1
On he ep esen a ion o majo s a osphe ic wa mings in eanalyses
Blanca Aya zagüena1,2, F oila M. Palmei o1a, Da id Ba ioped o2, Na alia Cal o1, Ul ike Langema z3 and
Kiyo aka Shiba a4
1Dp o. Física de la Tie a y As o ísica, Uni e sidad Complu ense de Mad id, Mad id, 28040, Spain
2Ins i u o de Geociencias (IGEO), Consejo Supe io de In es igaciones Cien í icas - Uni e sidad Complu ense de Mad id,
5
Mad id, 28040, Spain.
3 Ins i u ü Me eo ologie, F eie Uni e si ä Be lin, Be lin, 12165, Ge many.
4 School o En i onmen al Science and Enginee ing, Kochi Uni e si y o Technology, Kami, 7828502, Japan.
a now a : Ba celona Supe compu ing Cen e (BSC-CNS)
10
Co espondence o: Blanca Aya zagüena (baya [email protected])
Abs ac . Majo sudden s a osphe ic wa mings (SSWs) ep esen one o he mos ab up phenomena o he bo eal win e ime
s a osphe ic a iabili y, and cons i u e he clea es example o coupling be ween he s a osphe e and he oposphe e. A good
ep esen a ion o SSWs in clima e models is equi ed o educe hei biases and unce ain ies in u u e p ojec ions o
15
s a osphe ic a iabili y. The abili y o models o ep oduce hese phenomena is usually assessed wi h jus one eanalysis.
Howe e , he numbe o eanalyses has inc eased in he las decade and hei own biases may a ec he model e alua ion.
He e we compa e he ep esen a ion o he main aspec s o SSWs ac oss eanalyses. The examina ion o hei main
cha ac e is ics in he p e- and pos -sa elli e pe iods e eals ha eanalyses beha e e y simila ly in bo h pe iods. Howe e ,
disc epancies a e la ge in he p e-sa elli e pe iod han a e wa ds, pa icula ly o he NCEP/NCAR eanalysis. All da ase s
20
ep oduce simila ly he speci ic ea u es o wa enumbe -1 and wa enumbe -2 SSWs. A good ag eemen among eanalyses is
also ound o igge ing mechanisms, oposphe ic p ecu so s and su ace esponse. In pa icula , di e ences in blocking
p ecu so ac i i y o SSWs ac oss eanalyses a e much smalle han be ween blocking de ini ions.
1 In oduc ion
Majo sudden s a osphe ic wa mings (SSWs) cons i u e he mos impo an phenomena o he No he n Hemisphe e pola
25
s a osphe ic a iabili y in win e ime. They a e ab up wa mings o he pola s a osphe e ha lead o a decele a ion o he
pola o ex and a e e sal o he ypical wes e ly ci cula ion (And ews e al., 1987). SSWs can be classi ied in o wo di e en
ypes acco ding o he s uc u e o he pola o ex du ing he e en . Acco dingly, he pola o ex is ei he displaced om he
pola cap ( o ex displacemen , D SSWs) o spli in o wo pa s o simila size ( o ex spli , S SSWs) (Cha l on and Pol ani,
2007).
30
SSWs ep esen a clea example o s a osphe e- oposphe e coupling in bo h di ec ions. Fi s , hey a e usually p eceded by an
enhancemen o wa e ac i i y (e.g. Ma suno, 1971). Al hough his enhancemen can ake place in he lowe oposphe e, ecen
2
s udies ha e shown ha i o en happens wi hin he s a osphe e o opopause egion and depends on he s a osphe ic mean
low condi ions (Sjobe g and Bi ne , 2014; Bi ne and Albe s, 2017; de la Cáma a e al., 2017; Whi e e al., 2019). The sou ces
o upwa d-p opaga ing wa e ac i i y a e mainly loca ed in he mid- o-uppe oposphe e and co espond o anomalous
35
ci cula ion e en s such as a deepened Aleu ian low (e.g.: Ga inkel e al., 2010) o blocking highs, among o he s (e.g. Ma ius
e al., 2009; Nishii e al., 2011; Aya zagüena e al., 2011; Ba ioped o and Cal o, 2014). Based on he wa e ac i i y p eceding
SSWs, hey a e commonly classi ied in o wa enumbe 1 (WN1) o wa enumbe 2 (WN2) e en s (e.g. Bancalà e al., 2012;
Ba ioped o and Cal o, 2014). This classi ica ion p oduces subse s o e en s simila o he D/S ca alogue. Howe e , he e a e
di e ences since he o me is based on he p ecu so y wa e ac i i y while he D/S classi ica ion accoun s o he shape o he
40
pola o ex du ing he pos -wa ming phase (Bancalà e al., 2012). Depending on he ype o SSWs, he oposphe ic p ecu so s
a e di e en and/o loca ed in di e en geog aphical loca ions (Ma ius e al., 2009; Cohen and Jones, 2011; Bancalà e al.,
2012). In pa icula , di e ences in blocking p ecu so s a e la ge when SSWs a e classi ied in o WN1/WN2 han D/S
(Ba ioped o and Cal o 2014).
In e ms o downwa d coupling, he SSWs signal p opaga es downwa d and eaches he oposphe e as e ealed om
45
composi e analyses (Baldwin and Dunke on, 2001), al hough he e is s ill unce ain y abou his oposphe ic esponse when
analyzing indi idual e en s (e.g.: Ge be e al. 2009). One o he sugges ed ac o s ha may con ibu e o he sp ead o he
su ace signa u e o SSWs is he ype o e en . Howe e , while some s udies ha e shown ha only S SSWs ha e la ge e ec s
on su ace clima e (Mi chell e al., 2012, Se iou e al., 2013), o he s ha e no ound consis en di e ences be ween S and D
SSWs in i s signi ican su ace impac (Cha l on and Pol ani, 2007; Cohen and Jones, 2011). Thus, he e is no ye a consensus
50
in his ega d, p obably due o he di e ences in he algo i hms used o iden i y S and D SSWs (Maycock and Hi chcock,
2015). As o WN1 and WN2 SSW, hei su ace signa u e has no ye been explo ed.
SSWs a e a key elemen when analyzing s a osphe ic a iabili y. The equency and seasonali y o SSWs a e common me ics
o assess he e ec s o oposphe ic and oceanic phenomena on he pola nigh je (PNJ). These me ics a e also used o e alua e
he s a osphe ic esponse o clima e change (e.g.: Taguchi and Ha mann, 2006; Cha l on-Pe ez e al., 2008; Aya zagüena e
55
al., 2018). Indeed, in modeling s udies mos o hem use simula ions ha a e p e iously alida ed by compa ing hei esul s
wi h eanalysis da ase s (e.g.: Cha l on e al., 2007; McLand ess and Shephe d, 2009; Kim e al., 2017). Howe e , he numbe
o eanalyses has inc eased in he las decade, and al hough he obse a ional da a used in he assimila ion p ocess is he same,
he eanalysis models a e di e en , and so may he inal p oduc s be (Fujiwa a e al., 2017). As i happens wi h o he
a mosphe ic models, eanalyses also ha e biases and his can a ec he model e alua ion (Fujiwa a e al., 2017).
60
Due o quali y imp o emen s associa ed wi h he assimila ion o sa elli e da a, mode n eanalyses, such as ERA-In e im,
NASA-MERRA, and NCEP-CFSR, only co e he pos -sa elli e pe iod since 1979. This means ha he numbe o a ailable
eanalyses o assess he model pe o mance in he p e-sa elli e e a is smalle han in he pos -sa elli e pe iod. In addi ion, he
amoun o da a o assimila e is also limi ed in he o me pe iod. All his migh p oduce a i icial di e ences in esul s be o e
and a e he inclusion o sa elli e da a. Gómez-Escola e al. (2012) documen ed a change o some SSW ea u es om he p e-
65
sa elli e o he pos -sa elli e e a in NCEP-NCAR and ERA-40 eanalyses. Fo ins ance, he in a-seasonal dis ibu ion and he
3
ampli ude o he SSW-associa ed wa ming showed di e ences be ween bo h pe iods, po en ially due o a change in he ype
o he assimila ed da a. Wi h he a ailabili y o he new JRA-55 eanalysis, which is he only one ha applies an ad anced da a
assimila ion scheme o uppe -ai da a du ing he p e-sa elli e e a, e isi ing his opic seems app op ia e.
In his s udy, we aim o assess he pe o mance o he mos widely used eanalyses in ep esen ing SSWs. To do so, i s , he
70
main cha ac e is ics o SSWs a e examined o all da ase s o quan i y he deg ee o ag eemen ac oss eanalyses. Bo h p e-
and pos -sa elli e pe iods a e compa ed o in es iga e whe he disc epancies among eanalyses in he ep esen a ion o he
main SSW cha ac e is ics depend on he examined pe iod. Secondly, we add ess he dynamical o cing o SSWs in all da ase s,
including p ecu so s such as blockings. Finally, he su ace impac o SSWs e ie ed om he di e en eanalyses is analyzed.
Special emphasis is gi en o he assessmen and obus ness o he po en ial di e ences in he o cing and su ace impac o
75
WN1 and WN2 SSWs, as well as S and D e en s.
Ou wo k is a con ibu ion o he Chap e 6 o he S a osphe e- oposphe e P ocesses And hei Role in Clima e (SPARC)
Reanalysis In e compa ison P ojec (S-RIP) ini ia i e, which aims o assess s a osphe e- oposphe e coupling in eanalyses.
In he amewo k o his ini ia i e, a ew ecen s udies ha e add essed some aspec s o he ep esen a ion o pola s a osphe ic
a iabili y in eanalyses. In pa icula , Ma ineau e al. (2018) and Hi chcock (2019) also in es iga e SSWs- ela ed aspec s.
80
The o me analyzes he momen um budge du ing SSWs es ic ed o he pos -sa elli e pe iod, while Hi chcock (2019)
compa es he ep esen a ion o s a osphe e- oposphe e coupling in bo h p e and pos -sa elli e pe iod, wi h he emphasis on
he impac o including p e-1979 da a. Di e en om hese s udies, ou wo k p o ides a comp ehensi e in e - eanalyses
compa ison o he mos impo an and ypical aspec s and p ocesses associa ed wi h SSWs in bo h p e- and pos -sa elli e e as.
Addi ionally, we explo e u he he cha ac e is ics o WN1 and WN2 SSWs ha ha e no ye been in es iga ed.
85
The s uc u e o he pape is as ollows. The da a used and me hodology applied a e desc ibed in Sec ion 2. Sec ion 3 compa es
he pe o mance o he main cha ac e is ics o SSWs ac oss eanalyses. Sec ion 4 ocuses on he dynamical o cing o he
e en s and Sec ion 5 add esses he pe o mance o eanalyses in ep esen ing he su ace impac o SSWs. The main
conclusions a e summa ized in Sec ion 6.
2 Da a and me hodology
90
2.1 Da a
We ha e used daily da a om he ollowing eanalyses: ERA-40 (Uppala e al., 2005), ERA-In e im (Dee e al., 2011), JRA-
25 (Onogi e al., 2007), JRA-55 (Kobayashi e al., 2015), NASA-MERRA (Rienecke e al., 2011), NCEP-CFSR (Saha e al.,
2010), NCEP-DOE (Kanami su e al., 2002), and NCEP-NCAR eanalysis (Kalnay e al., 1996). Mo e de ails abou he
di e en eanalyses can be ound in Fujiwa a e al. (2017). Fo he compa ison ac oss di e en eanalyses, all da a was used
95
a he common egula g id o 2.5° lon x 2.5° la . When no di ec ly a ailable om he eanalysis cen e s, a i s o de
conse a i e emapping was applied.
4
The me hodology o he in e compa ison ollows he S-RIP speci ica ions. As such, he analysis has been ca ied ou o wo
di e en pe iods: his o ical (1958-1978) and compa ison (1979-2012). Gi en he pe iods co e ed by each eanalysis, only
ERA-40, NCEP-NCAR, and JRA-55 a e employed in he his o ical pe iod. In con as , all he abo e lis ed eanalyses a e used
100
in he compa ison pe iod wi h he excep ion o ERA-40, because i ends in 2002. The pe o mance o each eanalysis is
e alua ed agains a mul i- eanalysis mean (MRM), he ein conside ed as an “unbiased” e e ence. In he his o ical pe iod, he
MRM e e s o he a e age o he h ee eanalyses ha co e ha pe iod, while in he compa ison pe iod, he MRM is de ined
as he a e age o he mos ecen eanalyses o each cen e (ERA-In e im, NCEP-CFSR, JRA-55 and NASA-MERRA).
He ea e , anomalies o each eanalysis a e de ined as he depa u e o he ield om he daily clima ology o each eanalysis.
105
In he his o ical pe iod, he clima ology co e s he whole pe iod (i.e. 1958-1978), whe eas he compa ison pe iod uses he
1981-2010 baseline. Unless o he wise s a ed, s a is ical signi icance o he esul s is compu ed wi h a Mon e-Ca lo es o 1000
pe mu a ions, each one con aining he same numbe o cases and da es as he SSWs o each composi e bu wi h andom yea s
o occu ence.
2.2 C i e ia o he iden i ica ion o SSWs
110
We ha e used he lis o SSWs and common da es iden i ied in Bu le e al. (2017) and p o ided o he S-RIP ini ia i e
(Chap e 6), unless o he wise indica ed. Fi s , o each eanalysis, SSWs a e iden i ied based on he e e sal o he zonal mean
zonal wind a 60ºN and 10hPa be ween No embe and Ma ch, wi h a leas 20 days o sepa a ion be ween e en s. S a osphe ic
inal wa mings a e excluded by equi ing a leas 10 consecu i e days o wes e ly winds be o e he end o Ap il (Cha l on and
Pol ani, 2007). The i s day o e e sal o winds de e mines he da e o occu ence o he SSW ( he so-called cen al da e).
115
Common SSWs a e hose iden i ied by a leas wo o he h ee eanalyses in he his o ical pe iod and by a leas ou ou o
se en eanalyses in he compa ison pe iod a ound he same da e (usually wi hin one o wo days). The cen al da e o hese
common e en s is compu ed as he median o he cen al da es om he SSWs de ec ed o each eanalysis. Thus, wi h his
app oach, he same e en s and cen al da es apply o all eanalyses e en i he e e sal o he winds does no occu in all o
hem. This is use ul o ensu e ha he di e ences be ween da ase s a e no due o he selec ion o di e en e en s o da es. The
120
common SSWs a e lis ed in Table 1 o he compa ison pe iod.
Ne e heless, in he e y i s pa o ou s udy, we ha e add essed he opposi e ques ion and quan i ied he possible
disc epancies in he equency o SSWs among eanalyses when he same c i e ion is applied o all da ase s. In ha case, we
ha e imposed he WMO de ini ion o he iden i ica ion o SSWs in each eanalysis. The de ini ion is based on he e e sal,
wi hin 5 days) o zonal-mean zonal wind a 10hPa and 60ºN and zonal-mean empe a u e di e ence be ween 90ºN and 60ºN
125
a he same le el (Labi zke, 1981).
2.3 Types o SSWs
SSWs a e classi ied ollowing wo de ini ions: D s S SSWs, and WN1 s WN2 e en s. In his s udy, D and S SSWs we e
iden i ied acco ding o he algo i hm by K. Shiba a, which is simila o ha o Cha l on and Pol ani (2007). I is based on he
5
iden i ica ion o cyclonic o ices and hei ela i e sizes by means o he non-zonal absolu e o ici y a 10hPa om 5 days
130
be o e o 10 days a e (i.e. [-5,10]-day) wi h espec o he occu ence o an SSW, acco ding o he de ini ion o Sec ion 2.2.
Mo e speci ically, S SSWs a e iden i ied when wo local maxima o he absolu e o ici y a e loca ed diame ically opposed
and he size a io o he sec o s a ound hose maxima is la ge han 0.5 du ing a leas one o he 16-day pe iod su ounding
he SSW. O he wise he SSW is de ined as D. The e en s we e classi ied indi idually in each eanalysis. The classi ica ion
in o S/D e en s o common SSWs in he compa ison pe iod (used in Sec ions 4 and 5) was based on he p edominan ype o
135
each single e en ac oss he di e en eanalyses, ollowing a simila p ocedu e o ha employed o he iden i ica ion o he
common da es (Table 1).
WN1 and WN2 SSWs we e selec ed by applying a zonal Fou ie decomposi ion o he daily 50hPa geopo en ial heigh da a a
60ºN in o WN1 (Z1) and WN2 (Z2) ampli udes o he [-10,0]-day pe iod be o e each SSW (Ba ioped o and Cal o, 2014). An
SSW was de ined as a WN2 e en i [Z2] ≥ [Z1] (b acke s deno e he a e aged ampli ude o he [-10,0]-day pe iod be o e he
140
SSW) o i Z2 - Z1 ≥ 200 m a leas o one day wi hin he [-10,0]-day pe iod be o e he SSW. O he wise, he SSW was de ined
as a WN1 e en . See he lis o e en s o each ype in Table 1 and Ba ioped o and Cal o (2014) o mo e de ails on he
algo i hm.
2.4 Dynamical benchma ks
We ha e applied he ollowing diagnos ics p oposed by Cha l on and Pol ani (2007) o e alua e he dynamical signa u es
145
associa ed wi h he occu ence and de elopmen o SSWs:
- Ampli ude o he SSW in he middle s a osphe e (he ea e amp010) compu ed as he a ea-weigh ed mean 10hPa empe a u e
anomaly o e he pola cap (50°N-90°N) and a e aged o he [-5,5]-day pe iod wi h espec o he cen al da e o he e en .
- Ampli ude o he SSW in he lowe s a osphe e (he ea e amp100), de ined as amp010 bu a 100hPa. I p o ides a measu e
o he coupling be ween he middle and lowe s a osphe e a ound he occu ence o SSWs.
150
- Decele a ion o he PNJ (he ea e decelu), co esponding o he di e ence o he 10hPa zonal-mean zonal wind a 60°N
be ween he [-15, -5]-day pe iod p io o he cen al da e and he [0, 5]-day pe iod a e he cen al da e.
- Wa e ac i i y p io o SSW (he ea e ac wa ), compu ed as he a ea-weigh ed mean 100hPa me idional eddy hea lux (HF)
anomaly a e aged o e 45°N-75°N o he [-20,0]-day pe iod be o e he occu ence o he e en .
2.5 Upwa d-p opaga ing wa e ac i i y
155
The anomalous me idional eddy HF a e aged o e 45°N-75°N a di e en p essu e le els was used as a me ic o measu e he
upwa d p opaga ion o wa e ac i i y. This la i udinal band co esponds o he clima ological a ea wi h he s onges e ical
wa e p opaga ion om he oposphe e o he s a osphe e (Hu and Tung, 2003).
As a second s ep, he me hodology by Nishii e al. (2009) was applied o analyze he ole o di e en o cing p ocesses in he
occu ence o SSWs. This me hodology is based on he decomposi ion o daily anomalous eddy HF in o wo componen s,
160

6
which co espond o he in e ac ion be ween clima ological wa es and anomalous wa es (second and hi d igh hand e ms
o Eq. 1) and he inhe en con ibu ion o anomalous wa es ( i s igh hand e m o Eq. 1):
[𝑣∗𝑇∗]𝑎=[𝑣𝑎
∗𝑇𝑎
∗]𝑎+[𝑣𝑐
∗𝑇𝑎
∗]+[𝑣𝑎
∗𝑇𝑐
∗] (1)
whe e b acke s and as e isks indica e zonal mean and de ia ion om i , espec i ely, is me idional wind, T is empe a u e
165
and he a and c subsc ip s deno e daily anomalies and clima ological alues, espec i ely. Eq. 1 has been applied o each
p essu e le el.
2.6 Blocking de ini ions
The p ecu so ole o blocking in SSWs has been discussed ac oss s udies (see e.g., Cas anhei a and Ba ioped o (2010) o
an o e iew), al hough he e is no a clea consensus on his opic. The di e gen esul s o p e ious s udies may pa ially be
170
a ibu ed o di e en me hodologies o blocking de ec ion (e.g., Woollings e al., 2008). In his s udy, h ee di e en blocking
de ini ions ha e been used o add ess his ques ion. The h ee me hodologies use daily geopo en ial heigh a 500 hPa (Z500)
and span almos all app oaches o blocking de ini ion. The i s me hod is based on he occu ence o egional and pe sis en
me idional Z500 g adien e e sals ( he absolu e me hod, ABS; e.g., Sche e e al., 2006). The second me ic in ol es he
de ec ion o pe sis en and quasi-s a iona y Z500 anomalies, compu ed wi h espec o he local clima ological ield ( he
175
anomaly me hod, ANO; e.g., Sausen e al., 1995). Finally, a combined me hod o absolu e and anomaly Z500 ields ( he mixed
me hod, MIX) is used, p o iding a wo- old pe spec i e o blocking (Ba ioped o e al., 2010). Se e al c i e ia a e imposed o
ensu e ha he de ec ed episodes ep esen la ge-scale, quasi-s a iona y, and pe sis en high-p essu e sys ems. See Woollings
e al. (2018) o mo e de ails abou blocking de ini ions.
3 Main SSW cha ac e is ics
180
In his sec ion, he main signa u es o SSWs ( equency, ype o e en s and p ocess-based diagnos ics) a e analyzed o each
pe iod and compa ed among he di e en da ase s.
3.1 F equency, seasonali y and ype o e en s
Fi s , we ha e analyzed he esul s o he equency and ype o e en s when he same c i e ion is applied o each da ase .
Table 2 shows he mean equency o e en s and he a io o D o S SSWs o each pe iod and eanalysis. The main di e ences
185
a e ound in he his o ical pe iod when he eanalyses show a la ge sp ead in bo h equency and ype o e en s. In pa icula ,
he NCEP/NCAR eanalysis displays he esul s ha de ia e he mos om he o he wo da ase s, al hough he di e ences a e
no s a is ically signi ican a he 95% con idence le el (S uden ’s - es ). The sho pe iod o analysis and hence he educed
sample migh explain pa o hese disc epancies. Mo e impo an ly he una ailabili y o sa elli e da a in he p e-sa elli e e a
7
leads o a s ong dependency o he eanalysis da a in he s a osphe e on he cha ac e is ics o each eanalysis model. No e
190
ha NCEP/NCAR eanalysis is he only eanalysis wi h a low- op model and a lid in he s a osphe e (3hPa), whe eas JRA-55
and ERA-40 ha e he op in he mesosphe e (0.1hPa). The low op ypically dampens a iabili y close o he op and so, educes
he p obabili y o he occu ence o an SSW (Cha l on-Pé ez e al., 2013). In ac , he s anda d de ia ion o daily pola
empe a u e and zonal wind a 10 hPa in Decembe and Janua y o he his o ical pe iod is much lowe in NCEP/NCAR han
in he o he wo eanalyses, al hough he di e ences a e no s a is ically signi ican a he 95% con idence le el (F-Fishe es )
195
(Figu e 1a, c). In con as , a lowe le els, we do no ind such disc epancies (see 100 hPa empe a u e in Fig. 1b, d), suppo ing
ha he occu ence o SSWs du ing his pe iod is s ongly in luenced by he model pe o mance and hence should be
conside ed eanalysis-dependen .
Con e sely, in he compa ison pe iod, he e is a good ag eemen in bo h he equency and a io o D/S SSWs. Small
di e ences a e ound, pa icula ly, in he D/S a io, bu his migh be due o he speci ic h esholds o o he me hodological
200
issues o he applied c i e ion, since such de ia ion does no appea when classi ying SSWs in o WN1 and WN2 e en s
(Ba ioped o and Cal o, 2014). Mo e de ails abou hese classi ica ions o SSWs can be ound in he Chap e 6 o S-RIP.
Rega ding SSWs seasonali y, Figu e 2 shows he smoo hed seasonal dis ibu ion o SSW pe decade. This dis ibu ion has
been compu ed by coun ing he numbe o SSWs wi hin he ±10-day pe iods cen e ed on each win e days. Addi ionally, he
dis ibu ion has been smoo hed wi h a 10-day unning mean. Simila o he win e mean equency o SSWs, his o ical
205
eanalyses show he la ges sp ead in he seasonal dis ibu ion. A subs an ial pa o his sp ead is due o he NCEP/NCAR
eanalysis whose dis ibu ion is s a is ically signi ican ly di e en om ha o he o he wo eanalyses a a 99% con idence
le el ( wo-samples Kolmogo o -Smi no es ). In con as , ERA- 40 and JRA-55 dis ibu ions display simila (s a is ically
undis inguishable) dis ibu ions. In pa icula , hey show an inc easing SSW occu ence om ea ly win e ha maximizes in
Janua y and dec eases by la e win e (Fig. 2a), in ag eemen wi h he empo al e olu ion o he s anda d de ia ion o he zonal-
210
mean zonal wind a 60ºN and 10 hPa in he his o ical pe iod (Fig. 1c). In con as , SSWs o NCEP/NCAR a e mo e uni o mly
dis ibu ed wi h h ee sha p maxima in ea ly, mid and la e win e . The ea ly win e peak o SSWs in NCEP/NCAR ag ees well
wi h he clima ological pola s a osphe ic s a e, which shows a weake PNJ and a wa me pola s a osphe e han he o he
wo eanalyses (Fig. 1a and c). These NCEP/NCAR di e ences in he PNJ a e only s a is ically signi ican o he pola
s a osphe ic empe a u e and ERA-40 hough, likely due o he sho sample and he gene al la ge in e annual a iabili y o
215
he win e pola s a osphe e. Howe e , hey ag ee wi h an a i icial posi i e empe a u e end o 8ºC a 10 hPa o 1948-1998
in he NCEP/NCAR eanalysis, as documen ed by Badin and Domeisen (2014). On he o he hand, he lowe wind a iabili y
in Janua y in NCEP/NCAR would ag ee wi h he educed equency o SSWs in ha mon h and eanalysis, as compa ed o
he o he da ase s. In he compa ison pe iod he esul s a e simila ac oss eanalyses, which show s a is ically indis inguishable
dis ibu ions (Kolmogo o -Smi no es , Fig. 2b). In his pe iod, he maximum occu ence shi s o la e win e in all da ase s
220
compa ed o he dis ibu ions o ERA-40 and JRA-55 in he his o ical pe iod. Simila di e ences in he in a-seasonal
dis ibu ion o e en s we e al eady documen ed by Gómez-Escola e al. (2012) be ween he p e- and pos - 1979 pe iods.
Despi e he la ge unce ain y o he ea lie pe iod, hei dis ibu ions a e s a is ically signi ican ly di e en a he 99%
8
con idence le el and his esul suppo s he hypo hesis o mul i-decadal a ia ions in he in a-seasonal occu ence o SSWs,
which adds o he epo ed a iabili y in he o al win e equency o SSWs (Schimanke e al. 2011; Reichle e al. 2012;
225
Domeisen 2019).
3.2 P ocess-based diagnos ics
The p ocesses in ol ed in he occu ence o SSWs ha e been compa ed ac oss eanalyses by using he diagnos ics de ined in
Sec ion 2d. In his case, and in he es o he pape , we ha e used he common da es o SSWs o make su e he di e ences
ound ac oss eanalyses a e no due o he inclusion o di e en e en s.
230
Figu e 3 shows he s a is ics (mean, median and in e qua ile ange) o he dynamical benchma ks o all eanalyses in he wo
pe iods. A quick compa ison o he MRM o hese benchma ks o bo h pe iods e eals ha SSWs a e p eceded by a simila
anomalous s eng hening o wa e ac i i y a 100hPa, a e associa ed wi h a compa able decele a ion o he PNJ and ha e a
simila ampli ude in he middle and lowe s a osphe e in bo h pe iods. Only sligh di e ences a e ound in he median o
decelu and amp100 (compa e Fig. 3b,c wi h Fig.3 ,g). Howe e , gi en ha he median and mean o hese magni udes o one
235
pe iod a e included wi hin he in e qua ile ange o he o he , we can conclude ha SSWs cha ac e is ics a e simila in bo h
pe iods o s udy.
The compa ison pe iod shows good ag eemen among all eanalyses as all da ase s a e cha ac e ized by simila median, mean
and sp ead alues (Fig. 3e-h). Ne e heless, sligh de ia ions can be ound o NCEP/NCAR in he dis ibu ion o decelu,
which is shi ed owa ds lowe alues and shows a educed sp ead among e en s, as compa ed o he es o he da ase s (Fig.
240
3g). These de iciencies a e e en clea e in he his o ical pe iod, when a simila disc epancy is de ec ed in amp010 (Fig. 3a),
consis en wi h he educed s eng h and a iabili y o he PNJ in NCEP/NCAR eanalysis (Figs. 1c). As he de ia ion o decelu
in he NCEP-NCAR eanalysis is common o bo h pe iods, his migh poin o a bias o he model, whose e ec s a e ampli ied
in he i s pe iod by he lowe amoun o assimila ed da a. As men ioned be o e, his bias is e y likely linked o he low op
o he model and he low e ical esolu ion in he s a osphe e, p o ided ha he SSW cha ac e is ics a lowe le els (i.e.
245
amp100, ac wa ) do no di e much om hose o o he eanalyses. No e ha hese di e ences a e s ill no iceable in NCEP-
DOE, in ag eemen wi h Long e al. (2017) ha iden i ied simila biases in he clima ology and in e annual a iabili y o
empe a u e and zonal winds o bo h NCEP eanalyses. The model o NCEP-DOE is basically he same as ha o
NCEP/NCAR eanalysis al hough wi h an upda ed e sion (1995 s 1998) (Fujiwa a e al. 2017). This implies ha bo h
eanalyses use a model wi h a low esolu ion in he s a osphe e and wi h assimila ed empe a u e da a ins ead o di ec
250
adiances ha educe hei abili y o ep esen he s a osphe e (Fujiwa a e al. 2017). Despi e hei simila i ies, he NCEP-
DOE pe o ms be e wi h espec o he MRM, pa icula ly o decelu, a guably due o imp o emen s in oduced in he upda ed
e sion o he eanalysis model. P ima ily, NCEP-DOE was un wi h a new ozone clima ology (Kanami su e al., 2002). O he
di e ences in he concen a ion o CO2 o he adia ion scheme be ween bo h eanalyses migh also explain he di e ences
be ween bo h NCEP eanalyses (Fujiwa a e al., 2017).
255
9
A simila analysis has been ca ied ou sepa a ely o WN1 and WN2 SSWs in he compa ison pe iod (Fig. S1). All da ase s
ep oduce a simila beha io o bo h ypes o e en s and all diagnos ics, wi h he excep ion o he associa ed decele a ion o
he PNJ in he middle s a osphe e: WN2 SSWs a e ela ed o la ge decele a ions o he PNJ, p obably because hey a e usually
p eceded by a s onge pola o ex han WN1 SSWs (Albe s and Bi ne , 2014; Díaz-Du án e al., 2017). These esul s also
con i m he o e all good ag eemen ac oss eanalyses excep o he de iciency o NCEP/NCAR conce ning decelu.
260
Un o una ely, hese indings canno be con i med in he his o ical eanalyses due o he e y low equency o WN2 e en s
in ha pe iod (no shown).
4 Dynamical o cing
4.1 Upwa d-p opaga ing wa e ac i i y
Figu es 4 and 5 show he composi ed anomalous eddy HF, a ea-a e aged be ween 45° N and 75°N, a di e en le els a ound
265
he SSWs onse da e o he his o ical and compa ison pe iod, espec i ely. Only esul s om 300 o 10 hPa a e p esen ed, as
he [300-100] hPa laye co esponds o he communica ion egion o he s a osphe e- oposphe e coupling (de la Cáma a e
al. 2017), and he le els abo e his laye ypically show he s onges HF anomalies. The MRM shows a s ong anomalous
peak o HF a ound he cen al da e o SSWs in bo h pe iods. This s ong peak is p eceded by a weak pulse a ound [-20, -15]
days in he middle s a osphe e in he compa ison pe iod bu no in he his o ical one. The la ges di e ences ac oss eanalyses
270
a e de ec ed in he middle s a osphe e in ag eemen wi h Ma ineau e al. (2018), and hey a e mo e p onounced o he
his o ical han o he compa ison pe iod.
By applying he me hodology by Nishii e al. (2009) we ha e analyzed he con ibu ing ole o he di e en HF e ms o he
occu ence o SSWs. The MRM decomposi ion o he HF in he compa ison pe iod shows ha he s onges peak ([-5,0]-day
in e al) is mainly due o he ac ion o anomalous wa es ( i s igh hand e m o Eq. 1), albei wi h a ele an con ibu ion o
275
he cons uc i e in e ac ion be ween clima ological and anomalous wa es (second and hi d igh hand e ms o Eq. 1, Figs.
4c, e and 5c, e). Con e sely, he p eceding weake pulses o he compa ison pe iod seem o be mo e domina ed by he
in e ac ion e m. The ag eemen among eanalyses conce ning he ela i e oles o hese e ms is highe o he compa ison
pe iod, mainly in he middle s a osphe e, han o he his o ical pe iod (compa e Fig. 4d, s Fig. 5d, ).
Gi en he documen ed di e ences in he dynamical o cing o di e en ypes o SSWs (e.g. Smi h and Kushne , 2012;
280
Ba ioped o and Cal o, 2014), we ha e epea ed he analysis sepa a ely o WN1 and WN2 SSWs (Fig. 6). I has only been
done o he compa ison pe iod, due o he low sample size o WN2 e en s o he his o ical one. Al hough he e is no a
uni ocal ela ionship be ween D and S SSWs and WN1 and WN2 e en s (Waugh, 1997), ou esul s o WN1 and WN2 e en s
ag ee well wi h hose o Smi h and Kushne (2012) o D and S SSWs. WN1 e en s a e mainly igge ed by pe sis en bu
mode a ely in ense anomalies o HF du ing di e en pe iods ([-20, -15] and [-10, 0] days), which a e associa ed wi h he
285
cons uc i e in e e ence o clima ological and anomalous wa es (Figs. 6e and i). In con as , WN2 e en s a e ela ed o in ense
bu sho pulses o eddy HF in he i e days p io o he cen al da e. These pulses a e p edominan ly due o he anomalous
16
Cohen, J. and Jones, J.: T oposphe ic p ecu so s and s a osphe ic wa mings, J. Clima e, 24, 6562-6572, 2011.
Da ini, P., Cagnazzo, C., Gualdi, S., and Na a a, A.: Bidimensional diagnos ics, a iabili y, and ends o No he n
Hemisphe e blocking, J. Clima e, 25, 6496-6509, 2012.
Dee, D. P. and coau ho s: The ERA-In e im eanalysis: con igu a ion and pe o mance o he da a assimila ion sys em, Q. J.
R. Me eo ol. Soc., 137, 553-597, 2011.
480
De la Cáma a, A., Albe s, J. R., Bi ne , T., Ga cía, R. R., Hi chcock, P., Kinnison, D. E. and Smi h, A. K.: Sensi i i y o
sudden s a osphe ic wa mings o p e ious s a osphe ic condi ions. J. A mos. Sci., 74, 2857-2877, 2017.
Díaz-Du án, A., Se ano, E., Aya zagüena, B., Ábalos, M., and de la Cáma a, A.: In a-seasonal a iabili y o ex eme bo eal
s a osphe ic pola o ex e en s and hei p ecu so s, Clim. Dyn., 49, 3473–349, doi: 10.1007/s00382-017-3524-1, 2017.
Domeisen, D. I. V.: Es ima ing he equency o sudden s a osphe ic wa ming e en s om su ace obse a ions o he No h
485
A lan ic Oscilla ion, J. Geophy. Res., 124, 3180-3194.
Fujiwa a, M. and coau ho s: In oduc ion o he SPARC Reanalysis In e compa ison P ojec (S-RIP) and o e iew o he
eanalysis sys ems, A mos. Chem. Phys., 17, 1417–1452, 2017.
Ga inkel, C. I. and Ha mann, D. L.: Di e en ENSO eleconnec ions and hei e ec s on he s a osphe ic pola o ex, J.
Geophys. Res., 113, D18114, doi: 10.1029/2008JD009920, 2008.
490
Ga inkel, C. I., Ha mann, D. L., and Sassi, F.: T oposphe ic p ecu so s o anomalous No he n Hemisphe e s a osphe ic
pola o ices, J. Clima e, 23, 3282-3299, 2010.
Ga inkel, C. I., Bu le , A. H., Waugh, D. W., Hu wi z, M. M., and Pol ani, L. M.: Why migh s a osphe ic sudden wa mings
occu wi h simila equency in El Niño and La Niña win e s? J. Geophys. Res., 117, F19106,
doi:10.1029/2012JD017777, 2012.
495
Ge be , E. P., O be, C., and Pol ani, L. M.: S a osphe ic in luence on he oposphe ic ci cula ion e ealed by idealized
ensemble o ecas s. Geophys. Res. Le ., 36, L24801, h ps://doi.o g/10.1029/2009GL04091, 2009.
Gómez-Escola , M., Fueglis ale , S., Cal o, N. and Ba ioped o, D.: Changes in pola s a osphe ic empe a u e clima ology
in ela ion o s a osphe ic sudden wa ming occu ence, Geophys. Res. Le ., 39, doi:10.1029/2012GL053632, 2012.
Hi chcock, P.: On he alue o eanalyses p io o 1979 o dynamical s udies. A mos. Chem. Phys. 19, 2749-2764,
500
h ps://doi.o g/10.5194/acp-2018-879, 2019.
Hu, Y. and Tung, K. K.: Possible ozone-induced long- e m changes in plane a y wa e ac i i y in la e win e , J. Clima e, 16,
3027-3038, 2003.
Kalnay, E. and coau ho s: The NCEP/NCAR40-yea eanalysis p ojec , Bull. Ame . Me eo . Soc., 77, 437-471, 1996.
Kanami su, M. and coau ho s: NCEP-DOE AMIP II eanalysis (R-2), Bull. Ame . Me eo . Soc., 83, 1631-1643, 2002.
505
Kim, J., Son, S.-W., Ge be , E., and Pa k, H. –S.: De ining sudden s a osphe ic wa ming in clima e models: Accoun ing o
biases in model clima ologies, J. Clima e, 30, 5529-5546, 2017.
Kobayashi, S. and coau ho s: The JRA-55 eanalysis: Gene al speci ica ions and basic cha ac e is ics, J. Me eo . Soc. Japan,
93, 5-48, 2015.

17
Leh onen, I. and A. Y. Ka pechko: Obse ed and modeled oposphe ic cold anomalies associa ed wi h sudden s a osphe ic
510
wa mings, J. Geophys. Res., 121, 1591-1610, doi:10.1002/2015JD023860, 2013.
Limpasu an, V., Thompson, D. W. J., and Ha mann, D. L.: The li e cycle o he No he n Hemisphe e sudden s a osphe ic
wa mings, J. Clima e, 17, 2584-2596, 2004.
Long, C. S., Fujiwa a, M., Da is, S., Mi chell D. M., and W igh , C. J.: Clima ology and in e annual a iabili y o dynamic
a iables in mul iple eanalyses e alua ed by he SPARC Reanalysis In e compa ison P ojec (S-RIP), A mos. Chem.
515
Phys., 17, 14593–14629, 2017.
Ma ineau, P., Son, S-W, Taguchi, M., and Bu le , A. H.: A compa ison o he momen um budge in eanalysis da ase s du ing
sudden s a osphe ic wa ming e en s, A mos. Chem. Phys., 18, 7169-7187, 2018.
Ma ius, O., Pol ani, L. M., and Da ies, H. C.: Blocking p ecu so s o s a osphe ic sudden wa ming e en s. Geophys. Res.
Le ., 36, L14806, doi:10.1029/2009GL038776, 2009.
520
Ma suno, T.: A dynamical model o s a osphe ic sudden wa ming, J. A mos. Sci., 28, 1479–1494, 1971.
Maycock, A. C. and Hi chcock, P.: Do spli and displacemen sudden s a osphe ic wa mings ha e di e en annula mode
signa u es? Geophys. Res. Le ., 42, 10943-10951, 2015.
McLand ess, C., and Shephe d, T. G.: Impac o clima e change on s a osphe ic sudden wa mings as simula ed by he
Canadian Middle A mosphe e Model, J. Clima e, 22, 5449-5463, 2009.
525
Mi chell, D. M., G ay, L. J., Ans ey, J., Baldwin, M. P., and Cha l on-Pe ez, A. J.: The in luence o s a osphe ic o ex
displacemen s and spli s on su ace clima e, J. Clima e, 26, 2668–2682, 2013.
Nakagawa, K. I. and Yamazaki, K.: Wha kind o s a osphe ic sudden wa ming p opaga es o he oposphe e? Geophys. Res.
Le , 33, L04801, doi: 10.1029/2005GL024784, 2006.
Nishii, K., Nakamu a, H., and Miyasaka, T.: Modula ions in he plane a y wa e ield induced by upwa d-p opaga ing Rossby
530
wa e packe s p io o s a osphe ic sudden wa ming e en s: A case-s udy, Q. J. R. Me eo ol. Soc., 135, 39-52, 2009.
Nishii, K., Nakamu a, H., and O solini, Y. J.: Geog aphical dependence obse ed in blocking high in luence on he
s a osphe ic a iabili y h ough enhancemen and supp ession o upwa d plane a y-wa e p opaga ion, J. Clima e, 24,
6408-6423, 2011.
Onogi, K. and coau ho s: The JRA-25 eanalysis, J. Me eo . Soc. Japan, 85, 369-432, 2007.
535
Palmei o, F. M., Ba ioped o, D., Ga cía-He e a R., and Cal o, N.: Compa ing sudden s a osphe ic wa ming de ini ions in
eanalysis da a, J. Clima e, 28, 6823-6840, 2015.
Plumb, R. A.: Ins abili y o he dis o ed pola nigh o ex: A heo y o s a osphe ic wa mings, J. A mos. Sci., 38, 2514–
2531, 1981.
Reichle , T., Kim, J., Manzini, E., and K öge , J.: A s a osphe ic connec ion o A lan ic clima e a iabili y, Na . Geosci., 5,
540
783-787, 2012.
Rex, D. F.: Blocking ac ion in he middle oposphe e and i s e ec upon egional clima e. Tellus, 2, 196–211, doi:10.1111/
j.2153-3490.1950. b00331.x, 1950.
18
Rienecke , M. M. and coau ho s: MERRA: NASA’s Mode n-E a Re ospec i e Analysis o Resea ch and Applica ions, J.
Clima e, 24, 3624-3648, 2011.
545
Saha, S. and coau ho s: The NCEP clima e o ecas sys em eanalysis, Bull. Ame . Me eo . Soc., 91, 1015-1057, 2010.
Sausen, R., König, W., and Sielmann, F.: Analysis o blocking e en s om obse a ions and ECHAM model simula ions,
Tellus, 47A, 421–438, 1995.
Sche e , S. C., C oci-Maspoli, M., Schwie z, C., and Appenzelle , C.: Two-dimensional indices o a mosphe ic blocking and
hei s a is ical ela ionship wi h win e clima e pa e ns in he Eu o-A lan ic egion, In . J. Clima ol., 26, 233-249, 2006.
550
Schimanke, S., Kö pe , J., Spangehl, T., and Cubasch, U.: Mul i‐decadal a iabili y o sudden s a osphe ic wa mings in an
AOGCM, Geophys. Res. Le ., 38, L01801, oi:10.1029/2010GL045756, 2011.
Se iou , W. J. M., Mi chell, D. M., and G ay, L. J.: A p ac ical me hod o iden i y displaced and spli s a osphe ic pola o ex
e en s, Geophys. Res. Le ., 40, 5268–5273, doi:10.1002/g l.50927, 2013.
Sjobe g, J. P., and Bi ne , T.: S a osphe ic wa e-mean low eedbacks and sudden s a osphe ic wa mings in a simple model
555
o ced by upwa d wa e ac i i y lux. J. A mos. Sci., 71, 4055-4071, 2014.
Smi h, K. L. and Kushne , P. J.: Linea in e e ence and he ini ia ion o ex a opical s a osphe e– oposphe e in e ac ions. J.
Geophys. Res., 117, D13107, doi: 10.1029/2012JD017587, 2012.
Taguchi, M. and Ha mann, D. L.: Inc eased occu ence o s a osphe ic sudden wa mings du ing El Nio as simula ed by
WACCM, J. Clima e, 19, 324-332, 2006.
560
Taguchi, M.: Is he e a s a is ical connec ion be ween s a osphe ic sudden wa ming and oposphe ic blocking e en s? J.
A mos. Sci., 65, 1442-1454, 2008.
T eidl, R. A., Bi ch, E.C., and Sajecki, P.: Blocking ac ion in he No he n Hemisphe e: A Clima ological s udy, A mosphe e-
Ocean, 19, 1-23, 1981.
T ipa hi, O. P. and coau ho s: Examining he p edic abili y o he s a osphe ic sudden wa ming o Janua y 2013 using mul iple
565
NWP sys ems, Mon. Wea. Re ., 144, 1935–1960, 2016.
Uppala S. M. and coau ho s: The ERA-40 e-analysis, Q. J. R. Me eo ol. Soc., 131, 2961-3012, 2005.
Waugh, D. W.: Ellip ical diagnos ics o s a osphe ic pola o ices, Q. J. R. Me eo ol. Soc., 123, 1725-1748, 1997.
Whi e, I., Ga inkel, C. I., Ge be , E. P., Jucke , M., Aquila, V., Oman, L. D.: The downwa d in luence o sudden s a osphe ic
wa mings: Associa ion wi h oposphe ic p ecu so s, J. Clima e, 32, 85-108, 2019.
570
Wiedenmann, J. M., Lupo, A. R., Mokho , I. I., and Tikhono a, E. A.: The clima ology o blocking an icyclones o he
No he n and Sou he n Hemisphe es: Block in ensi y as a diagnos ic, J. Clima e, 15, 3459-3473, 2002.
Woollings, T.J., Hoskins, B. J., Blackbu n, M., and Be is o d, P.: A new Rossby wa e-b eaking in e p e a ion o he No h
A lan ic Oscilla ion. J. A mos. Sci., 65, 609–626, 2008.
Woollings, T., Cha l on-Pe ez, A., Ineson, S., Ma shall, A. G., and Masa o, G.: Associa ions be ween s a osphe ic a iabili y
575
and oposphe ic blocking, J. Geophys. Res., 115, D06108, doi:10.1029/2009JD012742, 2010.
19
Woollings, T., Ba ioped o, D., Me h en, J., Son, S-W., Ma ius, O., Ha ey, B., Sillmann, J., Lupo, A. R., and Sene i a ne,
S.: Blocking and i s esponse o clima e change. Cu . Clim. Change Rep., 4, 287-300, 2018.
580
20
Table 1: Classi ica ion o he common SSWs in o WN1 and WN2 e en s in he compa ison pe iod. (In b acke s he S/D classi ica ion).
WN1 SSWs
WN2 SSWs
29 02 1980 (D) 11 02 2001 (D)
04 03 1981 (D) 31 12 2001 (D)
04 12 1981 (D) 18 01 2003 (S)
24 02 1984 (D) 05 01 2004 (D)
23 01 1987 (D) 21 01 2006 (D)
08 12 1987 (S) 24 02 2007 (D)
14 03 1988 (S) 09 02 2010 (S)
15 12 1998 (S) 24 03 2010 (D)
26 02 1999 (S)
22 02 1979 (S)
01 01 1985 (S)
21 02 1989 (S)
20 03 2000 (D)
22 02 2008 (D)
24 01 2009 (S)
Table 2: F equency o SSWs pe decade and a io o o ex displacemen (D) s o ex spli (S) SSWs o each eanalysis and pe iod
o s udy.
585
His o ical pe iod
(1958-1978)
Compa ison pe iod
(1979-2012)
Reanalyses
F equency
(SSWs/dec)
Ra io D/S
F equency
(SSWs/dec)
Ra io D/S
ERA-40
6.2
1.6
NCEP-NCAR
4.8
0.7
6.2
1.6
JRA-55
5.7
1.0
6.8
1.2
ERA-In e im
6.2
1.6
JRA-25
6.5
1.8
NCEP-CFSR
6.5
1.4
NCEP-DOE
6.5
1.4
NASA-MERRA
6.5
1.2
21
590
Figu e 1: 21-day unning mean o he daily clima ology (solid line) and s anda d de ia ion (dashed line) in he his o ical pe iod
(1958-1978) o : (a) pola -cap (50ºN -90ºN) a e aged empe a u e a 10 hPa, (b) pola -cap (50ºN -90ºN) a e aged empe a u e a 100
hPa, (c) zonal mean zonal wind a 60°N and 10 hPa and (d) hea lux a 100 hPa a e aged o e 45ºN -75N. The le ( igh ) y-axis
595
e e s o he mean (s anda d de ia ion) in each plo . Thick lines indica e alues o ERA-40 o JRA-55 ha a e signi ican ly di e en
om hose o NCEP-NCAR eanalysis a he 95% con idence le el. Magen a c osses co espond o JRA-55 alues ha a e
signi ican ly di e en om ERA-40 ones a he 95% con idence le el (S uden ’s - es ).
600
No Dec Jan Feb Ma Ap
210
215
220
225
230
T (K)
Pola cap T @ 10hPa
0
2
4
6
8
T (K)
No Dec Jan Feb Ma Ap
210
215
220
225
230
T (K)
Pola cap T @ 100hPa
0
0.75
1.5
2.25
3
T (K)
No Dec Jan Feb Ma Ap
-10
0
10
20
30
40
u (m s-1)
Zonal mean u @ 60N & 10hPa
0
5
10
15
20
25
u (m s-1)
No Dec Jan Feb Ma Ap
0
5
10
15
20
Hea lux (K m s-1)
Hea lux @ 100hPa (45-75N)
ERA-40
JRA-55
NCEP-NCAR
s d. ERA-40
s d. JRA-55
s d. NCEP-NCAR
0
2.5
5
7.5
10
Hea lux (K m s-1)
a
c
b
d

22
Figu e 2. SSW o al equency dis ibu ion wi hin ±10 day pe iods om he da e displayed in he x-axis o : (a) he his o ical pe iod
(1958-1978) and (b) he compa ison pe iod (1979-2012).
23
605
Figu e 3. Box plo s showing he dis ibu ion o he dynamical benchma ks o SSWs (amp010, amp100, decelu and ac wa ) in he
his o ical (1958-1978) and compa ison (1979-2012) pe iods. The in e qua ile ange is ep esen ed by he size o he box and he ed
line (black c oss) co esponds o he median (mean). Whiske s indica e he maximum and minimum poin s in he dis ibu ion ha
a e no ou lie s. Ou lie s ( ed c osses) a e de ined as poin s wi h alues g ea e han 3/2 imes he in e qua ile ange om he ends
o he box. See ex o he de ini ion o dynamical benchma ks.
610
24
Figu e 4. (a) Composi ed ime e olu ion o he o al anomalous hea lux a e aged o e 45°N-75°N (K m s-1) a di e en p essu e
le els om 29 days be o e o 30 days a e he occu ence o SSWs in he his o ical (1958-1978) pe iod. Con ou in e al is 20 K m
s-1. (b) Same as (a) bu o he s anda d de ia ion o he eanalyses wi h espec o he MRM di ided by he squa e oo o he numbe
615
o eanalyses. Con ou in e al is 1 K m s-1. (c) and (d) Same as (a) and (b) bu o he in e ac ion be ween clima ological and
anomalous wa es. Con ou in e als a e 10 K m s-1 and 2 K m s-1, espec i ely. (e) and ( ) Same as (a) and (b) bu o he con ibu ion
o he anomalous wa es o he o al anomalous hea lux. Con ou in e als a e 10 K m s-1 and 2 K m s-1, espec i ely. Shading in (a),
(c) and (e) deno es s a is ically signi ican anomalies a he 95% con idence le el o he same sign in a leas 66.7% o all eanalyses
(Mon e-Ca lo es ).
620
25
Figu e 5. Same as Fig. 4 bu o he compa ison (1979-2012) pe iod.