Negative NAO and cold Eurasian winters: How exceptional was the winter of 96/96? Joël J.-M. Hirschi, Bablu Sinha National Oceanography Centre, Southampton, United Kingdom Introduction Since the mid 9s frequent positive phases of the North Atlantic Oscillation (NAO) linked to a higher-than-normal pressure difference between the Azores high and Icelandic low pressure centres and stronger westerlies (Hurrell, 99; Thompson and Wallace, 99; Osborn, 6) have led to higher winter temperatures over most of western Europe. Harsher winters were more frequent between the 9s and the 9s (see Figure 9 in Graham et al., 6). Even though mild winters similar to the recent ones did occur they alternated with colder winter seasons. The best remembered is probably the winter of 96/96 when temperatures were well below zero across most of Europe. Figure illustrates the unusual amounts of snow that occurred at lower elevations in Switzerland. With the exception of western Ireland, most of the British Isles was covered by a blanket of snow for most of the winter (Met Office, 6a). On an anecdotal note, the harsh conditions prevailing during that winter led a then relatively unknown Bob Dylan, who was staying in a London flat at that time, to burn furniture in order to keep warm (Williamson, ). On mainland Europe it was the only time in the twentieth century that large lakes such as Lake Constance or Lake Zurich were completely ice-covered and it was only in March that the ice finally broke up (Pfister, 999). Cold winters such as the one in 96/96 are generally associated with a negative NAO index characterized by weak westerlies (lower-than-normal pressure difference between Icelandic low and Azores high) and a reduced Atlantic influence in western Europe. The NAO index, and the late spring seasurface temperature pattern in the North Atlantic, can be used as indicators to forecast the conditions during the following winter (Graham et al., 6; Met Office, Figure. Impressions of winter 96/96 in Oberbalm (Schweizer Mittelland, Switzerland, photos courtesy of Ernst Hunziker). 6b). However, prediction of how much colder or warmer the winter season is going to be and in what regions the anomalies are likely to be most pronounced is subject to large uncertainties. A negative NAO index during the winter months does not always coincide with very cold winter conditions over the same regions. A recent example is the winter of /6. Despite the clearly negative NAO index the winter temperatures over most of Britain remained close to average values and were clearly higher than those experienced in 96/96. However, the winter /6 was the coldest winter since 96 in parts of central, eastern and southern Europe. This illustrates that even if the NAO index gives an indication of whether a winter is warmer or colder than normal, the spatial pattern of the largest temperature anomalies can vary. In this study, we use reanalysis data from the National Centers for Environmental Prediction and the National Center for Atmospheric Research (NCEP/NCAR, Kalnay, et al., 996) for surface temperature and sealevel pressure to compare the winter of 96/96 with other winters associated with a negative NAO index and very low temperatures over other regions of Eurasia. The purpose is to highlight how exceptional the winter of 96/96 was compared to other harsh winters in terms of amplitude and spatial extent of the cold temperature anomaly. The fact that the winter of 96/96 is so well remembered is partly due to the character of an average winter over most areas of Britain and western Europe; generally temperatures are above freezing and long spells of snow are rare. However, a shift in temperature by a few degc can transform the traditional green British winter into a snowy (white) one, thus making the difference even more obvious. In comparison, other areas of Europe/ Eurasia are characterised by average winters where snow and ice are common. Therefore, negative temperature anomalies are more likely to go unnoticed; a winter may be colder than normal but it does not change its colour. Cold winter anomalies Here we consider the spatial temperature anomaly pattern for the air temperature m above surface for particularly cold winters Weather February, Vol. 6, No.
Negative NAO and cold Eurasian winters Weather February, Vol. 6, No. during the period 96. The NCEP/NCAR reanalysis dataset used in this study is obtained from a numerical model into which observations have been assimilated (Kalnay et al., 996; Kalnay, ). Therefore, deviations between instrumental data and the reanalysis values are likely if values are compared at respective locations and times where instrumental data are sparse and the model might introduce features that don t reflect real climatic events (Waliser et al., 999). However, the surface temperature data that are mainly used in this study are likely to be one of the most reliable quantities: firstly a large number of surface temperature observations were included in the reanalysis; and secondly, temperature observations were subject to small observational errors. Eurasia The winter temperature anomalies for six cold winters during the 9 to 6 period covered by the NCEP/NCAR reanalysis dataset are illustrated in Figure. The values shown are differences between the mean temperature for December to February in a given winter (e.g. 96/96) and the mean winter temperature for all years between 9 and 6. In the second half of the th century four winters stand out as particularly cold in Eurasia: 9/96, 96/96, 96/969 and 9/9. Additionally, two recent cold anomalies are depicted for the winters 99/996 and /6. All these winters coincide with a negative NAO index but the spatial anomaly patterns for the temperature can be very different in each case, as shown in Figure : N S E S W N S N S E W E E /6 N N S E N 99/996 E E 9/9 W For the winter of 96/96 the largest temperature anomalies are centred over western Europe with values of degc in an area stretching from northern France to Poland and values of to degc over Britain. Mainly positive deviations are found southeast of a diagonal from western Africa to the northern Urals in Russia. N E N E N 96/969 W For the winter of 9/96 temperature anomalies below degc are found over western Russia and smaller anomalies extend into western Europe. In an area reaching from the Middle East to the Caspian Sea the winter of 9/96 was mainly warmer than average. 96/96 9/96 N W E N N S W Figure. Winter (Dec Jan Feb) temperature anomalies in degc. E E
Exceptionally cold conditions prevailed over much of Eurasia during the winter of 96/969. A vast area reaching from the Black Sea to eastern Siberia experienced temperature anomalies below degc compared to the 96 average. In central Siberia the values are lower than degc. Smaller temperature anomalies occurred in western Europe where the winter conditions were close to average in many areas. The winter of 9/9 affected an area of similar size as the winter of 96/969 but over Siberia the amplitude of the cold anomaly was less pronounced. The largest anomalies occurred in an area reaching from the Balkans to Scandinavia and into western Russia and Siberia. Especially over Finland and western Russia average winter temperatures were more than degc below normal. In 99/996 an area centred over Poland, northern Germany and southern Scandinavia was affected by temperature anomalies of about degc and only relatively small temperature anomalies extended into Russia. During that winter western Europe experienced small deviations ( to degc) compared to average conditions. With the exception of Scandinavia and most parts of Britain, the winter /6 was characterized by negative anomalies of to degc over Europe. The largest anomalies of degc were found in central Siberia. With anomalies of degc the Iberian Peninsula was colder during that winter than during the other winters discussed here. It is worth noting that during the winter of /6 exceptionally warm conditions prevailed over the Arctic. In contrast to 9/9, where the warm anomaly was confined to the Atlantic sector of the Arctic Ocean, a much larger area was abnormally warm in /6. North America Even though it is not the main focus of this article it is interesting to look at the temperature anomaly pattern over North America as well. One feature common to all winter temperature anomalies depicted in Figure is a positive temperature anomaly over Greenland and the Labrador Sea. This is the expected temperature anomaly during negative phases of the NAO (Thompson and Wallace, 99). Over the Labrador Sea warm anomalies are particularly pronounced for the winters 96/96 and 96/969 with values of more than degc. Over the North American continent itself there is no consistent pattern. The winter of 96/96 coincides with cold conditions over the eastern half of the continent while warmer than normal conditions prevailed along the west coast. A similar albeit westward-shifted pattern is found in 9/9 with a cold anomaly affecting the central areas and the west coast with the exception of Alaska where the temperature was warmer than normal. The winters of 96/969 and 99/996 are characterized by a negative anomaly along a diagonal from Alaska to the eastern USA and a warm anomaly in the western half of the USA. A similar pattern is seen for the winter of 9/96 but here the negative anomaly does not follow the entire diagonal from Alaska to the USA east coast and temperatures are above average over most of the USA. With the exception of the northwestern corner of the USA temperatures were well above average during the winter of /6 over the entire north American continent with the largest anomalies occurring over northern Canada. Absolute winter temperature minima As an indicator for how exceptional the winters discussed above are, it is useful to look at a map showing the years during which the lowest winter average temperatures were recorded between 9 and (Figure, top). The most striking Negative NAO and cold Eurasian winters Weather February, Vol. 6, No. 6 N N N A 96/96 B 9/9 C 9/96 D 96/969 N E E E E..... 6.. 9 96 96 96 9 99 Figure. Top: Coldest winters (Dec Jan Feb) between 9 and. The shaded areas indicate in which year the coldest winter temperatures were recorded. Bottom: Mean winter temperature anomalies for the years 9 to in the areas A D indicated in the top panel.
Negative NAO and cold Eurasian winters Weather February, Vol. 6, No. 6 feature here is how spatially coherent the patterns are. Between N and 6 N a large fraction of Eurasia ranging from the British Isles to eastern Siberia experienced the lowest winter average temperatures during one of the four pre-96 cold winters discussed before. The winter of 96/96 set record low temperatures over an area encompassing the British Isles, much of France, the BeNeLux states, Germany, Poland and Denmark. Adjacent to the eastern limit of the 96/96 area the lowest winter temperatures occurred during the winter of 9/9. Several countries in eastern Europe, Finland and parts of northwestern Russia experienced their coldest winter in that year. The winter of 9/96 caused with the lowest temperature in a small part of western Russia. The winter that set the record low temperature over the largest fraction of Eurasia was the winter of 96/969. A vast area reaching from the Urals to eastern Siberia experienced the coldest winter average temperatures since at least 9. The area most affected by this winter is more than half the size of North America and is larger than the areas linked to the winters of 96/96, 9/9 and 9/9 combined. In the following discussion, the areas where the lowest temperatures were recorded in 96/96, 9/9, 9/96 and 96/969 will be named with A, B, C and D, respectively (see Figure ). Looking at time-series of the average winter temperature for the areas A D from 99 to illustrates how cold the winter of 96/96 was in comparison to other cold winters (Figure, bottom). It also highlights if a winter anomaly is mainly reflected in one area or if it has an imprint on the temperatures in the other areas as well. The values for the temperature anomalies during the four pre-96 winters considered here are summarized for the regions A to D in Table I. For the winter season 96/96 the temperature anomaly was. degc in region A, which was by far the coldest winter season of the period considered here. The next coldest winters are those of 9/9 and 99/996 with anomalies of almost degc (Figure, bottom). Looking at the time-series for areas B to D reveals that temperature anomalies larger than that seen in region A in 96/96 are found for the other three very cold winters considered here. The temperature anomaly is.degc for the winter of 96/969 in region D and anomalies of. and. degc occurred during the winter seasons 9/96 and 9/9 in regions B and C, respectively. If the amplitude of the cold temperature anomalies in regions A D is considered, the winter of 96/96 is only ranked th between 9 and, Table I Standard deviation of temperature anomalies in the regions A D (see Figure ) compared to temperature anomalies for the winters of 9/96, 96/96, 96/969 and 9/9 over the same areas. Units are degc. Region C A D B Standard deviation.... anomaly 9/96.... anomaly 96/96 -.... anomaly 96/969....6 anomaly 9/9...6. the winter of 9/99 showing an anomaly of less than. degc in region C as well (Figure, bottom). The regions B and C form the transition between the two larger areas A and D and for the four winters discussed here, cold anomalies in either A or D are reflected in B and C as well. The peculiarity of the winter of 96/96 is not so much the amplitude of the negative temperature deviation but its westerly location. With an average value of. degc it was large enough to change the character of that winter in countries that normally experience temperatures above freezing. Compared to the anomaly of. degc seen during 96/969 over the vast region D, the winter of 96/96 actually appears rather modest. However, a large fraction of Region A is found over sea where temperature extremes are strongly reduced. This maritime influence is clearly reflected in the standard deviation of the temperature anomalies over region A, which is much smaller than that seen in regions B D (Table I). In regions B, C and D the standard deviation is up to. times larger. In analogy to region A, this can be explained by the more continental climate of these regions that favours temperature extremes. In 96/96 the anomaly of. degc is. times the standard deviation of degc in region A. In comparison, a ratio of. is found between the standard deviation in region D and the winter temperature anomaly for the winter of 96/969. This is truly remarkable considering the size of region D and the expected smoothing of extremes if averages are considered over larger areas. Pressure anomalies One intriguing feature is the easterly location of the maximum temperature anomalies during the winter of 96/969. The NAO index was negative and usually the largest imprint is expected to be found over western Europe, Scandinavia and the western half of Russia. However, in this case the largest temperature anomaly was most pronounced from the Urals to eastern Siberia (except the easternmost regions of Eurasia which remained unaffected). The anomaly pattern for the sea-level pressure is depicted in Figure. As expected from the negative NAO index, the winters of 9/96, 96/96, 96/969 and 9/9 coincided with higher-thannormal pressure at high latitudes and lower pressure at lower latitudes. However the exact pattern is different in each case. In 9/96, higher-than-normal pressure prevailed over northwestern Russia while pressure was lower over an area reaching from the western Mediterranean to central Asia. These pressure anomalies favoured the advection of cold continental air masses into western Russia where the largest anomalies occurred during that winter. A similar pattern that is shifted westward characterizes the winter of 96/96. There is a pronounced pressure anomaly dipole with higher than normal pressure over Iceland and lower than normal pressure over the Iberian Peninsula. This clearly favoured the outbreaks of cold continental air masses into western Europe that were typical for that winter. The sea-level pressure anomaly pattern for the winter of 96/969 is characterized by a more meridional pattern. A westward extension of the Siberian High led to a ridge of higher-thannormal pressure from northwestern Russia to the Caspian Sea, whereas pressures were lower than normal over the Iberian Peninsula and over northern China and Mongolia. The high-pressure ridge over Russia favoured the migration of arctic air masses into Siberia down to the Caspian Sea where subsequent radiative cooling led to the extreme cold anomaly. At the same time, the massive Siberian High effectively shielded off large parts of Russia from any maritime influence of Atlantic origin. On the other hand, the pressure difference between the western flank of the highpressure anomaly and the lower-pressure anomaly over southwestern Europe favoured the advection of air masses of southeastern origins into western Europe leading to the relatively small temperature anomalies observed in that winter. Finally, the winter of 9/9 was characterized by higher-than-normal pressures over northern Scandinavia. As in 9/96 this allowed cold continental air masses to move
6 N N N N 6 N N N N 6 N N N N 6 N N N N E E E E Figure. December to February sea-level pressure anomalies for the winters 9/96, 96/96, 96/969 and 9/9. Units are hpa and the contour interval is hpa. westward leading to the lowest temperatures in parts of eastern Europe and Scandinavia since at least 9. The winters of 96/969 and 9/9 coincided with a positive index of the Scandinavia pattern whereas the index was close to neutral in 9/96 and in 96/96 (NOAA, 6). A positive index for the Scandinavia pattern is normally associated with cold anomalies over central Russia and western Europe. Interestingly, this is in contrast to the temperature anomalies seen for the winter of 96/969. Despite the extreme cold anomaly over Siberia the temperature over western Europe was close to average (Figure ). It is also worth noting that the pressure anomaly during the winter of 96/969 was the peak of a phase of several consecutive years when the Siberian High was particularly strong between December and February (Panagiotopoulos et al., ). Monthly anomalies 6 6 6 6 6 6 6 6 The winter temperature anomalies discussed in the previous sections can result not only from a series of cold months but also from single, very cold months. Monthly temperature anomalies averaged over the regions C, A, D and B (Figure ) are depicted in Figure for the years 9/96, 96/96, 96/969 and 9/9, respectively. In 9/96, the low winter temperature resulted from a very cold December and February with anomalies of 9 degc and degc respectively. In January the temperature was only slightly below average. The cold anomaly in February 96 was not confined to region C but was felt in large parts of Europe, where it is remembered as one of the coldest months on record (e.g. Pfister, 999). In 96/96, all three winter months show monthly temperature anomalies below degc with the largest anomalies in January and February. The very cold winter of 96/969 is mainly a consequence of a temperature anomaly of 9 degc in January and February but, with a deviation of degc, December 96 was cold as well. Finally, in 9/9 low temperatures occurred in January and February. As in February 96, the cold January 9 was felt over large parts of western Europe. An interesting feature that can be seen in Figure is the duration of the cold anomalies in the different years. The amplitude and spatial extent of the cold anomalies of 96/969 stand out as exceptional, and as illustrated in Figure this winter was the peak of a prolonged cold phase. During five consecutive months (November 96 to March 969), the temperature was at least degc below average. In comparison, this duration is only three months in 96/96 and two months in 9/9. In 9/96, pronounced monthly anomalies (December, February) alternate with deviations of clearly less than degc (January, March). Conclusions The large scale patterns and fluctuations discussed here are unlikely to result from the data assimilation process, especially since we focus on areas with good data coverage, particularly over western Europe. The winter of 96/969 is not well documented. However, checking synoptic weather maps confirms the presence of an extreme cold anomaly in that year (Weather, 969). Of course, the smaller number of observations available during the first part of the 9 to period also means that the values associated with earlier events are subject to larger uncertainties than recent events. Nevertheless, looking at the dataset of Brohan et al., (6), which compiles observational data from to present also confirms that the temperature anomaly that occurred during the winter of 96/969 over Siberia was unique not only since 9 but probably since at least. Our study indicates that four very cold Eurasian winters (9/96, 96/96, 96/969 and 9/9) have set the lowest winter (December to February) temperatures recorded since 9 over a large part of Eurasia reaching from western Europe to eastern Siberia. Several winters over Eurasia were linked to more pronounced cold anomalies that affected much larger areas than the winter of 96/96. In terms of amplitude of the temperature anomaly the winter of 96/96 is only ranked th compared to other Eurasian winters during the 9 to period. The most extreme winter appears to be the one experienced in 96/969 with winter temperatures. degc below average over a vast area reaching from the Caspian Sea region to the Far East. The occurrence of this impressive anomaly was linked to the presence of a meridional high-pressure Negative NAO and cold Eurasian winters Weather February, Vol. 6, No.
Negative NAO and cold Eurasian winters Weather February, Vol. 6, No. Figure. Monthly temperature anomalies for the years 9/96, 96/96, 96/969 and 9/9, respectively. The letters in brackets (A, B, C and D, see Figure ) indicate over which regions the anomalies are calculated for the different years. anomaly ridge over western Russia that blocked off the Atlantic influence. In comparison the temperature anomaly over the area most affected by the winter of 96/96 was only. degc with maximum deviations of degc over several regions of western Europe. However, it has to be said that the temperature anomaly is likely to have been dampened by the maritime influence in this area. What makes the winter of 96/96 extraordinary is the westerly location of the cold temperature anomaly. It is interesting to note that from early 96 to early 96 sea surface temperatures were lower than normal around the British Isles (Sinha and Topliss, 6). The related cold air temperature anomaly combined with a negative NAO index in the winter of 96/96 might help to explain why the largest winter temperature anomalies occurred so far west during that year. Despite coinciding with a negative NAO index, the other harsh winters had the largest temperature anomalies over eastern Europe and Russia. The fact that the massive winter of 96/969 is much less documented than the winter of 96/96 is likely to be due to the fact that an anomaly of. degc is large enough to bring snowy and icy conditions in areas that normally experience winter temperatures above freezing. On the other hand, parts of the area most affected by the winter of 96/969 have winter mean temperatures of C or lower so that even an anomaly of. degc did not really change the character of the winter compared to an average one. With five consecutive months with temperature anomalies below degc (November 96 to March 969) the duration of the cold spell was exceptional as well. However, the long Siberian winter means that the coldest anomalies did not affect autumn 96 or spring 969 and therefore the duration of the winter of 96/969 must have been close to average. Acknowledgments We wish to thank NCAR/NCEP for making their reanalysis dataset available. We also thank two anonymous reviewers for their helpful comments. JH is funded through the NERC RAPID Climate Change Programme. References Brohan P, Kennedy JJ, Haris SFB, Tett I and Jones PD. 6. Uncertainty estimates in regional and global observed temperature changes: a new dataset from. J. Geophys. Res. : D6, doi:.9/jd6. Graham RJ, Gordon C, Huddleston MR, Davey M, Norton W, Colman A, Scaife AA, Brookshaw A, Ingleby B, McLean P, Cusack S, McCallum E, Elliot W, Groves K, Cotgrove D and Robinson D. 6. The /6 winter in Europe and the United Kingdom: Part How the MetOffice forecast was produced and communicated. Weather, 6: 6. Hurrell JW. 99. Decadal trends in the North Atlantic oscillation: regional temperatures and precipitation. Science. 69: 66 69. Kalnay E.. Atmospheric Modelling, Data Assimilation and Predictability. Cambridge University Press. pp. Kalnay E, Kanamitsu R, Kistler R, Collins W, Gandin L, Iredell YM, Saha S, White G, Wollen J, Zhu Y, Chelliah M, Janowiak EbW, Ropelewski CR and Jenne R. 996. The NCEP/NCAR reanalysis project. Bull. Amer. Meteor. Soc. : 9. MetOffice. 6a. Winter chills: 9 and 96. http://www.metoffice.com/ education/secondary/students/winter. html MetOffice. 6b. Winter outlook. http://www.metoffice.gov.uk/corporate/ pressoffice/6/pr6.html NOAA. 6. Northern hemisphere teleconnection patterns. Climate Prediction Center, http://www.cpc. ncep.noaa.gov/data/teledoc/scand.shtml Osborn TJ. 6. Recent variations in the winter North Atlantic Oscillation. Weather. 6:. Panagiotopoulos F, Shahgedanova M, Abdelwaheb H and Stephenson DB.. Observed trends and teleconnections of the Siberian High: a recently declining center of action. Journal of Climate. :. Pfister C. 999. Wetternachhersage. Jahre Klimavariationen und Naturkatastrophen (96-99). Verlag Paul Haupt. pp. Sinha B and Topliss B. 6. A description of interdecadal time-scale propagating North Atlantic sea surface temperature and their effect on winter European climate, 9. Journal of Climate. 9: 69. Thompson DWJ and Wallace JM. 99. The Arctic Oscillation signature in the wintertime geopotential height and temperature fields. Geophys. Res. Let. : 9. Waliser DE, Shi Z, Lanzante R and Oort AH. 999. The Hadley circulation: assessing NCEP/NCAR reanalysis and sparse in-situ estimates. Climate Dyn. : 9. Weather. 969. Weather logs for December 96, January 969 and February 969. Royal Meteorological Society. Williamson N.. The Rough Guide to Bob Dylan. Rough Guides Ltd. p.. Corresponding author: Dr Joël J.-M. Hirschi National Oceanography Centre, University of Southampton European Way, Southampton SO ZH, United Kingdom Email: jjmh@noc.soton.ac.uk doi:./wea.