Abstract

Long-term records of winter ice duration, formation, and breakup dates (1869–1996) and maximum thickness (1950–95) on Lake Baikal are analyzed to determine the nature of temporal trends and the relationship with the large-scale atmospheric circulation. There are highly significant trends of decreasing ice duration (and thickness) over the period, associated with later ice formation and earlier breakup dates. These trends are broadly in line with those of winter air temperatures in the region. Variability in Lake Baikal ice formation date, duration, and thickness is significantly related to winter temperatures over a wide area from the Caspian Sea to the Pacific and from northern India to the Kara Sea off the northern coast of Siberia. Thus, Lake Baikal ice cover is a robust indicator of continental-scale winter climate. Correlation and composite analysis of surface and upper-atmospheric fields reveal that interannual variability in ice cover is associated with a tripolar pattern of upper-level geopotential height anomalies. In years of high (low) ice duration and thickness, significant positive (negative) 700-hPa geopotential height anomalies occur over northern Siberia and the Arctic, complemented by negative (positive) anomalies over central-eastern Asia and southern Europe. This structure induces an anomalous meridional flow regime in eastern Siberia with cold (warm) temperature advection from the northeast (southwest) in years of high (low) ice duration and thickness. Analysis of the lower-tropospheric heat budget during years of extreme early and late ice onset indicates that horizontal temperature advection is largely responsible for the observed temperature anomalies. These circulation anomalies are associated with certain recognized patterns of Northern Hemisphere climate variability, notably the Scandinavian and Arctic Oscillation patterns. Significant correlations also occur between Lake Baikal ice cover and the Pacific–North American pattern in the previous winter. The component of variability in Lake Baikal ice cover unrelated to these modes of Northern Hemisphere climate variability is associated with the position and intensity of the Siberian high.

1. Introduction

Lake Baikal, situated in southeast Siberia, is one of the world's more unusual freshwater ecosystems. Estimated to be over 25 million years old, Lake Baikal is also the world's deepest (maximum depth 1642 m) and largest lake, in terms of water volume (23 015 km3), containing some 20% of the world's surface freshwater (Kozhov 1963). Lake Baikal is distinct from other deep lakes in that its entire water column is saturated with oxygen throughout due to regular overturning every spring and autumn (Weiss et al. 1991; Shimaraev et al. 1994). This oxygenation supports an extensive, and almost wholly endemic, deep-water fauna (Fryer 1991) and is one of the reasons Lake Baikal was declared a World Heritage Site in 1996.

Lake Baikal is an important location for paleoclimate studies and is a key site in the Pole–Equator–Pole (PEP II) transect in the International Geosphere–Biosphere Programme (IGBP) Past Global Changes (PAGES) Program (Dodson and Lui 1995). Several studies have utilized diatoms (siliceous unicellular algae that respond rapidly to their changing environment) and biogenic silica (derived from diatom frustules) to reconstruct paleoclimates over a variety of timescales from the last 150 years to the last 12 million years (e.g., Granina et al. 1992; Qui et al. 1993; Colman et al. 1995; Williams et al. 1997; Mackay et al. 1998; Karabanov et al. 2000; Bangs et al. 2000; Kashiwaya et al. 2001). While changes in biogenic silica concentration in the sedimentary record have been linked to orbital forcing and glacial cycles, the physical connection between climate and diatom productivity has yet to be determined. Laboratory and field observations indicate that a number of factors may be involved, many of which relate to ice cover on the lake, including light penetration and water temperature (Richardson et al. 2000), ice formation and water column mixing, length of ice-free season, snow cover, turbidity, and nutrient cycling (Colman et al. 1995; Shimaraev and Granin 1991; Jewson and Granin 2000; Mackay et al. 2000).

Seasonal ice cover is a notable feature of Lake Baikal, a consequence of the extreme continental climate in the region (the −20°C isotherm transects Lake Baikal during the December–February season). Generally, freezing begins in late October (January) in the north (south) of the lake while ice decay begins in late April (March) and is complete by in mid-June (May). Ice thickness ranges from about 1 m in the north to about 80 cm in the south. Cessation of freezing is thought to be relatively independent of air temperatures, occurring when air temperatures are still negative (Verbolov et al. 1965). Subsequent ice breakup is also complex, determined by air temperatures, the rate of heat flow through the ice to the main water body, upwelling of warmer waters, and wind patterns (Verbolov et al. 1965).

In recent decades much climate research has focused on the investigation of planetary-wide associations between parameters representing the atmosphere–ocean–terrestrial systems. Most previous teleconnection studies have focused on the low to midlatitude climates of Europe and North America, and very few studies have linked limnological processes with the large-scale atmospheric circulation. Given the importance of ice cover to the physical, chemical, and biological processes within Lake Baikal it is of considerable interest to understand the nature of climatic controls on variability in ice cover. Indeed, the sensitivity of diatom species to ice dynamics requires that the relationship of ice to the regional climate be quantified in order to interpret diatom records of paleoclimate. In this context, the aims of this study are twofold. First, to examine the Lake Baikal ice cover record and, second, to analyze the coupling of Lake Baikal ice cover and the large-scale climate, including the dominant patterns of Northern Hemisphere climate variability.

2. Data and methods

a. Data

Dates of ice formation (ION) and breakup (IOFF) have been recorded at the Lystvyanka, Siberia, station (51°52′N, 104°51′E) in the south of Lake Baikal continuously since 1869 (Magnuson et al. 2000). Data for the period 1869–1996 are available for this study and are transformed to a count relative to 1 January. The duration of ice cover (ID, in days) is derived from dates of ice formation and breakup. Ice thickness has also been recorded since 1950 and here we use the maximum ice thickness (IT), generally observed in late February. Throughout the rest of this paper reference to Lake Baikal ice cover refers specifically to conditions at Lystvyanka. Variability in ice cover at Lystvyanka is known to be highly representative of that throughout Lake Baikal.

Statistical analysis of climate data has revealed the presence of preferred patterns or modes of variability at large scales in the Northern Hemisphere (NH). The 14 patterns identified by Barnston and Livezey (1987) are utilized in this study. Only those patterns that reveal significant association with Lake Baikal ice cover are discussed further, namely, the Scandinavian (SCA), Polar–Eurasian (POL), west Pacific (WP), and Pacific–North American (PNA) patterns. We also consider the leading mode of interannual variability in the NH extratropical circulation, the Arctic Oscillation (AO) (Thompson and Wallace 2000a), and the Siberian high (SH), a thermodynamically induced area of high surface pressure extending over much of eastern Eurasia and a dominant feature of the Eurasian winter climate (Sahsamanoglou et al. 1991). An index of the SH is derived from standardized anomalies of sea level pressure (SLP) over the region 40°–55°N, 90°–110°E. Figure 1 depicts the patterns considered further in sections 3 and 4. Standardized indices of these dominant modes of NH atmospheric variability are used in this study and cover the period 1950–96 in all cases except the SH (1871–1996) and the AO (1900–96).

Fig. 1.

Modes of NH winter climate variability relevant to this study: (a) the AO (or NH annular mode) during DJF, (b) the SCA in OND, (c) the POL pattern in DJF, (d) the WP pattern in SON, (e) the PNA pattern in NDJ. Panels show the correlation between seasonal means of standardized time series of the patterns and NCEP–NCAR 700-hPa geopotential height. Contour interval is 0.2, positive (negative) contours are solid (dotted)

Fig. 1.

Modes of NH winter climate variability relevant to this study: (a) the AO (or NH annular mode) during DJF, (b) the SCA in OND, (c) the POL pattern in DJF, (d) the WP pattern in SON, (e) the PNA pattern in NDJ. Panels show the correlation between seasonal means of standardized time series of the patterns and NCEP–NCAR 700-hPa geopotential height. Contour interval is 0.2, positive (negative) contours are solid (dotted)

The National Centers for Environmental Prediction–National Center for Atmospheric Research (NCEP–NCAR) produces the most extensive set of atmospheric reanalysis data currently available (Kalnay et al. 1996). Data from 1948 to 1996, on a 2.5° grid were used in this study. We also utilize longer-term observations of near-surface air temperature (SAT; Parker et al. 1994) and SLP (Basnett and Parker 1997), on 5° × 5° global grid for 1869–1996 and 1871–1996, respectively, although there are periods of missing data in the earlier part of these records.

Lower-tropospheric (925 hPa) heat budgets for a given period were derived from NCEP–NCAR fields. The heat budget equation in pressure coordinates can be defined as

 
formula

where T is air temperature, U is the horizontal wind vector, ω is the vertical velocity, Cp is the specific heat capacity, R is the gas constant, and P is pressure. The first (second) term on the right-hand side represents horizontal (vertical) temperature advection. The diabatic heating rate Q is a function of shortwave and longwave radiative processes and latent and sensible heat fluxes and is derived as a residual.

b. Methods

To assess the relationship between variability in Lake Baikal ice cover variables and the large-scale structure of the atmosphere, the following techniques are adopted. First, correlation coefficients are derived from the time series of Lake Baikal ice cover variables with those of the NH circulation pattern indices, as well as those of the gridded SAT, SLP, and NCEP–NCAR atmospheric fields. It is well known that long-term observations of climate variables contain trends that can invalidate the assumptions of stationarity and data independence underlying correlation analysis. Accordingly, all data series were detrended of linear trends prior to correlation analyses. Second, composite mean anomalies were calculated for selected NCEP–NCAR variables (representing the large-scale circulation over the Lake Baikal region) for samples of the six most extreme years of high and low Lake Baikal ice cover (over the period 1948–96 coincident with NCEP–NCAR data). Comparison of the composite mean tropospheric heat budget and surface energy budget terms for extreme high and low ice cover was also conducted to identify the mechanisms through which the observed atmospheric circulation anomalies may result in Lake Baikal ice cover anomalies. Where appropriate, the mean anomalies are tested for statistical significance using a t test.

3. Results and discussion

a. Trends in Lake Baikal ice cover

On average, ice formation at Lystvyanka occurs on 10 January and ice breakup on 4 May (Table 1), resulting in 114 days of ice cover. However, ID shows a highly significant negative trend (r = −0.41) of −1.16 days decade−1, concentrated into two periods from around 1880–1930 and 1972–96 (Fig. 2, Table 1). Both ION and IOFF reveal significant trends toward later ice onset (by 1.1 days decade−1) and earlier ice breakup (by −0.51 days decade−1), respectively, over the study period (Fig. 2, Table 1). However, the time series are markedly different and the two variables have only a relatively weak (but significant) correlation of −0.23. The trend toward later ION dates is more consistent over the record despite notable multidecadal variability. In contrast, the trend toward earlier ice breakup occurred almost exclusively in the period 1880–1920, after which time no significant trend is observed (Table 1). This suggests that the observed trend of reducing ID in the decades of the end of the nineteenth and start of the twentieth centuries is a result of both later ice formation and earlier ice breakup, while the trend in recent decades is associated largely with the former. The IT reduces over the period 1950–95, with a large and significant declining trend (of −8.84 cm decade−1) since the early 1970s, in line with the observations of ID and ION (Fig. 2; Table 1).

Table 1.

Statistics of Lake Baikal ice cover variables recorded at the Lystvyanka station

Statistics of Lake Baikal ice cover variables recorded at the Lystvyanka station
Statistics of Lake Baikal ice cover variables recorded at the Lystvyanka station
Fig. 2.

Time series of (a), (b), (c), (d) Lake Baikal ice variables and (e) regional (50°–60°N, 100°–110°E) near-surface air temperature anomalies during DJF (relative to 1960–90). Dotted line in each case represents an 11-yr running mean

Fig. 2.

Time series of (a), (b), (c), (d) Lake Baikal ice variables and (e) regional (50°–60°N, 100°–110°E) near-surface air temperature anomalies during DJF (relative to 1960–90). Dotted line in each case represents an 11-yr running mean

These trends of declining Lake Baikal ice cover are broadly consistent with those observed in other Northern Hemisphere lakes and rivers (Magnuson et al. 2000). Such strong trends in Lake Baikal ice cover are unsurprising given observed positive trends in winter SAT in central Asia and eastern Siberia that, in recent decades, are greater than any other location globally (Jones et al. 1999). Over the region surrounding Lake Baikal (50°–60°N, 100°–110°E) trends in SAT (SATBAIKAL) of 0.3°C decade−1 have occurred during winter since 1892, with the greatest warming of 1.1°C decade−1 since 1970 (Fig. 2). In this context, the absence of a decline in IOFF since 1920 is intriguing. However, as shown in the next section, variability in IOFF is most closely associated with SATBAIKAL during April. April is unique among the autumn/winter/spring months in that there is no significant positive trend in SATBAIKAL over the period 1920–96 (not shown).

Model predictions of climate change for the twenty-first century using scenarios of anthropogenic greenhouse gas emissions generally show pronounced near-surface warming in the Lake Baikal region during winter. Indeed, one of the most consistent features of different model simulations is that the maximum warming occurs over the high-latitude NH landmasses during winter (Houghton et al. 2002). Model simulations under conditions of a 1% annual increase in CO2 (incorporating sulphate forcing) suggest mean December–January–February (DJF, hereafter all 3-month seasons will be abbreviated into an acronym using only the first letter of each month, respectively) temperature changes for the period 2071–2100 over the northern Asian region ranging from 4.8° to 9.3°C, relative to the 1961–90 period (Houghton et al. 2002). Assuming a linear dependence of Lake Baikal ice duration on winter temperature we can infer that ice duration may decrease by a further 15–28 days by 2071–2100 under this scenario.

b. Interannual variability in Lake Baikal ice cover

1) Relationship of Lake Baikal ice cover with local and regional temperatures

The ION is more variable at interannual timescales than IOFF, and is more closely related to ID(r = 0.86) than is IOFF(r = −0.77). The ID exhibits statistically strong associations with winter SAT (peaking in the NDJ season) over northern Asia, from the Caspian Sea in the west to the Pacific in the east, and from northern India to northern Siberia, a region extending over approximately 40° latitude and 100° longitude (Fig. 3a). The patterns and magnitude of correlations of SAT with ION and IT are very similar to those with ID, and are not shown. Lake Baikal ice cover, therefore, represents a strong index of climate at the wider continental scale. The IOFF shows far weaker associations with regional SAT, and significant correlations are restricted to a regions extending from Lake Balkhash to northeast Siberia, some 30° latitude and 50° longitude (Fig. 3b), consistent with the findings of Livingston (1999).

Fig. 3.

Correlation of various variables with seasonal SAT: (a) NDJ SAT with Lake Baikal ice duration 1869–1996, (b) FMA SAT with Lake Baikal ice breakup date 1869–1996, (c) SAT with the AO during DJF 1900–96, (d) SAT with the Siberian high during Dec 1871–1996. All data are detrended. Positive (negative) correlation coefficients are shown as solid (dotted) lines. Contour interval is 0.1, the zero contour is omitted, and correlations significant at the 0.05 probability level are shaded

Fig. 3.

Correlation of various variables with seasonal SAT: (a) NDJ SAT with Lake Baikal ice duration 1869–1996, (b) FMA SAT with Lake Baikal ice breakup date 1869–1996, (c) SAT with the AO during DJF 1900–96, (d) SAT with the Siberian high during Dec 1871–1996. All data are detrended. Positive (negative) correlation coefficients are shown as solid (dotted) lines. Contour interval is 0.1, the zero contour is omitted, and correlations significant at the 0.05 probability level are shaded

Correlations of ID with SATBAIKAL are strongest in the early winter period (Fig. 4a) peaking during December, indicating greater sensitivity to meteorological conditions during the early winter freezing period than the spring breakup period. The ION is strongly determined by SATBAIKAL during the November–December period immediately prior to ice formation. In contrast, IT has strong correlations with SATBAIKAL throughout the winter period, peaking in February (Fig. 4a). There are no significant correlations between the ice variables and seasonal SATBAIKAL during the preceding summer (Fig. 4a) in contrast to the findings of Verbolov et al. (1965). Assuming that the ION, ID, and IT closely represent the heat budget of the lake, it appears that ice processes are dominated by winter heat loss rather than summer heat gain. In contrast to the other ice variables IOFF is only weakly related to seasonal SATBAIKAL (Fig. 4a), and correlations peak during April (r = −0.4). The processes of ice breakup, therefore, appear to be distinct from those of ice formation. Given that ice breakup can begin in subzero temperatures, these results indirectly add weight to observations linking ice breakup to solar radiation transmission through clear ice (and thus to snow cover), wind patterns, and the timing and magnitude of spring river inflows into the lake (Verbolov et al. 1965; Magnuson et al. 2000). Indeed, the absence of a trend in IOFF in recent decades may be related to the observed increase in snow depth in recent decades throughout the Northern Hemisphere and specifically in Eurasia (Ye et al. 1998; Brown 2000), which may act to reduce the absorption by the lake of incident solar radiation in early spring.

Fig. 4.

(a) Correlations between seasonally averaged SAT in the Lake Baikal region (SATBAIKAL; 50°–60°N, 100°–110°E) and Lake Baikal ice cover variables of ION, IOFF, ID over the period 1893–1996, and IT over the period 1950–96. The critical correlation coefficient (at the 0.05 probability level) for variables ION, IOFF, and ID (IT) is 0.1946 (0.2875). (b) Correlations between SATBAIKAL and selected NH teleconnection patterns (all seasonally averaged) over the period 1950–96. The critical correlation coefficient (at the 0.05 probability level) is 0.2875. All data are detrended

Fig. 4.

(a) Correlations between seasonally averaged SAT in the Lake Baikal region (SATBAIKAL; 50°–60°N, 100°–110°E) and Lake Baikal ice cover variables of ION, IOFF, ID over the period 1893–1996, and IT over the period 1950–96. The critical correlation coefficient (at the 0.05 probability level) for variables ION, IOFF, and ID (IT) is 0.1946 (0.2875). (b) Correlations between SATBAIKAL and selected NH teleconnection patterns (all seasonally averaged) over the period 1950–96. The critical correlation coefficient (at the 0.05 probability level) is 0.2875. All data are detrended

2) Association with modes of Northern Hemisphere climate variability

The most notable features of the correlations of Lake Baikal ice cover variables with NH teleconnection indices are the strong associations with the AO and SCA patterns and the SH (Fig. 5). Generally, correlations between circulation indices and IOFF are relatively low reflecting weaker climatic control on ice breakup processes. The strong influence of the AO on ION, ID, and IT is not surprising given the considerable similarity in the broadscale structure of correlations of SAT with these ice cover variables (Fig. 3a) and those with the AO (Fig. 3c) over Eurasia. In the positive phase of the AO (Fig. 1a) regions poleward of 40°N experience enhanced westerlies in which the zonal component is directed across the climatological gradient of temperature in the East Asian region (Thompson and Wallace 2000a). As a result, strong positive temperature advection and SAT anomalies occur in central and northern Eurasia, including the Lake Baikal region (Figs. 3c and 4b), resulting in reduced ice cover and thickness (Fig. 5). Further, at the daily timescale, a decrease (increase) in blocking events over northern Russia and a consequent decrease (increase) in cold air outbreaks over East Asia is observed during the AO positive (negative) phase (Thompson and Wallace 2001). Indeed, much of the trend in Eurasian SAT observed over the latter decades of the twentieth century is related to the observed trend toward the positive phase of the AO (Thompson and Wallace 2000b).

Fig. 5.

Correlations between seasonally averaged indices of NH climate variability patterns and (a) Lake Baikal ice onset date, (b) Lake Baikal ice breakup date, (c) Lake Baikal ice duration, and (d) Lake Baikal ice thickness. All data cover the period 1950–96 and are detrended. The critical correlation coefficient (at the 0.05 probability level) is 0.2875

Fig. 5.

Correlations between seasonally averaged indices of NH climate variability patterns and (a) Lake Baikal ice onset date, (b) Lake Baikal ice breakup date, (c) Lake Baikal ice duration, and (d) Lake Baikal ice thickness. All data cover the period 1950–96 and are detrended. The critical correlation coefficient (at the 0.05 probability level) is 0.2875

The positive phase of the SCA pattern (Fig. 1b) is often associated with blocking anticyclones in the Scandinavian region, presenting a barrier to zonal westerly flow and positive temperature advection in northern Asia, east of the Caspian Sea toward the Lake Baikal region (Clark et al. 1999). This is most pronounced in the early winter (Fig. 4b). As a result, there are strong positive correlations between SCA and ID, IT and a negative correlation between SCA and ION (Figs. 5a,c,d). These findings are in accordance with previous studies that have documented the influence of the SCA pattern on Eurasian snow cover (Clark et al. 1999) and winter temperature variability in northern China (Yin 1999).

A stronger SH in early winter is associated with increased IT and earlier ION (Figs. 5a,d). The SH has the effect of blocking westerly flow in the Lake Baikal region. Increased SH intensity is associated with negative (positive) SAT anomalies over much of midlatitude East Asia including Lake Baikal (northwest Siberia) (Figs. 3d, 4b), presumably associated with cold (warm) temperature advection within the northerly (southerly) flow around the east (west) side of the SH (Clark et al. 1999). The POL pattern is apparent only in the midwinter months and as such exerts a significant influence only on the IT and IOFF variables (Figs. 5b,d). The POL positive phase is characterized by an enhanced circumpolar circulation (Fig. 1) and is thus similar to the positive AO phase in terms of the midlatitude circulation anomalies and influence on SATBAIKAL (Fig. 4b).

All Lake Baikal ice cover variables have statistically significant associations with certain Pacific-based circulation patterns in advance of the ice season, either during the autumn (in the case of the WP and PNA) or during the previous winter (in the case of the PNA; Fig. 5). The structure of the WP pattern exhibits a seasonal cycle and in autumn (SON) the center over Kamchatka extends westward to cover parts of East Asia (Fig. 1). Under these circumstances the influence on the zonal circulation in the Lake Baikal region is pronounced with the WP positive phase associated with an anomalous northeasterly circulation in the Baikal region (Fig. 1). The result is a negative correlation with ION and positive correlation with ID (Figs. 5a,c).

Notably, the PNA pattern during the early winter of the previous year (NDJ-1) has significant correlations with all Lake Baikal ice cover variables such that there may be some degree of long-term predictability in Lake Baikal ice cover. The positive (negative) phase of the PNA in NDJ-1 is associated with statistically significant negative (positive) SLP anomalies over northern Siberia in the following winter and, hence, anomalous southwesterly (northwesterly) flow over Baikal and potentially reduced (enhanced) ice cover (Fig. 7d). However, the mechanism by which such a delayed atmospheric response over Eurasia to an essentially Pacific-based feature could operate is unclear.

Fig. 7.

Correlation of (a) Lake Baikal ice duration with NDJ 700-hPa geopotential height, (b) Lake Baikal ice thickness with JFM 700-hPa geopotential height, (c) Lake Baikal ice thickness variability (unrelated to the AO) and Dec SLP, and (d) PNA and SLP (lagged 1 year) during the NDJ season. All data cover the period 1948–96. Positive (negative) correlation coefficients are shown as solid (dotted) lines. Contour interval is 0.1 and correlations significant at the 0.05 probability level are shaded

Fig. 7.

Correlation of (a) Lake Baikal ice duration with NDJ 700-hPa geopotential height, (b) Lake Baikal ice thickness with JFM 700-hPa geopotential height, (c) Lake Baikal ice thickness variability (unrelated to the AO) and Dec SLP, and (d) PNA and SLP (lagged 1 year) during the NDJ season. All data cover the period 1948–96. Positive (negative) correlation coefficients are shown as solid (dotted) lines. Contour interval is 0.1 and correlations significant at the 0.05 probability level are shaded

3) Atmospheric circulation anomalies associated with Lake Baikal ice cover

The mean state of the NH upper atmosphere in winter shows two major stationary midlatitude troughs over the east coasts of North America and East Asia and ridges over the eastern Atlantic and eastern Pacific, produced by land–sea thermal contrasts (Fig. 6a). At 700 hPa a ridge over Mongolia and central Asia is evident, associated with the Tibetan Plateau and the Siberian high (Fig. 6b). The spatial structure of correlation coefficients between Lake Baikal ice variables and NCEP–NCAR surface and upper-level fields is very similar to that of the NCEP–NCAR composite anomalies associated with the extreme years of high and low ice conditions. This adds confidence to our interpretation of these fields and for simplicity we mainly present the fields of correlation coefficients. Correlation patterns of geopotential height fields with ID and ION reveal circulation anomalies that are very similar (except that they are of opposite sign) and we have shown only the former.

Fig. 6.

Climatological mean (1948–2000) DJF (a) 200-hPa geopotential height (b) 700-hPa geopotential height, and (c) 700-hPa temperature

Fig. 6.

Climatological mean (1948–2000) DJF (a) 200-hPa geopotential height (b) 700-hPa geopotential height, and (c) 700-hPa temperature

Associated with variability in ID and ION is an anomalous upper-air wave train pattern in middle and high latitudes during early winter. This extends from the Atlantic to the west Pacific (Fig. 7a) and is most similar to the SCA pattern (Fig. 1b). This structure evolves to become more zonally symmetric by late winter (not shown). Thus, years of high ID and early ION are associated with an upper-level ridge over northwest Siberia and troughs over central East Asia and southern Europe in early winter. As a result, the polar low and central Asian ridge are weakened, reducing the poleward geopotential gradient in the East Asian region, resulting in a pronounced anomalous meridional flow regime. Composites of 700-hPa geopotential height fields show a near reversal of this pattern between high and low ID and ION years (not shown). Thus, during years of high ID and early ION pronounced northeasterly low-level wind anomalies and reduced wind speeds occur (Fig. 8). A reversal of the wind direction anomalies occurs in year of low ID and late ION. These wind anomalies are directed across the gradient of mean temperature (Fig. 6c) such that heat transfer anomalies result, providing a highly plausible mechanism for interannual variability in ID and ION.

Fig. 8.

1000-hPa wind vector anomalies (m s−1) during NDJ associated with composites of high (1991/92, 1974/75, 1949/50, 1969/70, 1968/69, 1966/67) minus low (1958/59, 1988/89, 1977/78, 1960/61, 1948/49, 1990/91) Lake Baikal ice duration years

Fig. 8.

1000-hPa wind vector anomalies (m s−1) during NDJ associated with composites of high (1991/92, 1974/75, 1949/50, 1969/70, 1968/69, 1966/67) minus low (1958/59, 1988/89, 1977/78, 1960/61, 1948/49, 1990/91) Lake Baikal ice duration years

Indeed, analysis of the contribution to the heat budget (at 925 hPa) of the individual terms of horizontal advection, vertical advection, and diabatic heating [Eq. (1)], demonstrates the central role of horizontal heat advection in explaining the differences between years of early and late ION. Considering the early winter cooling period (1 October–31 December) immediately prior to ice onset we have derived the cumulative values of each term in the heat budget, averaged over the Lake Baikal region (50°–57.5°N, 102.5°–110°E). These values are averaged for the samples of six extreme years of early and late ice onset on Lake Baikal (Fig. 9).

Fig. 9.

Daily cumulative heat budget terms based on Eq. (1) at 925 hPa averaged over the Lake Baikal region (50°–57.5°N, 102.5°–110°E): (a) temperature change, (b) horizontal advection term, (c) vertical advection term, and (d) residual diabatic heating term. Graphs show composite mean of the six earliest Lake Baikal ice onset years (1974, 1959, 1966, 1991, 1949, 1952) (thick line) and the six latest ice onset years (1958, 1951, 1963, 1982, 1986, 1977) (thin line with dots)

Fig. 9.

Daily cumulative heat budget terms based on Eq. (1) at 925 hPa averaged over the Lake Baikal region (50°–57.5°N, 102.5°–110°E): (a) temperature change, (b) horizontal advection term, (c) vertical advection term, and (d) residual diabatic heating term. Graphs show composite mean of the six earliest Lake Baikal ice onset years (1974, 1959, 1966, 1991, 1949, 1952) (thick line) and the six latest ice onset years (1958, 1951, 1963, 1982, 1986, 1977) (thin line with dots)

Both samples show rapid cooling over the 3-month period and the difference between the early and late ION composites in the amount of cooling is 5.3 K (Fig. 9a). This is very small relative to the magnitude of the individual terms. In all sample years, the majority of cooling is provided by the residual term, which is likely to be dominated by longwave radiative cooling (Fig. 9d). Atmospheric subsidence is a prominent feature of the winter climate over eastern Siberia, such that in all sample years vertical temperature advection contributes to warming (Fig. 9c). Although the horizontal advection term makes the smallest contribution to temperature changes the difference between early and late ION years is most pronounced in this term (Fig. 9b). It is striking that horizontal advection acts to warm the lower troposphere in the Lake Baikal region in the late ION years and cool it in the early ION years. Analysis of individual years (not shown) indicates that winter cooling and horizontal temperature advection are actually highly episodic, representing synoptic and subseasonal variability in the large-scale circulation. Thus, seasonal temperature anomalies in the Lake Baikal region are substantially controlled by the nature and intensity of individual episodic horizontal temperature advection events. The composites of the large-scale circulation (Figs. 7, 8) and the flow regimes associate with the relevant modes of NH climate variability (see previous section) provide a clear indication of the relationship between these episodic events and the planetary circulation.

The relationship between the atmospheric heat budget and the surface energy balance can be examined by considering equivalent composite mean values of the surface energy balance terms as estimated in the NCEP–NCAR data. Clearly, these terms are sensitive to the land surface scheme utilized in the NCEP–NCAR model and are derived at a rather coarse resolution that cannot resolve even surface features as large as Lake Baikal. Nevertheless, the results are instructive. NCEP–NCAR surface skin temperatures during early winter (OND) show a mean difference of some 4.1°C in the Lake Baikal region between years of late and early ION (Table 2). During the OND season the climatology shows that the Lake Baikal region experiences a net surface radiative flux deficit that is partly compensated for by a net sensible heat flux from the atmosphere to the surface (Table 2). Composite analysis of the surface energy balance terms for early and late ION years indicates that the largest difference occurs in the sensible heat flux component. Years of late ION are associated with an enhanced atmosphere to surface sensible heat flux, in contrast to the early ION years where the flux is reduced. Heat exchanges through latent heat fluxes, ground heat flux, and radiative fluxes are of less importance. This suggests that the main control on OND surface temperature anomalies is variations in the transfer of sensible heat from the atmosphere to the surface. Despite the limitations in the NCEP–NCAR reanalysis data these results (Figs. 79 and Table 2) together provide evidence of a direct link between large-scale atmospheric circulation anomalies, the regional atmospheric heat budget, and the surface energy balance consistent with the variations observed in Lake Baikal ice onset conditions.

Table 2.

Mean surface temperature and surface radiative and energy fluxes for the OND season averaged over the Lake Baikal region (51°–56°N, 103°–110°E). Values are shown for the climatological mean (1968–99) and for the composite mean for of the six earliest Lake Baikal ice onset years (1974, 1959, 1966, 1991, 1949, 1952) minus the six latest ice onset years (1958, 1951, 1963, 1982, 1986, 1977). All values are derived from NCEP–NCAR data

Mean surface temperature and surface radiative and energy fluxes for the OND season averaged over the Lake Baikal region (51°–56°N, 103°–110°E). Values are shown for the climatological mean (1968–99) and for the composite mean for of the six earliest Lake Baikal ice onset years (1974, 1959, 1966, 1991, 1949, 1952) minus the six latest ice onset years (1958, 1951, 1963, 1982, 1986, 1977). All values are derived from NCEP–NCAR data
Mean surface temperature and surface radiative and energy fluxes for the OND season averaged over the Lake Baikal region (51°–56°N, 103°–110°E). Values are shown for the climatological mean (1968–99) and for the composite mean for of the six earliest Lake Baikal ice onset years (1974, 1959, 1966, 1991, 1949, 1952) minus the six latest ice onset years (1958, 1951, 1963, 1982, 1986, 1977). All values are derived from NCEP–NCAR data

Ice thickness shows a structure of correlation with near-surface and upper-level fields similar to those with ID and ION, except that the upper-level height anomalies show a more zonal midlatitude–polar dipole (Fig. 7b), characteristic of the AO. This pattern persists into late winter, at which time correlations between IT and the AO are highest (Fig. 5d). The structure also bears close resemblance to the POL pattern (Fig. 1c). Composite low-level wind anomalies in the Lake Baikal region associated with years of high minus low IT (not shown) are very similar to those associated with ID and ION such that years of high IT are associated with anomalous northeasterly flow and cold advection, while years of low IT experience southwesterly wind anomalies and warm advection.

The component of Lake Baikal ice cover variability unrelated to the AO/SCA can be obtained by regressing the AO/SCA (during OND for ION and JFM for IT) against the ice time series and retaining the residuals. The large-scale circulation anomalies associated with the component of ION and IT ice cover variability unrelated to the AO are very similar and are primarily related to the Siberian high (Fig. 7c). A similar pattern is observed after removal of the linear influence of the SCA pattern (not shown). Thus variability in Lake Baikal ice cover unrelated to the AO or SCA may be modulated largely by shifts in the position and intensity of the SH. Gong et al. (2001) suggest that the AO influences winter temperatures over eastern China through its association with the SH. However, although during DJF AO and SH are related (r = −0.34) our results suggest an independent role for the SH in Lake Baikal ice cover. Zhang et al. (1997) suggest that the SH intensity may be modulated by ENSO, although we find no evidence of a significant ENSO influence on Lake Baikal ice cover.

4. Conclusions

Climatic control on Lake Baikal ice cover is of major limnological and ecological importance, as well as having potentially important implications for understanding the sedimentary record. All the ice cover variables (recorded at the Lystyanka station) are strongly related to winter temperatures throughout northern Asia. The results indicate that Lake Baikal ice cover responds strongly to regional-scale temperature variability, and is clearly a robust indicator of regional and continental-scale climate variability and change. Given the sensitivity of diatom species to ice cover characteristics this finding is important to our climatic interpretation of Lake Baikal sedimentary records.

Interannual variability in Lake Baikal ice cover is associated with coherent patterns of upper-level geopotential height anomalies, characterizing the amplitude and phase of the Eurasian wave train and NH annular mode. Consequently, strong associations are observed between Lake Baikal ice cover and indices of NH circulation patterns, notably the SCA and AO patterns. Our analysis further reveals that regional winter temperature variability (and thus Lake Baikal ice cover) is driven largely by the nature, frequency, and intensity of episodic horizontal temperature advection events. It is through such events that the identified patterns of NH climate variability (especially the SCA and AO) exert their influence on Lake Baikal ice cover. The results also raise the possibility of predictability in winter temperatures in the Lake Baikal region and in Lake Baikal ice cover. Further investigation of possible mechanisms by which the PNA may influence Lake Baikal one year later is suggested. That certain preferred modes of Northern Hemisphere variability may themselves be predictable (Rodwell et al. 1999; Baldwin and Dunkerton 2001) provides further potential for predictability in Lake Baikal ice cover.

Analysis of the surface energy budget terms, during the ice onset period at least, provides evidence of a direct and consistent link between tropospheric and surface temperature anomalies and hence the observed ice onset anomalies. Variations in the rapidity of ice onset appear to be associated largely with variations in sensible heat flux between the atmosphere and surface, resulting from variability in the large-scale atmospheric circulation as documented.

However, there are other climate parameters, including surface wind speed, that can influence ice formation and especially ice breakup processes (Verbalov et al. 1965), and that are also likely to be modulated by large-scale climate variability. NCEP–NCAR surface wind stress in the Lake Baikal region show little coherent association with the Lake Baikal ice cover variables (not shown). However, it is questionable whether the NCEP–NCAR fields can accurately represent the local-scale structure of wind speed variability, particularly that caused by the complex local topography.

Despite the limitations of using NCEP–NCAR data for understanding local-scale processes, our analysis of the surface energy budget at least provides a first-order estimate of the nature of surface–atmosphere interaction in the Lake Baikal region. There is, however, a need for further research into the local-scale lake–atmosphere interactions such that the remainder of variability in the ice cover variables, particularly ice breakup, may be understood. Such work requires long-term datasets of climate and limnology at the local scale.

The observed trends in Lake Baikal ice formation and breakup dates, duration, and thickness, suggesting reduced ice volumes since 1869 to the present, provide compelling evidence that the lake is responding to observed warming trends. This adds confidence to our interpretation of surface temperature observations in the region. In conjunction with observed positive trends in precipitation over Central Asia (Aizen et al. 2001) and snow depth in northern Russia (Ye 2000) the effects on lake ecology may be profound. There is a need for long-term ecological studies to establish this.

The observed trends in ice cover since the mid-1970s may be directly linked to a trend toward the positive phase of the AO over the same period. Indeed, this may indicate that the influence of the AO compared to that of other circulation patterns is stronger at decadal and multidecadal timescales. There are numerous possible explanations for the recent trend in AO including stratospheric ozone depletion (Volodin and Galin 1998), increasing concentrations of greenhouse gases and aerosols (Shindell et al. 1999), decadal variability in SST (Robertson et al. 2000), and volcanic eruptions (Kelly et al. 1996). As such, the strong association of Lake Baikal ice cover and the AO may have considerable implications for lake ecology given the possibility that future anthropogenic climate change may be expressed substantially through changes in amplitude and phase of existing modes of variability. In any case, there is a strong case that substantial declines in Lake Baikal ice cover may be expected in the future.

Acknowledgments

The authors are grateful to University College London for support. AWM also acknowledges support from The Royal Society under the UK-BICER Project and the EU Framework 5 Programme through the CONTINENT Project (EVK2-CT-2000-0057). The authors would also like to thank N.G. Granin for providing the Lystvyanka ice data.

REFERENCES

REFERENCES
Aizen
,
E. M.
,
V. B.
Aizen
,
J. M.
Melack
,
T.
Nakamura
, and
T.
Ohta
,
2001
:
Precipitation and atmospheric circulation patterns at mid-latitudes of Asia.
Int. J. Climatol.
,
21
,
535
556
.
Baldwin
,
M. P.
, and
T. J.
Dunkerton
,
2001
:
Stratospheric harbingers of anomalous weather regimes.
Science
,
294
,
581
584
.
Bangs
,
M.
,
R. W.
Battarbee
,
R. J.
Flower
,
D.
Jewson
,
J. A.
Lees
,
M.
Sturm
,
E.
Vologinova
, and
A. W.
Mackay
,
2000
:
Climate change in Lake Baikal: Diatom evidence in an area of continuous sedimentation.
Int. J. Earth Sci.
,
89
,
251
259
.
Barnston
,
A. G.
, and
R. E.
Livezey
,
1987
:
Classification, seasonality and persistence of low-frequency atmospheric circulation patterns.
Mon. Wea. Rev.
,
115
,
1083
1126
.
Basnett
,
T. A.
, and
D. E.
Parker
,
1997
:
Development of the Global Mean Sea Level Pressure Data Set GMSLP2.
Hadley Centre Climate Research Tech. Note CRTN 79, 16 pp
.
Brown
,
R. D.
,
2000
:
Northern Hemisphere snow cover variability and change, 1915–97.
J. Climate
,
13
,
2339
2355
.
Clark
,
M. P.
,
M. C.
Serreze
, and
D. A.
Robinson
,
1999
:
Atmospheric controls on Eurasian snow extent.
Int. J. Climatol.
,
19
,
27
40
.
Colman
,
S. M.
,
J. A.
Peck
,
E. B.
Karabanov
,
S. J.
Carter
,
J. P.
Bradbury
,
J. W.
King
, and
D. F.
Williams
,
1995
:
Continental climate response to orbital forcing from biogenic silica records in Lake Baikal.
Nature
,
378
,
769
771
.
Dodson
,
J.
, and
T.
Lui
,
1995
:
PEP II: The Austral–Asian Transect.
Palaeoclimates of the Northern and Southern Hemispheres, R. S. Bradley et al., Eds., Pages Report Series, Vol. 95-1, IGBP Pages, 43–64
.
Fryer
,
G.
,
1991
:
Comparative aspects of adaptive radiation and speciation in Lake Baikal and the great rift lakes of Africa.
Hydrobiologia
,
211
,
137
146
.
Gong
,
D. Y.
,
S. W.
Wang
, and
J. H.
Zhu
,
2001
:
East Asian winter monsoon and Arctic Oscillation.
Geophys. Res. Lett.
,
28
,
2073
2076
.
Granina
,
L. Z.
,
E. B.
Karabanov
,
M. K.
Shimaraeva
,
D. F.
Williams
, and
V. M.
Kuptsov
,
1992
:
Biogenic silica of Baikal bottom sediments used for palaeo reconstruction.
Int. Project Palaeolimnol. Lake Cenozoic Climate Newsl.
,
6
,
53
59
.
Houghton
,
J. T.
,
Y.
Ding
,
D. J.
Griggs
,
M.
Noguer
,
P. J.
van dor Lindon
,
X.
Dai
,
K.
Maskell
, and
C. A.
Johnson
,
Eds.,
.
2002
:
Climate Change 2001: The Scientific Basis.
Cambridge University Press, 881 pp
.
Jewson
,
D. H.
, and
N. G.
Granin
,
2000
:
How can present day studies of diatoms help in understanding past climatic change in Baikal?
Terra Nostra
,
9
,
29
33
.
Jones
,
P. D.
,
M.
New
,
D. E.
Parker
,
S.
Martin
, and
I. G.
Rigor
,
1999
:
Surface air temperature and its change over the past 150 years.
Rev. Geophys.
,
37
,
173
199
.
Kalnay
,
E.
, and
Coauthors
.
1996
:
The NCEP/NCAR 40-Year Reanalysis Project.
Bull. Amer. Meteor. Soc.
,
77
,
437
471
.
Karabanov
,
E. B.
,
A. A.
Prokopenko
,
D. F.
Williams
, and
G. K.
Khursevich
,
2000
:
A new record of Holocene climate change from the bottom sediments of Lake Baikal.
Palaeogeogr., Palaeoclimatol., Palaeoecol.
,
156
,
211
224
.
Kashiwaya
,
K.
,
S.
Ochiai
,
H.
Sakai
, and
T.
Kawai
,
2001
:
Orbit-related long-term climate cycles revealed in a 12-Myr continental record from Lake Baikal.
Nature
,
410
,
71
74
.
Kelly
,
P. M.
,
P. D.
Jones
, and
J.
Pengqun
,
1996
:
The spatial response of the climate system to explosive volcanic eruptions.
Int. J. Climatol.
,
16
,
537
550
.
Kozhov
,
M.
,
1963
:
Lake Baikal and Its Life.
W. Junk, 344 pp
.
Livingston
,
D. W.
,
1999
:
Ice break-up on southern Lake Baikal and its relationship to local and regional air temperatures in Siberia and to the North Atlantic Oscillation.
Limnol. Oceanogr.
,
44
,
1486
1497
.
Mackay
,
A. W.
,
R. J.
Flower
,
A. E.
Kuzmina
,
L. Z.
Granina
,
N. L.
Rose
,
P. G.
Appleby
,
J. F.
Boyle
, and
R. W.
Battarbee
,
1998
:
Diatom succession trends in recent sediments from Lake Baikal and their relationship to atmospheric pollution and to climate change.
Philos. Trans. Roy. Soc. London
,
B353
.
1011
1055
.
Mackay
,
A. W.
,
R. W.
Battarbee
,
R. J.
Flower
,
D.
Jewson
,
J. A.
Lees
,
D. B.
Ryves
, and
M.
Sturm
,
2000
:
The deposition and accumulation of endemic planktonic diatoms in the sediments of Lake Baikal and an evaluation of their potential role in climate reconstruction during the Holocene.
Terra Nostra
,
9
,
34
48
.
Magnuson
,
J. J.
, and
Coauthors
.
2000
:
Historical trends in lake and river ice cover in the northern hemisphere.
Science
,
289
,
1743
1747
.
Parker
,
D. E.
,
P. D.
Jones
,
A.
Bevan
, and
C. K.
Folland
,
1994
:
Interdecadal changes of surface temperature since the late 19th century.
J. Geophys. Res.
,
99
,
14373
14399
.
Qui
,
L.
,
D. F.
Williams
,
A.
Gvorzdkov
,
E.
Karabanov
, and
M. N.
Shimaraeva
,
1993
:
Biogenic silica accumulation and paleoproductivity in the northern basin of Lake Baikal during the Holocene.
Geology
,
21
,
25
28
.
Richardson
,
T. L.
,
C. E.
Gibson
, and
S. I.
Heaney
,
2000
:
Temperature growth and seasonal succession of phytoplankton in Lake Baikal, Siberia.
Freshwater Biol.
,
44
,
431
440
.
Robertson
,
A. W.
,
C. R.
Mechoso
, and
Y. J.
Kim
,
2000
:
The influence of Atlantic sea surface temperature anomalies on the North Atlantic Oscillation.
J. Climate
,
13
,
122
138
.
Rodwell
,
M. R.
,
D. P.
Rowell
, and
C. K.
Folland
,
1999
:
Oceanic forcing of the wintertime North Atlantic Oscillation and European climate.
Nature
,
398
,
320
323
.
Sahsamanoglou
,
H. S.
,
T. J.
Makrogiannis
, and
P. P.
Kallimopoulos
,
1991
:
Some aspects of the basic characteristics of the Siberian anticyclone.
Int. J. Climatol.
,
11
,
827
839
.
Shimaraev
,
M. N.
, and
N. G.
Granin
,
1991
:
Temperature stratification and the mechanisms of convection in Lake Baikal (in Russian).
Trans. USSR Acad. Sci.
,
321
,
831
835
.
Shimaraev
,
M. N.
,
V. I.
Verbolov
,
N. G.
Granin
, and
P. P.
Sherstyankin
,
1994
:
Physical Limnology of Lake Baikal: A review.
BICER Publishers, 81 pp
.
Shindell
,
D. T.
,
R. L.
Miller
,
G. A.
Schmidt
, and
L.
Pandolfo
,
1999
:
Simulation of recent northern winter climate trends by greenhouse gas forcing.
Nature
,
399
,
452
455
.
Thompson
,
D. W. J.
, and
J. M.
Wallace
,
2000a
:
Annular modes in the extratopical circulation. Part I: Month-to-month variability.
J. Climate
,
13
,
1000
1016
.
Thompson
,
D. W. J.
, and
J. M.
Wallace
,
2000b
:
Annular modes in the extratopical circulation. Part II: Trends.
J. Climate
,
13
,
1018
1039
.
Thompson
,
D. W. J.
, and
J. M.
Wallace
,
2001
:
Regional climate impacts of the northern hemisphere annular mode and associated climate trends.
Science
,
293
,
85
89
.
Verbolov
,
V. I.
,
V. N.
Sokolnikov
, and
M. N.
Shimaraev
,
1965
:
Hydrometeorological Conditions and Heat Balance of Lake Baikal (in Russian).
Nauka, 376 pp
.
Volodin
,
E. M.
, and
V. Y.
Galin
,
1999
:
Interpretation of winter warming on Northern Hemisphere continents in 1977–94.
J. Climate
,
12
,
2947
2955
.
Weiss
,
R. F.
,
E. C.
Carmak
, and
V. M.
Koropalov
,
1991
:
Deep-water renewal and biological production in Lake Baikal.
Nature
,
349
,
665
669
.
Williams
,
D. F.
,
J.
Peck
,
E. B.
Karabanov
,
A. A.
Prokopenko
,
V.
Kravchinsky
,
J.
King
, and
M. I.
Kuzmin
,
1997
:
Lake Baikal record of continental climate response to orbital insolation during the past 5 million years.
Science
,
278
,
1114
1117
.
Ye
,
H.
,
2000
:
Decadal variability or Russian winter snow accumulation and its associations with Atlantic sea surface temperature anomalies.
Int. J. Climatol.
,
20
,
1709
1728
.
Ye
,
H.
,
H. R.
Cho
, and
P. E.
Gustafson
,
1998
:
The changes in Russian winter snow accumulation during 1936–83 and its spatial patterns.
J. Climate
,
11
,
856
863
.
Yin
,
Z.
,
1999
:
Winter temperature anomalies of the North China plain and macroscale extratropical circulation patterns.
Int. J. Climatol.
,
19
,
291
308
.
Zhang
,
Y.
,
K. R.
Sperber
, and
J. S.
Boyle
,
1997
:
Climatology and interannual variation of the east Asian winter monsoon: Results from the 1979–95 NCEP/NCAR reanalysis.
Mon. Wea. Rev.
,
125
,
2605
2619
.

Footnotes

Corresponding author address: Dr. Martin C. Todd, Department of Geography, University College London, 26 Bedford Way, London WC1H 0AP, United Kingdom. Email: m.todd@geog.ucl.uk