Eastern Siberia. However much of Asia and especially Europe experienced a relatively mild winter.
This winter retrospective is consistent with my scientific publications related to the importance of Arctic influence, troposphere-stratosphere-troposphere coupling and the polar vortex on mid-latitude winter weather. This retrospective while based on the scientific literature is my opinion only.

Boundary Forcings

The climate community focuses on El Niño/Southern Oscillation or ENSO in making seasonal forecasts and an El Niño event was predicted by the models for winter 2018/19. El Niño favors milder temperatures in the Northwestern US and colder temperatures in the Southeastern US. El Niño also favors light precipitation in the Northwestern US and heavier precipitation in the Southwestern US including California. During the winter of 2018/19 a weak El Niño conditions did not materialize until late winter. Most winter forecasts, especially from the government forecast centers, relied heavily on the anticipated El Niño for their temperature and precipitation anomaly forecasts. In general, these forecasts performed poorly especially for temperature. As an example, NOAA’s Climate Prediction Center (CPC) United States (US) temperature forecast is shown along with the observed temperature anomaly forecast in Figure 1. My graphic of observed temperature anomalies is based on the relatively coarse NCEP/NCAR reanalysis and smoothing is applied so the anomalies don’t necessarily match station data, especially along the transition zone from positive to negative anomalies. Still the headline wouldn’t change, regardless of the dataset and the level of smoothing - ENSO derived temperature forecasts performed poorly and are almost the exact opposite of what was observed. I will no longer discuss ENSO in the remainder of the blog post.

Figure 1. a) Forecasted surface temperature anomalies from NOAA's Climate Prediction Center and b) observed surface temperature anomalies (°F; shading) for December 1, 2018 through February 28, 2019

At AER we use ENSO in producing seasonal forecasts, but in addition we have pioneered the use of Arctic boundary forcings in winter seasonal forecasting including Arctic sea ice but especially Eurasian snow cover in October. We have demonstrated using observational analysis and model perturbation experiments that extensive Eurasian October snow cover is related to/can force a strengthened Siberian high, increased poleward heat flux, a weak stratospheric polar vortex (PV), which culminates in an extended period of a negative Arctic Oscillation (AO). A negative AO is associated with below normal temperatures in the Eastern US and Northern Eurasia including Northern Europe and East Asia. Scientists including those at AER, have shown a similar atmospheric response to low Arctic sea ice. There are different ideas how variability in Arctic sea ice might influence winter hemispheric weather, but the trend has been a convergence to a similar set of mechanisms first proposed for Eurasian snow cover. Also there is growing consensus that it is Barents-Kara sea ice in the late fall and early winter that has the greatest impact across Eurasia. Therefore low Barents-Kara sea ice in November for example, favors a strengthened Siberian high, increased poleward heat flux, a weak PV and finally a negative AO. An important point in regard to the Siberian high, it strengthens or expands northwest of the climatological center. For low snow cover and/or high sea ice the opposite occurs.

October 2018 Eurasian snow cover extent (SCE) was above normal but lower than recent Octobers (Figure 2a). Snow cover advance was relatively slow for the first three weeks of October and then had a healthy spurt to end the month. Above normal SCE favored a sudden stratospheric warming (SSW) and a weak PV in mid-winter followed by a negative AO and cold temperatures across the NH mid-latitudes. I also compute the snow advance index (SAI) which is a measure of the pace or speed of the snow cover advance across Eurasia (see Cohen and Jones 2011 for more detail). The value was close to zero or neutral though I thought the slow start and fast finish of SCE during the month of October was at least in the “spirit” of a positive SAI value. Recently I have used the SAI mostly as an indicator of the timing of a possible PV disruption and I interpreted the SAI to favor an earlier PV disruption more so than a late PV disruption for winter 2018/19. But I readily admit this is highly speculative. It has worked well the past two winters but is no guarantee of future success.

Figure 2. a) Snow cover extent across Eurasia on October 31, 2018 shown in white, Arctic sea ice shown in light blue. Gray shading snows all points north of 60°N across Eurasia. b) Observed Arctic sea ice extent anomalies November 2018. Negative anomalies shown in blue shading.

Fall 2018 Arctic sea ice was below normal (Figure 2b) and for the first time I tried to estimate November 2018 Barents-Kara sea ice as a predictor in the AER model. I estimated Barents-Kara sea ice to be around 1 standard deviation below normal which was close to the observed value. Therefore, Arctic sea ice similarly favored a weak PV in mid-winter followed by a negative AO and cold temperatures across the NH mid-latitudes.

The quasi-biennial oscillation (QBO) was in its westerly phase. The QBO is a periodic oscillation of the zonal winds in the equatorial stratosphere and in the westerly phase the zonal winds are stronger. The westerly phase is thought to inhibit the strongest SSWs where the zonal wind reverses in direction and is referred to as a major midwinter warming.

And of course, the near record warm global atmosphere and ocean provided an overall warm backdrop heading into the winter of 2018/19.

Late fall/early winter

As mentioned above, October 2018 Eurasian snow cover extent was above normal due to a late month surge in snow cover advance. Above normal snow cover across Siberia in October favors a strengthened Siberian high in November and into December with the largest positive sea level pressure (SLP) anomalies northwest of the climatological center (see Figure 3a taken from Cohen et al. 2014). The rapid advance of snow cover at the end of October favored the northwestward expansion of the Siberian high in November which persisted for an unusually long time well into December (Figure 3b). Below normal sea ice in the Barents-Kara seas is also associated with the northwestward expansion of the Siberian high and the combination of the rapid advance in snow cover at the end of October and low Barents-Kara sea ice contributed to a long-lasting episode of blocking centered near Scandinavia for the months of November and December. From Figure 3, the northwestward expansion of the Siberian high is clearly evident as well as the “classic” tripole SLP anomaly pattern with relatively high pressure near Scandinavia/Urals and low pressure in both the North Atlantic and North Pacific Ocean basins that is the hemispheric circulation that is most favorable for disrupting the stratospheric PV.

Figure 3. a) Regression of November SLP anomalies (hPa) onto October monthly mean, October Eurasian SCE (contouring) and December meridional heat flux anomalies at 100 hPa, averaged between 40-80°N (shading). b) Observed average sea level pressure (hPa; contours) and sea level pressure anomalies (hPa; shading) across the Northern Hemisphere from November 1, 2018 through December 31, 2018.

This persistent and strong tripole pattern is optimal for forcing increased vertical transfer of Rossby wave energy (vertical wave activity flux or WAFz) and poleward heat flux. The WAFz plot in Figure 4 shows active WAFz throughout November and December. During the two-month period of active WAFz, the first and last pulses were notable for being the two strongest pulses of the two-month period – the first pulse in early November and the last pulse at the end of December.

Figure 4. Observed daily vertical component of the wave activity flux (WAFz) standardized anomalies, averaged poleward of 40-80°N from October 1, 2018 through March 31, 2019.

I have believed for a while now that conventional thinking of the behavior of the stratospheric PV as binary was overly simplistic and too limiting when trying to understand tropospheric influence, i.e., a strong PV and a weak or disrupted PV that is only really considered when a major mid-winter warming (MMW) is observed, which is identified when the zonal mean zonal wind reverses from westerly to easterly at 60°N and 10 hPa. I was able to find new ways to characterize PV behavior when graduate student Marlene Kretschmer visited me at AER. In our first collaborative paper we identified seven main clusters or categories of PV behavior from the strongest PV in cluster one and the weakest PV in cluster seven (see Kretschmer et al. 2018a). In a follow up paper instead of applying cluster analysis at 10hPa (the middle stratosphere) we applied it at 100 hPa (the lower stratosphere) where we identified five clusters (see Kretschmer et al. 2018b). Again, cluster one was the strongest PV and cluster five the weakest, but we also identified a new weak cluster (cluster four). We differentiated cluster four from five in that in cluster five the WAFz is absorbed in the polar stratosphere and in cluster four the WAFz is reflected off of the stratospheric PV back into the troposphere.

It turns out that the tropospheric response to a polar vortex disruption where WAFz is “reflected” is quite different from when WAFz is “absorbed.” The tropospheric response to a PV disruption where the WAFz is absorbed is the “classic” response to stratospheric PV disruptions. The tropospheric response is characterized by Greenland blocking, negative North Atlantic Oscillation (NAO), relatively cold temperatures across northern Eurasia and milder across North Africa, the Middle East and the North American Arctic. Though it was not included in the paper, our analysis did show that in about two weeks-time the cold temperatures overspread the Eastern US as well. Also, the tropospheric response is usually delayed relative to the WAFz pulses and the response can be long lasting of up to two months. In contrast the tropospheric response to a PV disruption where the WAFz is “reflected” is characterized by blocking near Alaska, relatively cold across much of Canada, the Eastern US and Central Asia and mild across Alaska and Europe. The response is not associated with a negative NAO but rather a negative West Pacific Oscillation (WPO). Also, the tropospheric response is usually rapid relative to the WAFz pulses and the response is of relatively short duration lasting on the order of days and up to two weeks.

I admit that I have not done a detailed analysis but the information in the provided figures are at least broadly consistent with my ideas. The WAFz through at least the end of November and possibly through early to mid-December is reflective and bouncing off of the stratospheric PV. In the stratosphere, the WAFz values are predominantly positive or enhanced upward but, in the troposphere, the WAFz are predominantly negative. To be clear a negative value does not necessarily mean downward WAFz but instead less upward and from the plot I cannot differentiate. As I just mentioned reflective WAFz results in blocking/high pressure as well as warming near Alaska with downstream troughing and cold temperatures across Central and Eastern North America first in the stratosphere and quickly followed in the troposphere. In Figure 5 I present the geopotential heights for both 10 hPa and 500 hPa from November 1-December 15, 2018. Figure 5a closely matches the “reflective” cluster 4 for the stratospheric PV (see Figure 1 from Kretschmer et al. 2018b) with positive geopotential height anomalies centered near Alaska and Eastern Siberia with downstream negative geopotential heights across central and eastern North America that extends into the North Atlantic.

Figure 5. a) Observed average 10 hPa geopotential heights (dam; contours) and geopotential height anomalies (m; shading) across the Northern Hemisphere for November 1- December 15, 2018 b) Observed average 500 hPa geopotential heights (dam; contours) and geopotential height anomalies (m; shading) across the Northern Hemisphere from November 1 - December 15, 2018.

The general circulation pattern in the stratosphere can be seen replicated in the troposphere with ridging/positive geopotential height anomalies across western North America with downstream troughing/negative geopotential height anomalies across eastern North America. Also, a region of negative geopotential height anomalies was observed in Central Asia both in the stratosphere and the troposphere. Probably the biggest difference was the strong blocking centered on Scandinavia in the troposphere and not the stratosphere. If, as I argue that the Scandinavian blocking is in part forced by high snow and low sea ice, this would make sense since the forcing is bottom up and not top down.

I do believe that the strong Scandinavian blocking coupled with a reflective stratospheric PV disruption can explain the large-scale surface temperature anomalies (Figure 6). The hemispheric temperature anomalies are consistent with the temperature anomalies associated with a reflective stratospheric PV disruption as seen in Figure 5a of Kretschmer et al. 2018b. Warm anomalies in Alaska and the West Coast of North America that extend into Eastern Siberia (though there was some regional cooling) with cold anomalies widespread across central and eastern North America. Europe is warm as well with downstream cold temperatures in Central Asia. But the temperature pattern is also consistent with the Scandinavian blocking with cold advection downstream of the block on northerly flow across Siberia and Central Asia and warm advection upstream of the block on southerly flow across Europe. It does seem to me the Scandinavian block and the tropospheric response to a reflective stratospheric PV disruption reinforced each other or constructively interfered across Eurasia during the six weeks of November 1 – December 15, 2018.

Figure 6. a) Observed average surface temperature anomalies ((°C; shading) for November 1 - December 15, 2018.

I would argue that during the month of December the pulses of WAFz transitioned from reflective to absorptive. The last WAFz pulse the second half of December impacted the stratospheric PV differently than the previous pulses. In Figure 7 I show the 10 hPa geopotential heights for the second half of December through January 2, 2019, the date an MMW was first achieved, and clearly the impact of the WAFz is different than earlier in the fall and early winter. November 1 – December 15 the stratospheric PV was stretched but the core of the PV/central Arctic was impervious to warming. After December 15 the WAFz resulted in strong warming of the entire polar stratosphere and the stratospheric PV is now displaced towards Scandinavia. This can also be seen in the plot of the polar cap geopotential heights (PCHs), which shows a similar transition (Figure 8). PCHs in the mid-stratosphere remained cold/negative in the fall through December 15 while PCHs in the troposphere remained warm/positive throughout this period.

Figure 7. a) Observed average 10 hPa geopotential heights (dam; contours) and geopotential height anomalies (m; shading) across the Northern Hemisphere for December 16, 2018 - January 2, 2019 b) Observed average 500 hPa geopotential heights (dam; contours) and geopotential height anomalies (m; shading) across the Northern Hemisphere for December 16, 2018 - January 2, 2019.

Often in the blog I discuss how if WAFz is resulting in a negative AO, high pressure and warming in the polar stratosphere then it is likely resulting in the opposite, a positive AO, low pressure and cooling in the polar troposphere. Also shown in Figure 7 are the 500 hPa geopotential heights for the second half of December through January 2, 2019. In comparison to previous weeks the circulation looks much more positive AO with a less amplified flow, low heights over the central Arctic, mostly positive heights over the mid-latitudes.

Figure 8. Observed daily polar cap geopotential height (i.e, area-averaged geopotential heights poleward of 60°N) standardized anomalies from October 1, 2018 through March 31, 2019.

Hemispheric continental temperatures are also relatively mild, including much of the US and Europe (Figure 9). The biggest exception is Asia with cold temperatures stretching from Western Asia, across Central and into East Asia. I also often mention that western North America and East Asia are frequently cold during these active WAFz/positive AO periods, so it is not surprising that it is cold across a large portion of Asia during this period. Still continued blocking across Scandinavia over to the Urals forced troughing downstream with cold temperatures across Asia.

Figure 9. a) Observed average surface temperature anomalies ((°C; shading) for December 16, 2018 - January 2, 2019.

Midwinter

The strong upward WAFz absorbed in the stratosphere at the end of December weakened the stratospheric PV and can be seen in the warm/positive PCHs from the end of December through all of January (Figure 8). In Figure 10 a&b I show the 10 hPa geopotential heights both in early and late January. The stratospheric PV disruption resulted in a PV split. Initially one daughter vortex was located over Eastern Europe, though it extends across all of northern Asia, with the second daughter over Southeastern Canada and New England.

Figure 10. Observed average 10 hPa geopotential heights (dam; contours) and geopotential height anomalies (m; shading) across the Northern Hemisphere for a) January 3, 2019 - January 14, 2019 and b) January 15, 2019 - January 31, 2019. Observed average 500 hPa geopotential heights (dam; contours) and geopotential height anomalies (m; shading) across the Northern Hemisphere from c) January 3, 2019 - January 14, 2019 and d) January 15, 2019 - January 31, 2019.

Again, in the blog I have often discussed that in my opinion there exists an initial tropospheric response and a delayed tropospheric response to PV disruptions. In Figure 10 c&d I also include the 500 hPa geopotential heights both in early and late January and in Figure 11 the surface temperatures for early and late January. In early January troughs with related cold temperatures exist in the troposphere directly beneath the two stratospheric daughter PVs one in Eastern Europe, though the troughing extended across northern Asia, and the other in Southeastern Canada and New England.

Figure 11. Observed average surface temperature anomalies ((°C; shading) for a) January 3, 2019 - January 14, 2019 and b) January 15, 2019 - January 31, 2019.

Initially the major daughter stratospheric vortex was over Eurasia and the minor daughter stratospheric vortex over North America, however over time the daughter vortex over North America deepened while the daughter vortex over Eurasia filled. In the second half of January the North American daughter vortex retrograded westward and became the major daughter vortex. The Eurasian daughter vortex drifted to the east and weakened and became the minor daughter vortex, though troughing continued to extend across much of northern Eurasia.

In parallel the tropospheric trough over North America deepened and retrograded westward. With the deepening trough, cold temperatures became more widespread across central and eastern North America. Over Eurasia the trough did become more expansive across Europe the second half of January with cold temperatures now extending from Eastern to Western Europe, though not all of Europe was cold. Cold also extended across all of northern Asia though its southern extend was quite limited and was mostly confined to Northern Siberia.

Overall, I think the troposphere responded to the stratospheric PV disruption close to our expectations with some notable exceptions. In Figure 12 (a figure I have shown often in the blog), I show the weekly temperature response for weeks one through three during and after a significant stratospheric PV disruption. This figure was included in the original Kretschmer et al. 2018a paper but was removed at the request of the reviewers. During the week of the PV disruption the largest and most widespread negative temperature departures are found across northern Eurasia with much weaker anomalies across North America. However, over time the temperature anomalies weaken across Eurasia but strengthen across North America so that by week three the greatest negative and most expansive temperature departures are found across central and eastern North America with weaker and more regional temperature anomalies across Eurasia. I think that the hemispheric temperature anomalies were broadly consistent with Figure 12 in January 2019. Negative temperature anomalies were initially deeper and more widespread across Eurasia, but during the month the negative temperature anomalies weakened across Eurasia while strengthening across North America.

Figure 12. Composite mean of detrended lagged near surface temperature for cluster 7 (weak PV) using a) a lag of 0-7 days, b) a lag of 7-14 days and c) a lag of 14-21 days. In all panels significant values (P<0.05) are indicated with hatches.

However, there are two notable exceptions that I don’t have a good answer for that separated the tropospheric response to a significant PV disruption in 2019 compared to 2018. First the initial negative temperature response to the PV disruption across Eurasia was weaker and more regional than might have been expected. Second no strong Greenland blocking developed following the PV disruption this is not only seen in Figure 10 but also in Figure 11, where there is a lack of strong surface warming in Northeastern Canada and Greenland as seen in Figure 12. I did often point out the surprisingly cold central Arctic this past winter and it is possible that the cold Arctic inhibited more widespread and persistent cold across Eurasia and Greenland blocking.

Late winter

I believe that persistence is strong in late summer and winter. It is my opinion that persistence could go a long way in explaining the observed surface temperature anomalies in February and March 2019. The tropospheric trough and associated cold surface temperatures that began along the eastern edge of North America in early January kept retrograding during January and into February where it was centered in Western Canada and the Western US for the month (Figures 13 and 14). They seemed to follow the North American daughter PV that retrograded westward during the month of January. I would argue that the widespread cold and deep snow cover established in January strongly favored for the persistence of cold into February.

Figure 13. Observed average 10 hPa geopotential heights (dam; contours) and geopotential height anomalies (m; shading) across the Northern Hemisphere for a) February 1 - February 28, 2019 and b) March 1, - March 31, 2019. Observed average 500 hPa geopotential heights (dam; contours) and geopotential height anomalies (m; shading) across the Northern Hemisphere from c) February 1 - February 28, 2019 and d) March 1, - March 31, 2019.

Across Eurasia the cold that was not as deep or as widespread as might be expected following a stratospheric PV disruption in January, continued to be regionalized in February. Europe was dominated by ridging and relatively mild temperatures (Figures 13 and 14). Sea ice was close to normal in most of the Arctic except in the Barents-Kara Seas, which would favor blocking/high pressure across Northern Europe. Instead in February and March the ridging was centered further south over Central Europe resulting in widespread mild temperatures and I don’t have an answer for the difference between my expectations and what was observed.

Figure 14. Observed average surface temperature anomalies ((°C; shading) for a) February 1 - February 28, 2019 and b) March 1, - March 31, 2019.

I think the surprise of how mild Europe was and the lack of Greenland blocking following the stratospheric PV disruption is related to the lack of “dripping” of warm/positive PCHs from the stratosphere to the troposphere. There was only one identifiable drip at the end of January and early February (Figure 8). Eurasian surface temperatures are more closely related to the phase of the AO/NAO than North American surface temperatures. So, the lack of warm/positive tropospheric PCHs, the overall positive AO/NAO was favorable for mild temperatures across Eurasia. I don’t have a good answer for the lack of a more emphatic descent of warm PCHs other than to say again the surprisingly cold Arctic might have interfered with stratospheric influence on the tropospheric circulation. This resulted in colder tropospheric PCHs and a more positive AO/NAO following other recent significant PV disruptions.

When the winds go easterly in the polar stratosphere inhibits WAFz and overall the WAFz was quiet for all of January and even into February and early March. Therefore, it is no surprise that the stratospheric PV continued to strengthen after the PV disruption in early January right into mid-March (Figures 8 and 13), when it became record strong. Overall not sure how much of an influence the stratospheric PV was in February and March on the tropospheric circulation. But if one were to extrapolate the tropospheric circulation from the stratospheric circulation, the flow is westerly, favoring mild temperatures, across Eurasia but northerly, favoring cold temperature, across western North America (Figure 13 and 14). I think this is a chicken and egg problem, but it seems that the circulation in the stratosphere supported relatively mild temperatures in Eurasia but cold temperatures in North America.

I do think that stratosphere-troposphere coupling resumed in March. In late February and early March more positive WAFz are observed quickly followed by negative WAFz (Figure 4). This striation WAFz pattern is similar to the pattern in November when I argue a “reflective” stratospheric PV disruption occurred. Similarly, the WAFz, the blocking near Alaska accompanied by mild temperatures (admittedly more impressive in the troposphere than in the stratosphere) and downstream cold across the US and Southern Canada but mild in Europe are all hallmarks of a reflective PV disruption. I would go further and argue that also like the reflective PV disruption in the early winter that transitioned to an absorptive PV disruption, this late winter reflective PV disruption transitioned into an absorptive PV disruption in April but will always be remembered as a Final Warming.

Observed winter circulation

In my opinion the rapid advance of snow cover at the end of October, the strengthened and westward displaced Siberian high during November and December, the active upward WAF, the weak PV, the drip of warm/positive PCHs from the stratosphere into the troposphere was a textbook example of snow-atmosphere coupling.

To support my argument, I show again the PCH plot for the winter with a composite of PCHs from years when Eurasian October snow cover extent was high minus years when it was low (Figure 15) and the same for the WAFz plot (Figure 16). The SCE composites were created by my former colleague Jason Furtado but never published. For much of fall 2018, the tropospheric PCHs are warm/positive while the stratospheric PCHs are cold/negative. This is consistent with the composite using SCE. Similarly, the active WAFz observed in December 2018 is consistent with the composite using SCE. Finally, the significant PV disruption characterized by warm/PCHs in the stratosphere and the subsequent downward propagation or dripping of warm/positive PCHs are very similar in 2019 and in the SCE composite, and the timing is nearly perfect. I think the similarities between winter 2018/19 and the SCE composite are very striking. Of course, the devil is in the details and the biggest difference between winter 2018/19 and the SCE composites is that in the SCE composites the warm/positive emphatically descend from the stratosphere to the troposphere and that warm/positive PCHs dominate the lower troposphere from mid-January through mid-February. In 2019 the same descent of warm/positive from the stratosphere to the troposphere is much more tepid with the signal getting hung up at the tropopause for an extended period. Only a weak and short-lived warm/positive PCHs is observed to descend into the lower troposphere in late January and early February. I don’t think that the weak tropospheric response is related to the advance of SCE but instead the only explanation that I have offered is the surprisingly cold central Arctic this winter.

Figure 15. a) Observed daily polar cap geopotential height (i.e, area-averaged geopotential heights poleward of 60°N) standardized anomalies from October 1, 2018 through March 31, 2019. b) composite of polar cap geopotential height from October 1 through March 31 for winters when October Eurasian snow cover extent is high minus low.

In Figure 17 I show the winter mean (December-February) circulation in the mid-stratosphere (10 hPa) and mid-troposphere (500 hPa). The impact of the stratospheric PV disruption/split in January is evident on the winter mean anomalies. Positive height anomalies cover the Arctic with negative height anomalies across the mid-latitudes readily recognizable as a classic negative AO pattern. But this pattern did not translate into a similar pattern in the troposphere. Instead the Central Arctic is characterized by negative height anomalies with mostly positive height anomalies across the mid-latitudes. This is yet another illustration of the weak downward propagation of geopotential height anomalies from the stratosphere to the troposphere.

Figure 16. Observed daily vertical component of the wave activity flux (WAFz) standardized anomalies, averaged poleward of 40-80°N from October 1, 2018 through March 31, 2019. b) composite of WAFz from October 1 through March 31 for winters when October Eurasian snow cover extent is high minus low.

Not surprisingly positive heights cover Alaska, but the ridge is not aligned along a north-south axis, which is typical but rather along a northeast-southwest axis that contributed to a mild winter across Alaska but also allowed cold air to slide into Western North America. Troughing in Eastern Canada pulled the cold air into Eastern Canada and Northern New England as well. Ridging centered over Western Europe resulted in a mild winter for most of Europe with relatively colder air in Southeastern Europe and especially North Africa. More troughing in Central Asia and Eastern Siberia supported more cold temperatures in those regions.

Figure 17. a) Observed average 10 hPa geopotential heights (dam; contours) and geopotential height anomalies (m; shading) across the Northern Hemisphere for December 1, 2018 - February 28, 2019 b) Observed average 500 hPa geopotential heights (dam; contours) and geopotential height anomalies (m; shading) across the Northern Hemisphere from December 1, 2018 - February 28, 2019.

I have to admit that I am surprised by the positive height anomalies focused over Greenland. Not sure how that happened but I did check and that is not an error of my plotting. Still the low heights elsewhere across the Arctic is the surprise of the winter and will be interesting to see if it turns out to be a one off or a more persistent feature.

Finally, I include an extended animation of the stratospheric PV from December through early February that shows the variability in the PV and the long-lived PV split during all of January in Figure 18.

*Click for ANIMATION

Figure 18. Polar vortex animation. a) Observed average 10 hPa geopotential heights (dam; contours) and geopotential height anomalies (m; shading) across the Northern Hemisphere for December 1, 2018 - February 5, 2019.

Winter forecasts

The main predictors in the AER winter forecast are October Eurasian SCE, the Barents-Kara sea ice anomaly and El Niño. In mid-October I published the AER winter forecast on the National Science Foundation website. That forecast is relatively mild across the US and Canada as I was expecting below normal October Eurasian SCE when the forecast was published. I also published an updated NH version of the forecast that was colder due to the late surge in Eurasian SCE in the blog in November (Figure 19). A US only version was published in the Washington Post. Snow cover was high and sea ice low so therefore the model forecast was consistent with a weak polar vortex and a negative AO. The model predicted cold temperatures across Northern Asia, Northern Europe and seasonably cold for much of the US, centered in the west-central US. Much of the remainder of the NH was predicted to be mild. The observed winter temperatures are also shown in Figure 19. The model forecast was mixed with the model performing better across North America than Eurasia as a whole. The biggest model error was that predicted temperatures were generally too cold compared to the observations in Northern Europe. In North America the model was generally too warm except in Southeastern US where the model was too cold.

Figure 19. Predicted December, January and February 2018/19 surface temperature anomalies from the a) NMME suite of models, b) C3S suite of models both initialized November 1, 2018, c) the CFSv2 model and d) the AER statistical model initialized November 8, 2018. e) the observed surface temperature anomalies for December, January and February 2018/19. Smoothing was applied to the statistical model and observed surface temperature anomalies.

I also include the national multi-model ensemble (NMME- an ensemble of North American models) forecast for NH temperatures, the NOAA CFSv2 and the European model ensemble (C3S) in Figure 19. As is the case every winter now, the dynamical models predict almost universal above normal temperatures across the NH continents. And once again the NMME winter temperature forecasts are in general too warm. The one region where the models predict closer to normal or slightly below normal temperatures is in the Eastern US and this is related to the predicted El Niño. I attribute the consistent model error of being too warm to model deficiencies in simulating high latitude surface-atmosphere coupling. And similar to last winter the occurrence of a polar vortex disruption is of little consequence in the dynamical models. Amazingly, the CFS did predict a cold winter in the Central Arctic. Not sure why but that was quite the feat! Otherwise the model did poorly like the rest.

Concluding remarks

I believe that the winter of 2018/19, like winter 2017/18 demonstrated the importance of the polar vortex on our winter weather and the benefit of using predicted behavior of the polar vortex in winter forecasts. I started alerting to the strong possibility of a PV disruption on Twitter as early as early November and predicted a significant PV disruption in late December/early January in the blog dated November 19, 2018. But there is large event-to-event variability especially in the response in NH surface temperature anomalies and a correct forecast of a PV disruption does not necessarily translate into a correct forecast of surface temperature anomalies.

I also believe that Arctic boundary forcings are the best available predictors of the possible behavior of the polar vortex. I am not going to get into the details here but ENSO, the Madden Julian Oscillation (MJO) and the QBO are all proposed as influencers of the stratospheric PV and all were in different phases in winter 2017/18 and winter 2018/19. Eurasian snow cover and Barents-Kara sea ice were in the same phase though I readily admit two years is too small of a sample to draw conclusions from but still an interesting footnote.

In conclusion, this fall highlighted the importance of the northwestward expansion of the Siberian high referred to as Scandinavian/Ural blocking in initiating a significant PV disruption more than any other fall that I can recall. Scandinavian/Ural blocking variability has been linked to anomalies in Eurasian SCE and Barents -Kara sea ice. I argue that above normal SCE and below normal sea ice contributed to the strong and persistent Scandinavian/Ural blocking in the late fall and early winter. The Scandinavian/Ural blocking in turn triggered the stratospheric PV disruption/split in January. The PV disruption was critical to understanding the general circulation of the NH during the entire winter of 2018/19.