Comparison of stratospheric evolution during the major sudden stratospheric warming events in 2018 and 2019

Using Modern‐Era Retrospective analysis for Research and Applications, Version 2 (MERRA‐2) data in the northern hemisphere at the 10 hPa level, we compared the stratospheric evolution of temperature and geopotential height during two major sudden stratosphere warming events (SSWs) that occurred in the Arctic winter of 2018 and 2019. In the prewarming period, poleward temperature‐enhanced regions were mainly located around 120°E with a displaced vortex and around 120°E and 60°W with splitting vortices. The evolution of geopotential height indicated that these temperature‐enhanced regions were both on the western side of high‐latitude anticyclones. In the postwarming period, the polar vortex turned from splitting to displacement in the 2018 SSW but from displacement to splitting in the 2019 SSW. Both transitions were observed over the Atlantic region, which may have been caused by anticyclones moving through the polar region. Our findings revealed that the evolution of the anticyclone is important during SSWs and is closely related to temperature‐enhanced regions in the prewarming periods and to transitions of the polar vortices in postwarming periods.


Introduction
Sudden stratospheric warming (SSW) is a large-scale meteorological event that always occurs with the winter polar stratospheric temperature increasing sharply within several days (Matsuno, 1971;Andrews et al., 1987). This polar event is reported to have a large impact on the atmosphere and on the ionosphere at other latitudes (e.g., Ayarzagüena et al., 2011;Gong Y et al., 2013Gong Y et al., , 2016Gong Y et al., , 2018aCoy and Pawson, 2015;Butler et al., 2017;Ma Z et al., 2017Ma Z et al., , 2018Ma Z et al., , 2020Xiong JG et al., 2018;Li N et al., 2020). A major SSW event is associated with a reversal of the zonal mean zonal wind (from eastward to westward) at 10 hPa of 60°N and a posit-ive zonal mean temperature difference between 90°N and 60°N at the 10 hPa level. In general, the first date of the wind reversal is defined as the central date of the major SSW event. During major SSWs, the polar vortex moves southward from the North Pole or splits into two daughter vortices, which are always classified as "displacement-type SSW" events and "split-type SSW" events. The displacement-type SSWs are generally believed to be caused by strong planetary waves with a zonal wavenumber of 1 (wave 1), whereas the split-type SSWs are mainly due to the enhancement of planetary waves with a zonal wavenumber of 2 (wave 2; e.g., Charlton and Polvani, 2007;Karpechko et al., 2018;Rao J et al., , 2019aRao J et al., , 2019b. The statistical characteristics of the planetary waves during major SSWs and the two vortex types have also been verified by modeling analyses (e.g., Cao C et al., 2019;Liu SM et al., 2019). However, the polar vortices reveal different statuses before and after the central dates during some atypical events, Recently, in 2018 and 2019, two major SSWs occurred continuously during the Arctic winters (e.g., Rao J et al., , 2019bRao J et al., , 2020. The 2018 SSW, with a central date of 12 February 2018, was found to be related to the extremely cold winter over the European region (King et al., 2019;Lü ZZ et al., 2020) and ionospheric perturbations over the China sector (Liu GQ et al., 2019). The 2018 SSW was excited by the upward wavenumber 2 planetary waves, which were mainly related to the Ural and Alaska blockings . The 2019 New Year SSW had a central date of 02 January 2019. Lee and Butler (2020) proposed that the 2019 SSW was triggered by a dominant wave 1 pattern without the wave 2 amplification seen in the 2018 SSW, whereas splitting polar vortices appeared again after the central date of the 2019 SSW. With the splitting vortices in the postwarming period, the 2019 SSW event was then regarded as a mixed-type (displacement-split) event (Rao J et al., 2019b). It is interesting to note that the splitting vortices could be observed in both wave 1-dominant and wave 2-dominant SSW events. The transitions of the polar vortices complicate the atmospheric variations during major SSW events. However, the stratospheric similarities and differences between conventional SSWs and atypical SSWs with transition vortices are not fully understood. The stratospheric conditions during different SSWs with transition vortices may also exhibit large discrepancies in the prewarming and postwarming periods.
In the present study, we compared the stratospheric evolution in the northern hemisphere at 10 hPa during the major SSWs in 2018 and 2019, focusing on the temperatures, geopotential heights, and planetary waves separately in the prewarming and postwarming periods. Our primary goal was to present the differences between the two warming events and to investigate the similarities of SSWs with transition vortices. The data and methodology applied in this study are presented in Section 2. The results and discussion are given in Section 3. Conclusions are provided in Section 4.

Data and Methodology
Comparisons of the stratospheric evolution during the two major SSWs were made based on temperature and geopotential height data at 10 hPa from the Modern-Era Retrospective analysis for Re-search and Applications, Version 2 (MERRA-2) reanalysis data set (Gelaro et al., 2017). Variations in the stratospheric temperatures were investigated to track the warming process. The evolution of the polar vortices was then presented via maps of the geopotential heights. Low values of the geopotential heights represent cyclones (polar vortices), whereas high values denote anticyclones. Note that the geopotential height data obtained from MERRA-2 and the geopotential data downloaded from the European Centre for Medium-Range Weather Forecasts Interim Reanalysis (ERA-Interim) revealed a consistent distribution of cyclones and anticyclones (Dee et al., 2011). The evolution of cyclones and anticyclones is discussed from 15 days before the central dates to 15 days afterward, whereas the results are shown every other day to avoid having an excessive number of figures. Throughout the 31-day period, the amplitudes of the stationary planetary waves (SPWs) with wavenumber 1, wavenumber 2, and wavenumber 3 were calculated according to the geopotential height data at 60°N and 10 hPa by the harmonic fitting analysis introduced by Lu X et al. (2018). The evolution of the North Atlantic Oscillation (NAO) indices was accessed from the National Centers for Environmental Prediction Climate Prediction Center (NCEP CPC) website (https://www.cpc.ncep.noaa.gov/). Note that the central date of each SSW in the present study was defined as the first date of the wind reversals at 60°N based on MERRA-2 data. The onset date of each SSW was defined as the date with the maximum positive temperature difference between 90°N and 60°N. Our definitions of the onset dates may differ slightly because of the definition by wind reversal, but they were used mainly to reveal the temperature enhancements in the polar region (Gong Y et al., 2019;Ma Z et al., 2020). In addition, definitions of the SSW type have varied in different studies. As mentioned above, different analytical methods can produce different types when classifying the same SSW event. In the present study, we analyzed the vortices based on the daily evolution of the maps of geopotential heights and discuss the transitions of the polar vortices. Our conclusion is not dependent on the classification of the vortex type.

Prewarming Comparisons
Because the transitions of polar vortices occur after the central dates, our comparisons between the two major SSWs were divided into two periods. One was in the prewarming period (before the central dates). The other was during the postwarming period (after the central dates). Figure panels 1a and 1b present the daily mean temperatures in the northern hemisphere at 10 hPa in the 15 days prior to the central dates of the 2018 and 2019 SSWs, respectively. As shown in Figure 1a, lower temperatures were mainly in the western hemisphere from day −15 to day −5 in the 2018 SSW. Meanwhile, higher temperatures propagated continuously from the mid-latitudes around 60°E northeastward to high latitudes around 120°E, but they were unable to occupy the entire Arctic region. On day −3, in addition to the temperature-enhanced region around 120°E, a warming area occurred at the midlatitudes around 60°W. This new warming area expanded poleward into the high latitudes in the next 2 days and connected in the polar region with the other poleward temperature-enhanced region around 120°E on day −1. Compared with the 2018 SSW, the temperature evolution in the 2019 SSW ( Figure 1b) revealed large differences in longitudinal distribution. The warming process appeared only in the eastern hemisphere and developed northeastward from the mid-latitudes around 60°E to the high latitudes around 120°E. However, the temperature-enhanced route in the western hemisphere was absent in the prewarming period of the 2019 SSW. In addition, the warming air in the polar region occurred much earlier in the 2019 SSW (around day −5), and the temperature in the polar region seemed to decrease from day −5 to day −1. In fact, the zonal mean temperature difference between 90°N and 60°N reached the maximum on day −5 (28 December 2018), which can be considered the onset date of the 2019 SSW (Ma Z et al., 2020). The occurrence and duration of the poleward temperature-enhanced process in the prewarming peri-od were not specifically related to the time of the central date but may have been associated with the onset dates of major SSWs. Figure 1 indicates that the poleward temperature-enhanced process may have occurred earlier (before the onset date) than the wind reversal at 60°N (central date). Our results also revealed that the temperature-enhanced regions may have had different longitudes during the different SSW events.
The evolution of geopotential heights in the prewarming periods of the 2018 and 2019 SSWs are presented in Figure 2. Following previous studies (e.g., Rao J et al., , 2019bLee and Butler, 2020;Lü ZZ et al., 2020), the region with a high-value geopotential height suggests an anticyclone, whereas the area with a lowvalue geopotential height indicates a cyclone (polar vortex). As shown in Figure 2a, the polar vortices did not have significant displacement features from day −15 to day −7. The anticyclone  around the 180° meridian, known as the Aleutian High (Harvey and Hitchman, 1996), was located stably over the northern Pacific Ocean. From day −5, the anticyclone over the Atlantic Ocean became stronger and moved poleward together with the Aleutian High. These two poleward anticyclones began to push the polar vortex after day −3 and split it into two daughter vortices on day −1. Thus, the 2018 SSW was classified as a split type in previous studies on the basis of these vortex distributions (e.g., Rao J et al., , 2019aHarada et al., 2019;Lee et al., 2019). Unlike the longlasting split vortices that occurred before the onset of the wellknown 2009 SSW event (e.g., Ayarzagüena et al., 2011), the vortex in the prewarming period of the 2018 SSW split suddenly because of the sharply enhanced anticyclone over the Atlantic region on day −3. The abruptly enhanced anticyclone may also have made the 2018 SSW event less likely to be forecast well by models (e.g., Lee et al., 2019). The evolution of geopotential heights in the prewarming period of the 2019 SSW revealed only a displaced polar vortex with the Aleutian High, as shown in Figure   2b, which is similar to the results shown by Rao J et al. (2019b). The polar vortex first tilted eastward to the Eurasian region and then extended westward to the Atlantic region after day −11. This process may be related to the attenuation of the high-latitude zonal winds at the beginning of the SSW event. It is interesting that the polar vortex weakened and seemed to become elongated from Europe to North America after day −7. On day −5 and day −3, the polar vortex was centrally located over the Atlantic region from 0° westward to 60°W. Nevertheless, the evolution of the polar vortex did not show any splitting distribution in the prewarming period of the 2019 SSW. Therefore, the 2019 SSW exhibited a displacement type based on the evolution of geopotential heights in the prewarming period (e.g., Rao J et al., 2019b).
As shown in Figures 1 and 2  inal dependencies may have been related to the evolution of anticyclones in the prewarming periods. For instance, the distribution of the air temperature was zonally symmetrical, based on the split vortices in the 2018 SSW, which may have been caused by the two-peak anticyclonic structure. The Aleutian High and the Euro-Atlantic anticyclone could have forced the northward winds on the western side and promoted the poleward propagation of warming air (Kozubek et al., 2015). In general, the Aleutian High persists in the stratosphere during most Arctic winters (e.g., Lastovicka et al., 2018). The Aleutian High continuously transfers warming air to high latitudes during winters, for example, from day −15 to day −7 in the 2018 SSW. Thus, the poleward temperature-enhanced phenomenon can commonly be seen around 120°E, especially before major SSW events. However, the anticyclone sometimes arises unexpectedly in a short time over the Euro-Atlantic region, leading to a poleward temperature enhancement around 60°W. The evolution in the prewarming period of the 2018 SSW further indicates that the poleward temperature-enhanced processes occurring around both 120°E and 60°W finally connect over the polar region. This evolution is conducive to classifying the 2018 SSW as a split-type SSW event in the prewarming periods (e.g., Rao J et al., , 2020. The frequency with which an anticyclone occurs over the Euro-Atlantic region is not high during Arctic winters (e.g., Lastovicka et al., 2018), which is also consistent with the lower frequency of split-type events in the prewarming periods (e.g., Choi et al., 2019). Comparisons of the 2018 and 2019 SSWs in the prewarming periods revealed that the longitudinal dependencies of the poleward temperature-enhanced regions were closely accompanied by anticyclones and polar vortices.   This secondary enhancement had a higher maximum value (more than 260 K) than in the prewarming period. The other enhancement was a gradual weakening over the European region from day 1 to day 9 and its disappearance after day 11. In the postwarming period of the 2019 SSW, a secondary enhancement of air temperature was also observed at high latitudes while the SSW was located over the Eurasian region. Note that we observed a secondary enhancement at high latitudes in both postwarming periods of the two major SSWs. These secondary enhancements extended the warming time in the stratospheric polar region after the onset dates of the major SSWs. As shown in Figure 3a, the warming air over the Eurasian region moved toward the North American region via the polar region from day 1 to day 5, which may have been responsible for the secondary en-hancement that was concentrated over the North American sector in the postwarming period of the 2018 SSW. However, the secondary enhancement in the postwarming period of the 2019 SSW was likely due to the warming air shifting westward from the Pacific Ocean to the Eurasian region.

Postwarming Comparisons
In addition to propagating warming air in the stratosphere, the polar vortices shown in Figure 4 also indicate significant transitions in the postwarming periods in both the 2018 and 2019 SSWs. The polar vortices were not symmetrical, having a 180° longitudinal difference over North America and Asia after the central date of the 2018 SSW. The Asian vortex became weaker and propagated westward to the Atlantic region from day 1 to day 3. From day 5 to day 7, the weakening vortex combined with the North American vortex, which ended the previous splitting distributions. The recombined vortex over the North American region  and the major anticyclone over the Eurasian region then propagated westward together from day 9 to day 15, and it developed a displaced vortex in the postwarming period of the 2018 SSW. This was the first time that the recombination process in the postwarming period of a split-type SSW event had been observed, and it was obviously different from the multiple vortices that formed after the 2009 split-type SSW (e.g., Manney et al., 2009). Figure  In the postwarming period of the 2019 SSW, we also observed a transition of the polar vortex. As shown in Figure 4b, the polar vortex began to split into two daughter vortices after the central date and were separated on both the western and eastern sides of the Atlantic region. Rao J et al. (2019bRao J et al. ( , 2020 suggested that the difference in longitude between the two vortices was less than 120°, which was due to the enhancement of wave 3. Note that only one anticyclone (the Aleutian High) occurred in the postwarming period of the 2019 SSW. However, the polar vortex still split without the anticyclone over the Euro-Atlantic region, as shown in Figure 2a. The splitting vortices gradually controlled the North American region and the Eurasian region from day 9 to day 15, al-  SSWs are worth noting because the evolution of the polar vortices during SSWs can further influence the surface climate (e.g., Mitchell et al., 2013;Liu Y and Zhang YL, 2014;Nath et al., 2016;Rao J et al., 2020). However, what causes these transitions in the polar vortices or what the similarity is between these transition SSWs is not known.
In general, planetary wave activities are considered a significant driver of the atmospheric dynamics during SSWs (e.g., Azeem et al., 2005;Goncharenko et al., 2012;Shi CH et al., 2017;Cao C et al., 2019). The vortex types during SSWs are believed to be connected to the dominant planetary waves in the prewarming periods; that is, wave 1 plays a primary role in displacement events, whereas wave 2 dominates in split events (e.g., Harada and Hirooka, 2017;Liu SM et al., 2019). To further understand the wave evolution during the 2018 and 2019 SSWs, the day-to-day amplitudes of the SPWs with wavenumber 1 (SPW1), wavenumber 2 (SPW2), and wavenumber 3 (SPW3) were calculated at pressure levels of 10 hPa based on the geopotential height at 60°N. Figure 6 illustrates the evolution of the SPW1, SPW2, and SPW3 from day −15 to day 15. In the prewarming periods of the 2018 SSW, the amplitudes of wave 1 decreased by about 1 km and the amplitudes of wave 2 increased during the descending periods of wave 1. Wave 2 was larger than wave 1 from day −5 to day −1 in the 2018 SSW, which dominated closer to the central date and was expected to be consistent with the typical split-type vortices. Wave 1 decreased by about 0.6 km in the prewarming period of the 2019 SSW and dominated from day −10 to day 0 with a relatively stable amplitude of 0.9 km. Wave 2 had a peak amplitude of only about 0.4 km around day −10 and decreased to nearly zero on day 0. Dominant wave 1 and decreased wave 2 were related to the displacement vortex in the prewarming periods of the 2019 SSW, as shown in Figure 2b. Thus, the status of the vortices could be explained by the evolution of the dominant waves in the prewarming periods in both SSW events. In the postwarming period of the 2018 SSW, wave 2 decreased sharply, whereas wave 1 again dominated from day 0. The recombination of the vortices in the 2018 SSW was very likely due to this transition of the wave amplitudes. According to Choi et al. (2019), the 2018 SSW can be classified as a split-displacement-type SSW, similar to the 2007 SSW, because wave 1 dominated wave 2 during the entire postwarming period after wave 2 dominated during the prewarming period. However, the splitting vortices in the 2019 SSW were not consistent with the most prominent SPWs. Rao J et al. (2019b) attributed the splitting vortices in the 2019 SSW to enhancements of the tropospheric wave 3. The SPWs also indicated an enhancement of the stratospheric wave 3 around day 2, and three vortices were observed on day 5, as shown in Figure 4b. Nevertheless, wave 2 had the least influence on the splitting vortices around the central date of the 2019 SSW. As shown in Figure 6 (right), the amplitudes of wave 2 were smaller than 0.2 km until day 8, although the dominant wave 1 kept weakening during the postwarming period. Lee and Butler (2020) also proposed that the 2019 SSW had a splitting vortex without the wave 2 amplification.
According to the postwarming vortices shown in Figure 4, the recombination and splitting vortices were all observed over the Atlantic region (0°-60°W). Choi et al. (2020)  Pawson, 2015; Ma Z et al., 2017;Gong Y et al., 2018bGong Y et al., , 2019 was a displacement-split-type event with the splitting vortices over the Atlantic region just after the central date, as shown in Figure 5b, which is very similar to the 2019 SSW. Comparison of the NAO indices during these four major SSWs revealed a weak correlation between the negative NAO phases and the transitions of the vortices. For instance, the 2013 SSW had a positive-phase NAO index from day −12 to day 4, whereas a negative-phase NAO index appeared around the central date of the 2019 SSW. The NAO indices were both in a positive phase before day 5 in the 2018 SSW and the 2009 SSW, and splitting vortices were observed in both events. Thus, the negative phase of the NAO index was not responsible for the splitting vortices in the 2018, 2009, and 2013 SSWs, which indicates that the transitions of the polar vortices in the postwarming period may be further related to other factors over the Atlantic sector.
In addition to the enhanced planetary waves and the negative phase of the NAO, the transitions of the polar vortices seem to be influenced mainly by the development of anticyclones over the polar region. In the postwarming period of the 2018 SSW, the recombination vortices may have been due to a weakening of the Euro-Atlantic peak, which moved into the Pacific region through the polar region, as shown in Figure 4a, leaving uncontrolled areas over the Atlantic region. This evolution may have allowed the weak vortex around 60°E to recombine with the strong western vortex under the westward wind, forcing in the postwarming period of this major event. The splitting vortices in the postwarming period of the 2019 SSW may have been related to the highvalue geopotential height moving in the opposite direction, which was developing from the Aleutian High toward the Atlantic region via the polar region. This process may have cut the polar vortex into two daughter vortices. Harvey et al. (2002) reported a splitting vortex in 1999 that was led by a merger of stratospheric anticyclones, indicating that variations in the anticyclones could influence the distribution of the polar vortices. Coy and Pawson (2015) found splitting vortices similar to the 2019 SSW over the Atlantic region after the central date of the 2013 SSW. In their discussion, they proposed that the major wave forcing was found only in the Pacific sector and was not obvious in the Atlantic sector in the postwarming period of the 2013 SSW. As shown in Figure 5b, the anticyclone revealed a clear propagation in the postwarming period of the 2013 SSW, which moved from the Pacific sector to the Atlantic sector and segregated the splitting vor-tices into the North American region and the Eurasian region. The single direction of the anticyclone in the 2019 SSW also developed from the Pacific sector toward the Atlantic sector. It remains to be studied whether the Pacific wave forcing is responsible for the shifts of the anticyclone during all the displacement-split events. Nevertheless, the transition vortices observed in the 2013, 2018, and 2019 SSWs may all have been related to the evolution of anticyclones between the Pacific region and the Atlantic region, crossing the polar region. It is still unclear what is controlling this evolution. Statistical analysis is needed to further study the triggering mechanism of the SSWs with transition vortices.

Conclusions
Stratospheric evolution of the temperature and geopotential height during two major SSW events that occurred in February 2018 and January 2019 were investigated based on MERRA-2 reanalysis data. In the prewarming period, poleward enhancement of the stratospheric temperature from the mid-latitudes to the polar region had a clear longitudinal dependence. The temperature enhancement occurred around 120°E before the 2019 SSW with a displacement vortex, whereas it appeared at both 60°W and 120°E before the 2018 SSW with a splitting vortex. In the postwarming period, an unexpected recombination of the polar vortex was observed in the 2018 SSW, whereas splitting vortices were found in the 2019 SSW. This was the first time a recombination vortex had been reported in the postwarming period of a split-type SSW. A secondary enhancement of wave 1 and a sudden decrease of wave 2 may have been responsible for this recombination. However, the evolution of wave 1 and wave 2 cannot fully explain the splitting vortices in the postwarming period of the 2019 SSW. The enhancement of tropospheric wave 3 may have been responsible for the splitting vortices in the 2019 SSW (e.g., Rao J et al., 2019b). It is interesting that we observed all the transitions of the polar vortices over the Atlantic region, whereas negative phases of the NAO indices, as suggested by Choi et al. (2020), may not always be suitable to explain the splitting vortices. The stratospheric evolution of the geopotential heights indicates that the recombination vortex in the 2018 SSW may have been due to the Euro-Atlantic anticyclone moving into the Pacific region through the polar region, whereas the splitting vortices in the 2019 SSW may have been related to the anticyclones moving from the Aleutian High toward the Atlantic region via the polar region. Our results suggest that the Atlantic region may be a crucial region during the SSWs with transition vortices. Hence, future studies are needed to further investigate the mechanisms of the transition of polar vortices in the Atlantic region.

Acknowledgments
The MERRA2 data can be obtained from NASA (https://disc.gsfc. nasa.gov/datasets?page=1&project=MERRA-2). This study was supported by the National Natural Science Foundation of China (grants 41574142 and 41531070), the Specialized Research Fund for State Key Laboratories, and the National Science Foundation (grant AGS-1744033).