The Siberian High (SH) is a dominant circulation system over the Eurasian continent during winter. Its climatological mean central pressure is the highest in the Northern Hemisphere (NH), exceeding 1030 hPa (Takaya and Nakamura, 2005). Severe winters are manifested by extremely low temperatures over Central Asia due to a continuous radiative cooling associated with the coldest air masses centred usually over northern Mongolia. The SH influence, characterized by excessively low surface temperatures, affects regions extending well beyond its source area (Panagiotopoulos et al., 2005). Furthermore, fluctuations in the SH atmospheric circulation modulate the East Asian monsoon in wintertime (Jhun and Lee, 2004). This monsoon is reflected as a persistent northerly flow over East Asia, on the edges of the eastern flank of the SH and the western flank of the Aleutian low to the east (Zhang et al., 1997). The build-up of this high-pressure system results from three essential factors: upper level convergence, caused by dynamic and thermodynamic factors, continuous cold air advection in the upper troposphere in tandem with surface radiative cooling (Ding and Krishnamurti, 1987; Ding, 1990). This accumulation of anomalous cold air at the surface was found to be one of the essential preconditions for the strong amplification of this thermal high (Takaya and Nakamura, 2005). Severe winter weather associated with SH bitter cold air outbreaks, called dzuds often brings serious damages to agriculture and major disruptions in economic activities due to livestock mortality in Mongolia (Lau and Lau, 1984; Begzsuren et al., 2004).
The purpose of this study is to examine the synoptic atmospheric conditions that favour the initiation and evolution of this high-pressure system on several temporal scales.
The article is structured as follows: a description of the data and methodology, including the domain and the atmospheric layer representing the SH is given after the introduction. In Section 3, the synoptic climatological features of the SH are presented, accompanied by a description of historical trends in the SH intensity as manifested by minimum surface temperatures. Section 4 presents the results of an analysis of large-scale circulation patterns favourable for the SH build-up, to assess how its variability is related to climatic fluctuations over other places in the NH. In Section 5, spatial correlations between the SH intensity and various atmospheric variables are presented. Atmospheric conditions featuring the most extreme SH winter and spell are described in Sections 6 and 7, respectively, followed by a conclusion section.
2. Data and methodology
The data set used in this study is based on ground station observations of daily minimum temperatures during the winter season, in the domain between 40–60°N and 80–120°E.
As gridded surface temperatures derived from various models are affected by the model treatment, observations of daily minimum temperatures, obtained from the Royal Netherlands Meteorological Institute (Koninkiljk Nederlands Meteorologisch Instituut, 2011), available at (http://climexp.knmi.nl/selectdailyseries.cgi), were preferred over reanalysis or model data.
The domain was set along the definitions used by previous studies. However, there is no consensus on the boundaries of this area. Panagiotopoulos et al. (2005) define the SH as confined between 40–65°N and 80–120°E, while Gong and Ho (2001) restrict its northern boundary to 60°N but extend its western flank to 70°E. In this study, the overlapping area between these domains (i.e. 40–60°N and 80–120°E), used also by Wu and Wang (2002b), Wu et al. (2006) and Jhun and Lee (2004), is considered to best represent the SH core. Figure 1(a) shows that the chosen domain indeed captures well the geographical location of the SH activity.
Since the SH acquires its maximum intensity during the winter, all data were extracted for the months of December through February (DJF) as in the study by Wu and Wang (2002b), Ding and Krishnamurti (1987) and Panagiotopoulos et al. (2005).
There are three main reasons for focusing on minimum temperatures rather than sea-level pressure (SLP) or other atmospheric parameters for representing the SH intensity. First, when characterizing trends and, more importantly, extremely low temperature episodes generated by this dominant cold anticyclone, minimum temperatures are the most relevant observed variable determining the impacts of this synoptic system (e.g. Begzsuren et al., 2004). Second, the use of this parameter avoids possible artifacts caused by the extrapolation of surface pressure to sea-level values (Panagiotopoulos et al., 2005). Third, anomalous cold air precedes the strong amplification of the high, and serves as an indicator to identify the build-up of high pressure over central Asia (Takaya and Nakamura, 2005). Furthermore, Gong and Ho (2001) found a strong coupling between the SH and surface temperature in which the centre of cooling is co-located with the high's centre.
Another criterion in sorting the data relates to the altitude of the stations in the domain. Since the SH is of a thermal nature, implying a small vertical depth (Panagiotopoulos et al., 2005), only stations located in an altitude that does not exceed 700 m a.s.l. were used for the analyses. In addition, these stations are regarded as ‘ground stations’ by the World Meteorological Organization (2011; see Publication No. 9, Vol. A, available at: http://www.wmo.int/pages/prog/www/ois/volume-a/vola-home.htm), and as such report the geopotential height at 1000 hPa as the first standard level.
The daily minimum temperature data obtained were trimmed for each station so that only the longest consecutive period for which data were complete was considered. Unfortunately, data for most stations were available only to the late 1990s. Finally, the longest overlapping record for which the data were available for a substantial number of stations was found. The application of the various criteria (i.e. domain boundaries, maximum altitude and data availability) yielded a period of 60 years, extending from the winter of 1937–1938 to that of 1996–1997 with a total of 11 stations. Table I introduces these stations.
Table I. Stations names, locations (decimal degrees) and altitudes
Altitude (m, a.s.l.)
3. Climatological setting and trends of the SH
In this section, the general climatology of the SH is presented in two different manners. The first part of the section introduces the SH's synoptic climatology along three standard tropospheric levels—the geopotential height at 1000 hPa (GPH1000), which illustrates well the spatial extent of the SH near the surface, the geopotential height at 500 hPa (GPH500) showing the location of convergence and divergence zones caused by upper level circulation, and the zonal wind component at 200 hPa (U200), indicating the jet stream position. The synoptic charts were derived from the National Centers for Environmental Prediction and the National Center for Atmospheric Research (NCEP/NCAR) Reanalysis website (http://dss.ucar.edu/pub/reanalysis/) along a model developed by Kalnay et al. (1996) and Kistler et al. (2001). For purposes of orientation, the SH domain is marked by a rectangle in all charts shown in this article.
The second part of this section illustrates the trends of the SH observed for minimum temperatures for the 60 year period on which this study focuses.
3.1. Synoptic climatology
As the long-term mean (LTM) chart shown in Figure 1(a) indicates, the SH is the most dominant synoptic pattern over Asia during the winter, with a central core exceeding a LTM GPH1000 of 270 m located over western Mongolia. The SH boundaries stretch from the Caspian Sea in the west to the Pacific Ocean in the east, and from the northern most part of the Eurasian continent to the South China Sea in the south. This thermal high is bounded near the surface by two other large-scale synoptic systems: the eastern flank of the Icelandic Low (IL) to the northwest and the Aleutian low to the east. The strong horizontal pressure gradient formed between the SH and the latter generates the monsoonal northerlies characterizing the East Asian winter monsoon.
Figure 1(b) depicts the LTM of the GPH500 level. Evidently, the domain is found westward of the upper air trough located over East Asia, implying air convergence at mid-tropospheric layers.
Further support aloft is supplied by the East Asian jet stream. The convergence zone in the entrance of this jet is located over the southern edge of the SH domain. The climatological position of this upper tropospheric circulation feature is shown in Figure 1(c).
3.2. Trends in SH intensity
The most salient feature of the SH is its extreme cold associated weather. To illustrate the inter-annual variability in SH minimum temperatures over the domain and detect possible trends, daily records from all 11 stations were averaged for the 3 months (DJF) to produce a single figure representing the severity of each winter. Then, annual standardized anomalies were calculated by subtracting the averaged 60 values from each year's value and dividing it by the 60 years' standard deviation. Additionally, a 5 year centred moving average has been applied to detect pronounced trends within the data set. Results of these analyses are shown in Figure 2.
Clearly, the SH is undergoing a phase of warming. Moreover, the warming over the domain has accelerated during the second half of the period, with a rate considered to exceed warming at any other location on the globe (Jones et al., 1999; Houghton et al., 2001). These arguments are supported by a few findings. First, the continuously rising moving average trend during the second half of the period implies that not only warming is taking place, but it is stable and exhibits only minor fluctuations. Second, the linear trend calculated for this period is steeper than that which encompasses the entire 60 years, showing an average increase of 0.18 °C/annum with comparison to 0.07 °C/annum from 1937–1938 through 1996–1997 (both trends are significant at the 99% confidence level). Lastly, of the standardized anomalies calculated for the second half of the period, only nine are negative, with all the last 12 years being positive, indicating milder winters during this period. These findings are supported by the study of Sahsamanoglou et al. (1991) and Panagiotopoulos et al. (2005), which found a significant decline in SLP values over Siberia since the late 1970s.
4. Correlations between SH intensity and standard oscillations
Despite the dominance and large spatial extent of the SH, its non-local effects on climate and associations with other oscillatory modes in the NH are an area of limited knowledge (Panagiotopoulos et al., 2005). Furthermore, previous papers (Wu and Wang, 2002a; Panagiotopoulos et al., 2005) yielded contradictory evidence regarding the association of the SH with other climatic variations in the NH, and whether these relationships are substantial.
The purpose of Chapters 4 and 5 is to address this issue and examine the relation between the SH intensity as defined by surface minima temperatures to NH standard oscillations and to various atmospheric parameters.
Before turning to describe the methodology in use, a terminological reference has to be made. In this section, the term ‘oscillation’ was preferred over ‘teleconnection’, even though both are used synonymously in meteorology (Felix et al., 2010). ‘Oscillation’ solely describes large-scale seesaw patterns in shallow tropospheric layers between confined regions, as compared with teleconnection, which is often used whenever a linkage between weather changes occur over widely separated regions of the globe (Glickman, 2000). Therefore, standard oscillations, as described further on, are a more appropriate term to describe their pressure fluctuations against the SH intensity.
Establishing relationships between pressure patterns or other atmospheric variables is commonly done by examining the correlation between time series. To analyse the relationship between the SH intensity and standard oscillations, monthly standardized anomalies were derived along the same method described in Section 3.2. The SH standardized anomalies were correlated with monthly (DJF) indices of various NH standard oscillation indices and the El-Nino Southern Oscillation (as Nino 3.4) index for the period of 1950 to 1997 (years of available data for the NH indices), resulting in an overall 143 monthly values. NH standard oscillations include the North Atlantic Oscillation, East Atlantic Pattern, East Pacific-North Pacific Pattern, Pacific-North American Pattern, East Atlantic-West Russia Pattern, Pacific Transition Pattern, Scandinavia Pattern, North Pacific Index, Arctic Oscillation (AO), Pacific Decadal Oscillation and the North Sea-Caspian Pattern.
The results obtained in the correlation analysis indicate that the SH intensity is not strongly associated with most of the oscillations examined in this study. Among all the 12 relationships, the correlation of the SH with the AO is the highest, by far exceeding other correlations at R = 0.46 (p value < 0.0001). Of a resembling nature to the AO (Wu and Wang, 2002a), the second strongest correlation of the SH is with the North Atlantic Oscillation with R = 0.26 (pvalue = 0.002).
Since the AO has the strongest association with the SH, an elaboration of the AO index is given. The monthly AO index is constructed by projecting the monthly mean GPH1000 height anomalies poleward to 20°N onto the leading Empirical Orthogonal Function (EOF) mode of these heights, shown in Figure 3, in a spatial resolution of 2.5°(lat) by 2.5°(long). The time series of these anomalies is then normalized by the standard deviation of the index (see http://www.cpc.ncep.noaa.gov/products/precip/CWlink/daily_ao_index/history/method.shtml).
The AO, first identified by Lorenz (1951) and further elaborated by Thompson and Wallace (1998), exhibits a negative phase characterized by an intensified polar high accompanied by relatively shallow mid-latitude cyclones. In the positive phase, the polar high is weak and confined to the polar region, enabling the penetration of warm and moist air driven by the intensified jet stream on the edge of mid-latitude cyclones in their deep mode.
The annual correlation coefficient between the SH minimum temperature anomalies and the AO index for the 60 year record is 0.53 (p value < 0.0001), implying warmer temperatures over Siberia during a positive AO. As indicated by Gong and Ho (2001), the generally positive phase of the AO and the weaker SH (as manifested by SLP) observed since the middle of the 1970s have yielded high correlations. The annual anomalies based on minima temperatures reaffirm this finding: the correlation between the AO and the SH for the first half of the period (i.e. 1937–1938 to 1966–1967) is insignificant, while the correlation for the second half (1967–1968 to 1996–1997) reaches 0.71 (p value < 0.0001). Although the trend featured by the AO and the SH can affect the results obtained in the correlation analysis, the series were not de-trended because the assumption that the trend of both systems results from the effect of a common external factor is not met.
A possible explanation to the positive correlation found between the AO and the SH is as follows: the intensified westerly zonal wind at mid-latitudes and the weak polar high during the positive phase of the AO lead to favourable conditions in which warm and moist air is driven further into Eurasia, thus contributing to milder winters over Siberia. When the leading mode of the AO is negative, advection of warm air to Siberia is limited, thus enabling the amplification of the SH. In addition, a strong SH blocks the passage of extratropical migrating cyclones crossing Europe, thereby reducing warm advection over its source area (Panagiotopoulos et al., 2005).
5. Spatial correlations of the SH intensity with synoptic fields
An additional way to explain the variability in SH intensity is by examining its spatial correlations with various synoptic fields. For this sake, the SH monthly minimum temperature standardized anomalies were spatially correlated with variables such as the GPH and temperature at 1000 hPa, the GPH500 and U200. The synoptic fields used in our analyses were extracted from NCEP/NCAR Reanalysis archive (Kalnay et al., 1996; Kistler et al., 2001), available at http://www.esrl.noaa.gov/psd/data/correlation/. These analyses are based on the period of 1948 to 1997, as data preceding 1948 are not available at this site. To provide insights that can be attributed to processes that encompass the entire globe, the analyses are not restricted to the NH.
Starting with the lowest atmospheric level indicative of atmospheric circulation, the GPH1000 shows two centres of strong correlations with the SH minimum temperatures (Figure 5(a))—a positive correlation over the tropical Pacific Ocean (R ≥ 0.7) and a negative correlation northwest to the SH domain (|R | ≥ − 0.8).
Both these centres of correlations imply that a weakening of the equatorial low and a deepening of the eastern flank of the IL are associated with a warmer SH. Although the association between the SH and the equatorial low is difficult to interpret, it emphasizes that the SH is indeed linked with other circulation patterns far beyond its immediate domain.
The strong negative correlation northwest to the domain can be interpreted as an IL in its deepest mode. This mode contributes to an increased frequency of intense Atlantic cyclones crossing Siberia, which lead to warm air masses advection over this region, as suggested by Rogers and Mosley-Thompson (1995).
The negative correlation (|R | ≥ − 0.5) between the temperature at 1000 hPa over the southern tip of Greenland and SH anomalies (Figure 5(b)) indicates that a colder core of the IL is associated with warmer temperatures over Siberia during the winter. This is consistent with a deepening of the eastern flank of the IL, mentioned above.
Furthermore, the drop in GPH500 over the eastern part of the IL (Figure 5(c)), deepening this cyclone at the surface and associated with a milder SH, is manifested by a strong negative correlation (|R | ≥ − 0.7). A resembling pattern was also reflected at GPH700 (not shown).
At higher tropospheric layers, the strong positive correlation (R ≥ 0.7) observed between U200 just north to the domain and the SH minimum temperature (Figure 5(d)) implies that a strengthening of the wind there is associated with less severe weather over Siberia. Noteworthy, a decline in the intensity of the subtropical jets over both hemispheres is also observed during milder winters over the domain, when the reduced longitudinal thermal gradient results in weak thermal winds, as supported by Panagiotopoulos et al. (2005).
6. Extreme SH
Extreme cold events are a source of major concern for the inhabitants of Siberia (Lau and Lau, 1984; Begzsuren et al., 2004). The purpose of the following two sections is to address such discrete events, by analysing the synoptic scale processes that lead to their formation, and to point at their anomaly as derived from climatological LTM values, shown in Section 3.
The analyses focus on the most extreme events generated in two different temporal scales. Section 6.1 deals with the seasonal scale, analysing the coldest winter during the 60 year record. Section 6.2 analyses the coldest spell found in the data.
6.1. The extreme winter of 1968–1969
As shown in Figure 4, the 1968–1969 winter is clearly the coldest one in the record. The significant correlation found between the AO and the SH is apparent by the lowest AO annual index value during the 60 year period, coinciding with the coldest SH winter observed.
The circulation pattern of this exceptional frigid winter is shown for GPH1000 and GPH500. The position of the jet stream as displayed by U200 was examined but is not shown.
The positive GPH500 anomaly over the Barents Sea (Figure 6(d)) is demonstrated by a weakening of the trough at mid-tropospheric layers. This weakening is further manifested by a shallow IL with ∼100 m above its LTM depth (Figure 6(c)). The retreat of the IL allows a westward expansion of the associated SH cold core (Figure 6(a)), accompanied by a shallowing of the ridge at mid-tropospheric layers (Figure 6(b)). At higher altitudes, a second streak of the sub-tropical jet is formed southwest to the domain.
6.2. The cold spell of 21–30 January 1969
Extreme cold spells are expected to feature even more severe conditions than those generated by an extremely cold season. Moreover, due to their smaller temporal scale, it is easier to draw valuable insights on the evolution of such spells. To detect extreme spells, a methodology that takes into account all winter days for the 60 year record (5,415 days per station) was implemented.
First, only the coldest 5% (271 days in each station) were used for the analysis. Then, an arbitrary threshold set on a minimum of 5 consecutive days was used to filter extreme cold spells in each station from the remaining days. To distinguish between spells generated by large-scale synoptic processes from those of a more local nature, two additional stages were required. Assuring that cold spells are of a regional extent was achieved by scanning the lists of individual spells for overlapping dates in at least 6 of the 11 stations. The new list was then scanned again under the threshold of 5 consecutive days used earlier, to generate the final list of SH cold spells.
Using this graded method guarantees a very robust threshold, where regional spells necessarily comprise common individual spells recorded in the majority of the stations, not merely inconsecutive overlapping dates. This methodology yielded an overall three regional extreme cold spells: 15–23 January 1947, 7–11 January 1951 and 21–30 January 1969. The last spell was chosen as the most extreme due to its longer duration and, more importantly, due to its extensive spatial coverage. This spell affected all stations in the domain, whereas the first and second spells were observed in only nine and six stations, respectively.
During the extreme spell, the average minimum temperature of all 11 stations was − 41.6 °C, ranging between − 24.6 °C (Zajsan station) and − 53.1 °C (Vitim station), both recorded on 23 January 1969. Figure 7 illustrates the observed daily minimum temperatures averaged over the domain during the respective winter. The running mean was set to average 10 days, as the length of the spell, highlighting its severity.
The evolution of the extreme spell was analysed by examining composite charts of the standard atmospheric levels used in the previous section. For this sake, successive 10 day composites were compared in 1 day increments, such that the first composite reflects the conditions preceding the event (i.e. 11–20 January 1969) and the last representing the days following the spell (i.e. 31 January to 9 February 1969). Consequently, 21 charts were scrutinized for each atmospheric level.
Of the 21 charts, four composites are presented for GPH1000, representing the four stages that best reflect the temporal evolution of the event at the surface: the period preceding the spell, the beginning of the spell—covering its first half, the spell centre and finally, the 10 days following the spell.
Although previous research suggests that local processes such as diabatic cooling as well as upper level air flow convergence contribute to a rapid intensification of the SH and, in turn, to a cold outbreak (Ding and Krishnamurti, 1987; Ding, 1990), Figure 8(a)–(d) demonstrates that the factors that lead to this extreme spell originate far from the SH domain. However, it should be noted that the processes that led to the outbreak of this cold spell are not indicative of a common mechanism, as the examination of the extreme events of 1947 and 1951 revealed that these were generated by the processes mentioned above.
Before the spell, the GPH1000 is higher at the North Pole than over Siberia. This pattern indicates anomalous conditions, since the coldest air masses during the boreal winter are most commonly located over northern Eurasia. At this stage, an extension of the IL in its deep mode is observed over northern latitudes, preventing the supply of extremely cold air masses from polar regions to the domain. The SH is not yet exposed to the frigid air and maintains a weak core over western Mongolia (Figure 8(a)).
In the first stage of the event (Figure 8(b)), a weakening of the eastern flank of the IL enables the onset of polar air advection to mid-latitudes. This bitter cold air accumulates over northwestern Russia.
Figure 8(c), representing the centre of the event, shows a weakening of the IL and its eastern flank in tandem with a westward migration, which further reduces the chances of warm advection to Siberia. This weakening is explained by unfavourable upper air support as indicated by the ridge located over northwestern Europe (Figure 9(a)). Noteworthy, during this phase, the subtropical jet stream is divided into two separate streaks (not shown). At this stage, the polar air mass progresses eastwards, leading to the build-up of the SH in its pronounced mode, resulting in a GPH1000 of 300 m, 30 m higher than its LTM, equivalent to two standard deviations.
The termination of this episode is marked by a deepening of the sub-polar low aloft (Figure 9(b)), enabling a recurring intensification of the IL at the surface as observed before the event (Figure 8(d)). The intensification of the IL leads to the generation of an easterly low pressure belt with two cyclonic centres (marked by a dashed arrow in Figure 8(d)), which enhances the isolation of the extremely cold air pool, pushing it further south and leading to the dissipation of the SH over the domain.
The important role of the IL in modifying temperatures over Siberia is consistent with the study by Rogers and Mosley-Thompson (1995), who based their study on monthly time averaged global gridded point data rather than on ground observations daily records.
Contradictory to the IL, the Aleutian low was found as non-active in modulating the SH activity. This is reflected in the sections analysing the SH correlations with standard oscillations, its spatial correlations with various synoptic fields and its extreme modes on short and long time spans.
A comprehensive analysis of daily surface minimum temperature observations from 11 stations, carried out in inner Eurasia for the winters of 1937–1938 to 1996–1997 enabled to shed light on recent trends in SH intensity and global and synoptic scale factors associated with its fluctuations. Evidently, global warming is considerable over Siberia, accelerating in recent periods. The linear trends augmented from 0.07 °C/annum for the whole 60 year period to 0.18 °C/annum for the second half of the period.
Over the 60 year record, three regional extreme cold spells were identified. To address the mechanism of the SH build-up and dissipation, the most severe cold spell over these 60 years was analysed.
The IL was revealed as the most important synoptic scale system modulating the SH activity as supported by the following:
•A strong positive correlation with the AO, implying a deep IL during mild SH winters.
•Strong spatial correlation between the IL and the SH temperature, indicating warmer SH during the deep mode of the IL.
•In the extreme spell analysed, besides the warm advection associated with the IL, its location and depth have shown an ability to control the supply of air advected from polar regions to the SH domain.
The first author gratefully acknowledges additional funding given by the Amiran Fund, Hebrew University of Jerusalem. The authors would like to thank Tamar Sofer for her assistance in the preparation of the figures and Baruch Ziv for his insightful comments on earlier versions of this article. We would like to thank two anonymous reviewers for their helpful suggestions and comments.