Exceptionally cold and mild winters in Europe (1951–2010)

Extreme thermal conditions appear to occupy an important place among research subjects at a time of climate warming. This study investigates the frequency, duration and spatial extent of thermally anomalous winters in Europe during the 60 years between 1951 and 2010. Exceptionally cold winters (ECWs) and exceptionally mild winters (EMWs) were identified using the statistical criterion of plus/minus two standard deviations from the long-term winter temperature (January–December) recorded at 60 weather stations. It was demonstrated that ECWs have occurred more frequently and covered larger territories than EMWs and that they may occur anywhere in Europe, while EMWs were limited to its southern and western parts. ECWs are characterised by greater absolute temperature anomalies, as anomalies greater than |6.0 °C| account for 35 % of ECWs, but only for 8 % of EMWs. The greatest anomalies are found in the east of the continent. The largest territory affected by an ECW included 24 stations in 1962/1963, while the equivalent among the EMWs included 11 stations in 2006/2007. The study also confirmed an expected trend whereby ECWs diminished in frequency in favour of EMWs in the second half of the 60-year study period.


Introduction
In the tem perate climate zone, w hich covers m ost o f the E u ropean continent, w inter has the m ost variable therm al condi tions o f all the seasons. These are the result o f the typical pattern o f atm ospheric circulation dom inant at the time, w hich alternates betw een air m ass advection from over the A tlantic Ocean, w hich favours m ild winters, and a build-up o f station ary high pressure systems that block such advection and allow inflow o f cold air from the north or east and lead to very low tem peratures and very cold w inters (e.g. K ossow ska-Cezak 1997 ;Jaagus 2006;Bardin 2007; U gryum ov and K h ar'kova 2008; Van den Besselaar et al. 2010). The occurrence o f such severe winters, as w ell as shorter w inter spells w ith very low temperatures and, to a lesser extent o f very m ild w inter sea sons, has long been the object o f universal attention as they have affected all the populations o f the territories in question an d h a v e h a d a m u ltifa c e te d in flu e n c e o n th e ir liv e s (Błażejczyk and M cG regor 2007;M aignan et al. 2008).
A ccounts o f such exceptional w inters can be found in historic chronicles. A study b y R. Girguś and W. Strupczewski (W yjątki..., 1965) offers valuable inform ation in this respect, including the following examples: In the year o f our Lord 1076, a very severe winter befell the lands and their largest rivers, such as the Saone, Rhone, Rhine and Loire in Gaul; the Elbe, Vistula and Danube in Germania; and the Po in Italia which were seized by ice so strong that they fell silent immediately after the smaller ones and the astounded local population w ent over them as if on solid ground (W y ją tk i., 1965, p. 17). The w inter [1306] w as so severe that betw een the D an ish mainland, its islands and Sweden all the seas turned into perm anent bridges for fourteen w eeks and even longer ( W y ją tk i., 1965, p. 26).
The w inter o f 1322/1323: A lso betw een Norway, England and France m any ships were caught b y ice in the open seas, so that merchants w a lk e d o n th e ice to p ay v isits to each o th e r for entertainm ent" ( W y ją tk i., 1965, p. 30).
O n the following page: D uring the tim e betw een St. A ndrew 's D ay [30 N ovem ber 1322] and L aetare Sunday [6 M arch 1323], the w eather w as so cold that m erchants carried their goods on carts across the sea from N orw ay to Sweden and back and there w ere inns and taverns on the seas w here they consum ed their beer and food. A lso m erchants v is ited from ports in Prussia and Livonia and there w ere also taverns in the m arketplace (W y ją tk i., 1965, p. 31).
M ild w inters w ere m uch rarer: W inter o f that y ear [1412] w as exceptionally w arm w ithout ground frost or any frost in general to the extent that even in Lithuania, a cold and frosty land, people already had vegetables to eat and flowers around Pre sentation D ay [2 February], w hich w as regarded as a great w onder and a veritable miracle ( W y ją tk i., 1965, p. 41).
In that year [1493], the w inter in January and February w as so m ild that trees blossom ed in orchards, grass grew tall, birds nested, but in M arch frost destroyed every thing ( W y ją tk i., 1965, p. 85-86).
These are ju st a sample o f numerous accounts o f particularly severe w inter seasons, which ceased to occur in Europe due to th e p erio d o f clim ate-w arm in g o b serv ed since th e m id nineteenth century. Nevertheless, anom alously cold winters still occur with serious consequences ranging from affecting daily life, disorganising the econom y to even influencing historic events, such as a series o f three exceptionally cold winters during the Second World War. In this last example, in January 1942, the temperature in the w ar zone in Russia dropped to -5 6 °C (Bronnimann 2005). O n the other hand, exceptionally m ild winters can also have adverse effects as they disturb the cycle o f natural processes (M aignan et al. 2008).
This study pursues this still current topic and focuses on the frequency, duration and spatial extent o f anom alous w inter seasons in Europe betw een the m id-tw entieth century and 2 0 1 0 .
The research hypothesis adopted here proposes that the observed increase in w inter tem perature is expressed b y a decline o f the frequency o f EC W s an d an increase in the frequency o f EM W s. This pap er continues from previous studies b y the authors (Tw ardosz and K ossow ska-C ezak 2013a, b , 2015a, b) on exceptionally h ot and cold summer seasons in Europe. The intention is to provide com prehensive understanding o f therm al anom alies in w inter using a single m ethod o f their identification, long-term observation data and covering the w hole o f Europe, as opposed to a w ide body o f fragm ented research targeting different areas and periods (e.g. B aur 1954; G raham et al. 2 0 0 6 ; H irschi an d Sinha 2 0 0 7 ; Hirschi 2008;Cattiaux et al. 2010;Wang et al. 2010;Ouzeau et al. 2011;B uchan et al. 2014).

Data and methodology
The study is based on the average m onthly air temperatures recorded at 60 w eather stations in continental Europe and the British Isles.  (Table 1). For the purposes o f the study, the stations were num bered from 1 to 60 going from w est to east in eight belts o f 5° o f latitude from north to south. Four stations are located below 40° N , 14 stations in the belt 40-45° N , 10 stations each in the belts from 45 to 60° N , seven stations in 6 0 -6 5° N , four stations in 65-70° N and a single station above the latitude 70° N ( Table 1) T his datab ase is one o f fo u r g en erally av ailab le d ata sources (M oberg et al. 2006) offering high spatial resolution and large num bers o f com plete series with high quality data confirm ed b y hom ogeneity testing (W ijngaard et al. 2003). The authors, as w ell as m any others, have used this database in the past (e.g. C ony et al. 2008; Van den Besselaar et al. In this study, the underlying assum ption is that w inter lasts betw een D ecem ber and February. A nom alous w inters are con sidered to occur w hen the average air tem perature recorded by a given station differs from the corresponding multi-annual average  by at least tw o standard deviations. Exceptionally cold w inters (ECW s)-t< tav. -2ct and excep tionally m ild w inters (EM W s)-t> tav.+2tr) w ere distin guished on the basis o f this assumption. Exceptionally cold and m ild w inter m onths (ECM s, EM M s) w ere identified using th e sam e m eth o d . T he sta n d ard d ev iatio n criterio n w as a d o p te d b y a n a lo g y to th a t u s e d fo r d is tin g u is h in g    (Table 1).

Frequency of exceptionally cold winters and exceptionally mild winters
ECW s w ere recorded by all the stations, w ith their num ber ranging from 1 to 3 per station and reaching 4 at 3 stations only (Table 1); in southern and w estern Europe, their num ber tended to be 1-2, w hereas the equivalent figure w as 2 -3 in N orthern and Eastern Europe. The num ber o f ECW s and E M W s found at individual stations (3 ± 2) differed from their norm al distribution (ca. 5 % or 3). The difference is explained b y the fact that as a natural phe nom enon, the w intertim e distribution o f average daily tem per ature does not follow the norm al distribution pattern. Indeed, it is to be expected that a single case o f an extremely high or low average temperature m ay occur thus significantly altering the standard deviation and, consequently, the n u m b er o f anom alous values identified.
The highest frequency o f ECW s w as observed in the de cade 1960/1961-1969/1970 (Table 2) EM W s were m uch less frequent than ECW s (Table 2)  A s can be seen, the num ber o f therm ally anom alous winters differs across Europe. This is attributable to the differences in temperature o f the air m asses carried into Europe in winter. These can be relatively w arm m aritime polar air m asses (mP) and strongly cooled continental polar (cP) or A rctic (A) air m asses. The overall predom inance o f w estern circulation causes the former (mP) to occur m ore frequently, in particular over W estern and Central Europe, leading to m ild w inters w ith tem peratures above average b ut n ot changing m uch over time. As mentioned above, m ost o f the ECW s (82.5 %) saw excep tionally cold months (ECMs), and in extreme cases, all 3 months were ECMs. By contrast, the EM W s were characterised by less frequent exceptionally mild months (EMMs).
O ver the 60 years under study, two w inters proved to be ECW s and EM W s at the same time. They w ere the w inters o f Tem perature (°C)  (Tables 3 and 4).

Coverage and thermal characteristics of ECWs and EMWs
The years w ith ECW s and EM W s, as w ell as the num ber o f the stations w ith their codes (as listed in Table 1), are show n in Tables 3 and 4 (Calendars o f E C W s and EM W s). A s m en tioned above, E C W s w ere recorded in 18 years over areas varying in size, i.e. b y different num bers o f stations-from 1 to 24. However, for the m ost part, ECW s were recorded by only a single station or b y two neighbouring stations; there w ere seven such w inters (Table 3). Those w inters w hich were recorded b y 1-2 stations, i.e. fewer than 5 % o f all the stations, w ill be disregarded in this study. The analysis w ill focus on those w inters that were recorded concurrently b y at least three stations, i.e. b y at least 5 % o f the stations). There w ere 11 such ECW s. EM W s w ere identified in 13 years. M ost o f them were also only recorded by 1 or 2 stations (7 E M W s in total), even though n o t always by neighbouring ones. A s it turned out, there w ere cases w hen ECW s w ere observed b y stations dis tant from each other-three or even m ore altogether.
This study w ill focus on those EM W s that were recorded by at least three stations in the same part o f Europe. There were four such EM W s. The one covering the largest area w as recorded b y 11 stations (Table 4).
The characteristics o f the ECW s and EM W s described in this study com prise their location in Europe and the indicative area covered, num ber o f stations, average air temperature and degree o f anom aly (i.e. air tem perature deviation from the corresponding m ulti-annual average), as w ell as the average m axim um and m inim um tem peratures and the num ber o f days w ith a m inim um temperature below 0 °C and m axim um tem perature below 0, -1 0 and -2 0 °C.

Exceptionally cold winters
The E C W o f 1953/1954 w as the w inter w ith the second largest coverage (16 stations) and w ith the greatest negative air tem perature anom aly in the 60-year period (the coldest w inter in the study area). The w inter covered the area o f Southeastern Europe (Fig. 2), stretching from Belgrade and D ebrecen in the w est to K azan and O renburg in the east and from the Black Sea and foreland o f the Caucasus in the south to K azan in the north. For m ost o f the stations, January and/or February were exceptionally cold m onths (ECM s) plus D ecem ber in the southernm ost area (3 E C M s in Istanbul an d Sim feropol). The tem perature anom aly A t in 5 southern stations exceeded 3 sta n d ard d ev iatio n s; ev ery w h ere, ex cep t fo r Sochi, it exceeded -4 °C, reaching -5 to -7 .0 °C in the east and -8.6 °C in Rostov-on-Don. It w as the greatest negative w inter temperature anom aly in the years 1951-2010 in Europe and one o f 5 cases w hen the negative anom aly exceeded -8 °C ( Table 5). The entire area saw days w ith 24-h temperatures below -1 0 °C (approx. 30 m ore than on average in the central and eastern parts) and in the east even below -2 0 .0 °C.
The E C W o f 1955/1956 covered the area o f northeastern Europe, w here it w as recorded b y ten stations (Fig. 2). It m ay be presum ed that the area o f the ECW w as n ot m uch larger than during the w inter 2 years previous to it w hereby the clearly low er num ber o f w eather stations resulted from the  1953/195416 12,14,15,16, (17), 18, 24, 25, 26, 27, 28, 35, 36, 37, 38, 47 Dec.-3 stations, Jan.-10 stations Feb.-15 stations 1955/1956 Dec.-8 stations Feb.-6 stations 1962/1963 24 (5), (6), 7, (8), (9), (10), 11,12,13,19,20,21,22,23,24,25,29,30,31,32,33,39,41,49 Dec Station numbers printed in italics mean that the average winter temperature at that station met the formula t<4, 3o\ A number in brackets means that the station had no single month in the category in that year a This is not a summary of all ECWs, but of just the months that decided that the winter was anomalous low est density o f the netw ork o f stations in that part o f Europe. The E C W w as recorded b y stations from K ursk and Saratov in the south to Saint Petersburg and Ivdel in the north, as w ell as in Vardo, and from Saint Petersburg in the w est to Ivdel and Y ekaterinburg in the east. D ecem ber and F ebruary w ere ECMs. Everywhere, except for Vardo, the tem perature anom aly A t w as greater than -5 .0 °C, exceeding -7 .0 °C in the north and reaching its highest value in Vologda w here it w as -8 .2 °C. In Saint Petersburg, Moscow, Vologda and Kazan, it w as the coldest w inter in the six decades (Table 5). The entire area, except for Vardo, saw days w ith 24-h temperatures b e low -2 0 .0 °C; they w ere m ore num erous than on average, betw een 10 days in Saint Petersburg and 25 days in Ivdel. U nlike the E C W discussed above, the E C W o f 1962/1963 w as recorded in W estern Europe b y a record num ber o f 24 stations, from M adrid to Rom e in the south to Edinburgh and Bergen in the north and from L a Coruna and Valentia in the w est to Warsaw, Chernivtsi and Sofia in the east (Fig. 2). January w as an ECM at nearly all locations. In the northwest, D ecem b er an d /o r F eb ru ary w ere also ex cep tio n ally cold m onths; three ECM s w ere recorded in Paris, W urzburg and De Bilt. The only areas w ith no ECM were the Iberian Penin sula and the coast o f the M editerranean Sea. E ven though m ost o f the tem perature anom alies in the northeast o f the continent reached three standard deviations (t> tav.+3(j), the absolute value o f anom alies w as low er than for the above ECW s in the east o f Europe w ith A t usually ranging betw een ca. -2 .0 and -5 .0 °C and its highest value recorded in W arsaw A t= -6 .7 °C (Table 5). W estern Europe ow es its low er negative air temperature anom alies in w inter to Station numbers printed in italics mean that the average winter tempera ture at that station met the formula t>fav.+3<j. A number in brackets means that the station had no single month in the category in that year the m itig atin g effect o f the w aters o f th e A tlantic O cean (Hirschi and Sinha 2007), in contrast to the cooled land in central, and even m ore so, Eastern Europe w hich is largely responsible for the significant falls in air tem perature. O ver m ost o f the area, the w inter o f 1962/63 w as the coldest in all the 60 years and w as the only anom alously cold w inter in the south and west, even though the average temperature in the southern-and w esternm ost areas rem ained above zero.
The E C W o f 1963/1964 w as observed b y three stations in Central Europe: in Vienna, D ebrecen and Belgrade. Decem ber and January were ECMs. The temperature anom alies ranged betw een -3 .9 °C in V ienna and -5 .4 °C in D ebrecen (the latter experienced its coldest w inter in the six decades).
T he E C W o f 1 9 6 5 /1 9 6 6 co v e red th e S can d in av ian P eninsula (7 stations; Fig. 3). In m o st cases, F ebruary w as an EC M , w ith no E C M reco rd ed b y the southern stations. It w as the coldest w inter in the area, w ith the tem per ature anom aly A t betw een -3 .0 °C in Vardo and -7 .5 °C in Vaasa (Table 5). In the northern, inner part o f the Scandinavian Peninsula, there w as not a single w inter day w hen the m axi m um tem perature exceeded 0 °C, w ith m ore than 20 days recording 24-h air tem peratures below -2 0 °C (few er than 10 days on average). aThe lowest in 60 years; a value in italics that the temperature meets the criterion t< tav -3<r; -no data The E C W o f 1968/1969 covered the easternm ost areas o f Europe (8 stations) and M insk (Fig. 3). January and Decem ber w ere ECM s in Ivdel and Yekaterinburg. The air temperature an om aly A t w as h ig h er th an -5 .0 °C across the sta tio n s, e x c e e d in g -8 .0 °C in P e c h o ra a n d O re n b u rg ( A t= -8 .1 and -8 .2 °C, respectively). It w as the coldest w inter in the 60 years for nearly all the stations. The average temperature o f -2 5 .4 °C in Pechora w as the low est o f all the cases studied here (Table 5) H ere too m axim um tem peratures rem ained below -2 0 .0 °C for over h alf the days in th at w inter (2 0 days on average). S u ch a strong decline in air tem perature in the area w as due to a b lo ck caused b y a ridge o f h ig h pressure o ver w estern R ussia (H irschi and S inha 2 0 0 7 ).
The E C W o f 1969/1970 w as observed b y four stations in central Europe: in Berlin, Warsaw, Stockholm and Liepaja. D ecem ber w as an E C M and February too, b ut only in S tock holm. The tem perature anom aly A t ranged betw een -4 .6 °C in V ienna and -5 .6 °C in Liepaja. In the w eather stations on the Baltic Sea, it w as the coldest w inter in the 60-year period.
The E C W o f 1978/1979 w as recorded b y six stations in the n o r th o f E u ro p e : D e B ilt a n d B e rg e n , L ie p a ja a n d A rkhangelsk, Syktyvkar and Pechora. In the last four stations, D ecem ber w as the ECM . G iven the high latitudinal span o f the ECW, the therm al conditions w ithin its coverage differed greatly: in the w esternm ost areas, the average air temperature w as approx. -1 .0 °C (A t -3 .0 to -4 .0 °C), and in the eastern m ost ones, it w as low er than -2 0 .0 °C (anom aly A t -7 .0 to -8 .0 °C). In the w est, there w ere only 30 days w ith 24-h tem peratures below zero, w hereas in the east they were re corded throughout the w inter; often, there w ere days w ith m axim um temperatures below -2 0 .0 °C. For Pechora, it w as one o f the two coldest w inters in the 60 years, w ith the tem perature reaching -2 5 .4 °C on average, i.e. as low as during the w inter o f 1968/1969 (Fig. 3, Table 5).
The E C W o f 1980/1981 w as recorded b y three stations on the Italian Peninsula and on the Adriatic Sea: in Rome, C ro tone and Split. D ecem ber and January were ECM s in Crotone, but it w as the coldest w inter in the 60 years for R om e alone. The air tem perature anom aly w as approx. -2 .0 °C.
The E C W o f 1984/1985 w as recorded in two areas: in the south-from Belgrade and Sofia to K iev and Simferopol (6 stations), and in the north-w ithin the Kajaani-Arkhangelsk-V ologda triangle (3 stations). In Belgrade and Sofia, there w ere no ECMs, and February w as an ECM for the rem aining stations. In the tw o northernm ost stations, January w as also an ECM. For these last stations, it w as the coldest w inter in the 60-year period, w ith the tem perature anom aly A t reaching -7 .1 °C in K ajaani and -8 .0 °C in A rkhangelsk. For the re m aining stations, A t ranged from -2 .6 °C in Sofia to -5 .5 °C in K iev and -6 .4 °C in Vologda (Fig. 3, Table 5). D ays with 24-h tem peratures below -1 0 °C were observed in both areas: in the south, there w ere fewer than 2 0 such days (in the west, even few er than 5), and in the north, there w ere up to 60 days, including approx. 30 days w ith m axim um tem peratures below -2 0 °C. A nother ECW recorded by at least three stations appeared only 25 years later-in the 2009/2010 season; it w as the only EC W in the tw enty-first century, i.e. "a cold extreme in a warming climate" (Cattiaux et al. 2010). It was observed by three seaside stations in the northwest o f Europe: in Valentia and Edinburgh (with no ECM, temperature anom aly A t o f -2.1 and -2 .4 °C, respectively) and in Bergen (ECM =January, A t= -3 .4 °C; here, it was the coldest w inter in the 60-year period).
Strongly negative air tem perature anom alies in W estern and N orthern Europe are observed at the tim e o f negative p h ases o f the N o rth A tlan tic O scillatio n (C attiaux et al. 2010;Wang et al. 2010;O uzeau et al. 2011;B uchan et al. 2014). During such spells, Central and Eastern Europe sees m ore anticyclonic circulation from the north and east or the formation o f highs directly over the area (Kossowska-Cezak 1997).

Exceptionally mild winters
The  (Fig. 4). February w as an EM M, plus Decem ber in the Iberian Peninsula. The tem perature anom aly in the m ain area w as betw een 1.4 °C in Lisbon and 3.0 °C in Bordeaux (Table 6). A part from Lisbon, it w as the w arm est w inter in six decades across the locations. In the m ain area, there w ere nearly no cases o f temperatures dropping below 0 °C. These w ere observed in Chernivtsi, b ut to a m uch lesser extent than on average Table 6).
The E M W o f 1994/ 1995 w as recorded by only three sta tions in two areas far apart from each other: Brest in the w est and N aryan-M ar and Pechora in the north-east. However, this EM W is notew orthy because the two last stations recorded the greatest positive air tem perature anom aly in the 60-year peri od, A t=7.3 and 7.7 °C respectively; these were the only cases o f an anom aly greater than 6.0 °C.
The E M W o f 1997/1998 w as recorded by three stations in different parts o f Europe that were located far apart from one another ( Table 4).
The E M W o f 2006/2007 covered the greatest area in the 60 years (Fig. 4). It w as recorded b y 11 stations across the belt stretching from Paris and D e Bilt in the northw est and C roto ne, Sofia and C onstanta in the southeast, as w ell as in the distant station o f Orenburg. There w as no EM M in the w estern  (Table 6 ). F or some parts of W estern Europe, it w as the w arm est w inter since temperature records using in struments began (Hirschi 2008). T hat winter, there w as n o t a single day w ith a 24-h tem perature below 0.0 °C (except for O renburg; Table 6 ). The m ild w inter w as follow ed b y a w arm sp rin g , w h ic h g re a tly ac c e le ra te d th e g ro w th o f p la n ts (M aignan et al. 2008). A s is asserted b y m any researchers, m ild w inters in W estern and N orthern Europe tend to accom pany positive phases o f the N orth Atlantic O scillation (NAO), w hich is attributed to higher than norm al pressure differences betw een the Icelandic L ow and the A zores High. A t such times, Central and Eastern Europe sees an increased frequency o f cyclonic types o f circulation from the w est and a fall in or even absence o f circu latio n from th e n o rth and th e east ( K o s s o w s k a -C e z a k 1 9 9 7 ; J a a g u s 2 0 0 6 ; I s a y e v a n d Sherstyukov 2008; Sidorenkov and Orlov 2008; A nisim ov e tal. 2011 ).

Conclusions
This study is based on the assum ption that w inter periods (from D ecem ber till February) are therm ally anom alous w hen their average air temperature differs from the corresponding m ulti-annual average   D epending on the part o f Europe, the term s "co ld " and "m ild " m ay denote w inters differing substantially in term s o f th e rm a l c o n d itio n s -in th e M e d ite rra n e a n B a sin , a "severe" w inter m ay actually be free from tem peratures below zero, and in the far north, even "m ild " w inters m ay involve prolonged spells o f freezing weather. The apparent paradox results from the fact that the identification o f tem perature anom alies w as based on the average tem peratures recorded in a given area to w hich its residents are accustomed.
It w as found that there are great differences in the frequen cy, location and area covered by individual anom alous w in ters, as w ell as in the level o f air temperature anom aly for both types o f anom alous winters.
1. E C W s are m ore frequent across Europe, w hile EM W s appear only in the south and w est o f the continent (they w ere only recorded b y h a lf o f the stations).
2. ECW s are recorded m ore frequently than EM W s-18 and 13 w inters respectively in the 60 years. 3. In extreme cases, ECW s cover larger areas than EMWs. 6 . The above com parison indicates that the num erous histor ical accounts o f very severe w inters and few accounts o f m ild w inters are n o t accidental b u t rath er indicative, am ong other things, o f a significant characteristic o f E urope's current climate. E ven though the second h alf o f the six decades saw few er ECW s and m ore EM W s ( 1 9 5 1 / 1 9 5 2 -1 9 7 9 / 1 9 8 0 : 11 E C W s , 3 E M W s , 1980/1981-2009/2010: 7 and 10, respectively), one may always expect spells o f temperature w ell below zero, short or long lasting, notably in the east and north o f the conti nent. This results from the nature o f the circulation o f air over Europe w hich is dom inated b y advection o f w arm air m asses from over the Atlantic, w ith intermittent periodic blocks causing the forced descent o f cold A rctic or conti nental polar air from the north or the east (positive or negative N A O phase). In addition, the prolonged presence o f high-pressure system s in w inter tim e contributes to further radiative cooling o f the air during long cloudless w inter nights (especially in the north). This pattern is con firmed b y an increased frequency o f ECW s in the early part o f the study period w hen the negative N A O phase w as also m ore frequent (Hurrell and National Center for Atm ospheric Research Staff 2014). 7. In the light o f the authors' previous study on exceptionally hot and cool summ ers (Twardosz and Kossow ska-Cezak 2015b) exceptionally cold w inters occurred less frequent ly than exceptionally hot summ ers (18 vs. 24 over the 60year period), but on average covered larger areas (6 vs. 4 stations; exceptionally m ild w inters and exceptionally cool summ ers accounted for 3 each). In the second h alf o f the 60-year period, exceptionally cold w inters becam e increasingly rare and th e ir spatial coverage b eg an to shrink, as opposed to exceptionally hot summers, w hich recorded gains in both frequency and spatial coverage.