Title Formation of the Ice Core Isotopic Composition

Author(s) Ekaykin, Alexey A.; Lipenkov, Vladimir Ya.

低温科学, 68(Supplement), 299-314 Citation Physics of Ice Core Records II : Papers collected after the 2nd International Workshop on Physics of Ice Core Records, held in Sapporo, Japan, 2-6 February 2007. Edited by Takeo Hondoh

Issue Date 2009-12

Doc URL http://hdl.handle.net/2115/45456

Type bulletin (article)

Note IV. Chemical properties and isotopes

File Information LTS68suppl_022.pdf

Instructions for use

Hokkaido University Collection of Scholarly and Academic Papers : HUSCAP Formation of the Ice Core Isotopic Composition

Alcxcy A. Ekaykin·· •• , Vladimir Va. Lipcnkov"

'" Hokkaido Ulliversily, Sapporo, Japan, ekaykill@aari,lIl1'.ru •• Arctic and All/arctic Research Instilule, SI. Petersburg. Russia. [email protected]

Abstract: Main processes of the ice core isotopic In this work we overview the processes leading to the composition formation are overviewed. Theory of formation of the vertical profile of isotopic composition isotope-temperature relationship is discussed and of an ice core. using the extensive data set obtained at confirmed by a number of experimental data. The the Russian Vostok Station, central Antarctica. At first , factors related to wind-driven spatial snow we present the theoretical background of the redistribution and post-depositional isotopic changes relationship between the stable water isotope content in that may aller or weaken this relationship, are also precipitation and air temperature (Section 2). In Section considered. For high-resolution isotopic time-series 3 we consider the limitations of isotopic method due to obtatned at sites wilh low accumulation of snow, Ihe the fonnation of non-climatic noise as a result of signal-to-noise ratio is shown to be as low as 0.25, depositional and post-depositional processes in the which means that noise accounts for about 80 % of the upper snow thickness. 1n order to estimate the signal-to­ total variance. [t is demonstrated that "classical isotopic noise ratio in vertical isotopic profiles, the data on local method" (based on the present-day geographical spatial isotopic distribution in the vicinity of Vostok isotope-temperature slope) underestimates the amplitude Station are used. Temporal variability of isotopic of past temperature changes in Antarctica. The most content of the Vostok precipitation and snow thickness likely reason for the discrepancy is the change in the is then compared with the instrumental record of air moisture source conditions. After correction for the temperature (Section 4) to derive the local isotope­ latter, the paleo-temperature reconstructions produced temperature calibration coefficient needed for the deep by the isotopic method become consistent with those ice core-based paleo-reconstructions. We also use the obtained from borehole temperature measurements. We high-resolution isotopic profi les from deep snow pits to show that in the case of the Vostok ice core, both reconstruct Vostok temperature history over the past approaches lead to the same temperature shift of 1000C 200 years thus demonstrating the validity of the method between LGM and the present time. The isotopic within a shorter time scale. Finally, the isotopic record composition of the basal part of the Vostok ice core, from the deep Vostok borehole is considered in Section comprising frozen subg[ac ial Lake Vostok water, is also 5. The paleotemperature reconstruction obtained with discussed. isotope approach is compared with that inferred from borehole temperature measurements. The most reliable Key words: Antarctica. Vostok station, ice core, estimate of the amplitude of the past temperature isotopic composition, paleo-climate. Lake Vostok changes at Vostok, consistent with results of both methods, is presented. Section 6 of the paper deals with formation of the isotopic composition of the basal I. lntroduction and outline section of the Vostok ice core comprising frozen water of subglacial Lake Vostok. Stable water isotopic composition (Of) and 8 80) of ice cores is commonly known as one of the most 2. Theory of isotope-temperature valuable paleo-climate proxies for temperature relationship. Modeling isotopic content of conditions of the precipitation fonnation in the past [15]. precipitation The observed relationship between air temperature and isotopic content of the precipitation (both geographical In this section we discuss the theoretical background and temporal) has been robustly con finned of the relationship between the stable water isotope experimentally and explained theoretically (see review content of precipitation and air temperature. The main in [20], for example). Isotopic data extracted from deep principles of modeling the isotopic composition of ice cores has allowed reconstructing paleo-temperature liquid, mixed and solid precipitation are also given. over about 120 ka in Greenland [60] and 800 ka in The relationship between isotopic composition of Antarctica [40]. precipitation and tempemture of its fo nnation is based On the other hand, the limitations of the isotopic on the "isotopic depletion" of moisture in the paleo-thermometer method are also known [2, 33 , 39, precipitating air mass due to isotopic fractionation 46. 54], which makes the absolute values of isotope­ during phase transfonnation of water. Since saturation temperature reconstructions somewhat uncertain. water vapor pressure is less for heavy water molecules 16 18 16 (HD 0 and H2 0) than for light molecules (H2 0), the

-299- concentration of heavy isotopes in a liquid phase is mean value of a from the beginning of the process to the higher than in a vapor phase equilibmted with this liquid. current moment. So, the isotopic content of water vapor of an air mass The equilibrium fmctionation coefficient a (aD for fomled over the ocean is reduced compared to that of deuterium and a l8 for oxygen IS) is by definition equal the ocean water. As the cooling of the air mass proceeds, to ratio of heavy isotope concentmtion in liquid to its the water vapor condenses and new portions of concentmtion in water vapor being in equilibrium with precipitation are enriched in heavy isotopes with respect the liquid: to the vapor remaining in the air mass thus making the R lil(!lid vapor more and more isotopically depleted (Fig. I). a = --- (always > I), (3) Obviously, in the course of further cooling both vapor R,·up-o, and condensate become isotopically lighter due to the and Can also be determined as mtio of saturation washing out of heavy water molecules during pressure of light molecules to that of heavy ones. In turn, precipitation formation. the fractionation coefficients are tempemture-dependent. The temperature functions of fractionation coefficients are discussed in (70]. condensation 1--<'I '\., Unlike the condensation in the atmosphere, the evaporation of water from the ocean takes place under non-equilibrium conditions characterized by under­ '"Of'O transport qoY,ransport G saturated water vapor above the sea surface. As a -to ,," .u- I ~ consequence, a "kinetic isotopic effect" that occurs due to the slower diffusion of heavy molecules leads to evapou"tion precipitation §if ?:~~~~...-----1 larger effective fractionation coefficients as compared _ -20"" ice sheet with those at equilibrium conditions. The first ::::: theoretical model satisfactorily describing the kinetic 0 .. effect was developed by Merlivat and Jouzel (1979) [57]. According to the authors, the isotopic composition ocean continent ) of water vapor forming over the sea surface, li,{), equals E UATOR POLE to: I I-k 8,., = (I + 8~ ) --- - I , (4) Figure I, Nalwal water cycle and isotope a I - kh Fac/iolla/ioll of oxygen 18 (Fom [20]) where ooe is the isotopic composition of the ocean water, Cl is the equilibrium fractionation coefficient, II is Thus, the isotopic composition of precipitation is the relative humidity of the air over the water surface proportional to the mtio F of current amount of water in and k is the coefficient that accounts for the molecular the air mass to its initial amount. In turn, F depends on and turbulent diffusivity of water vapor. The latter the difference of condensation temperature at the coefficient differs by 12 % for D and 180, and is current time and at the beginning of the distillation dependent on wind regime, though in some cases it can process. Isotopic composition of liquid precipitation (op) be took constant and equal to 0.006 (for oxygen IS) can be sufficiently well predicted by a Rayleigh [36]. distillation approach which is based on the assumption The equation (4) is based on the assumption that the that condensation takes place in the dynamic and air mass is fomled in a closed system with no vapor isotopic equilibrium and that new portions of being in the air before evaporation begins, which may condensate are removed immediately from the air mass be far from reality [II]. It was shown however that [15J' using this assumption for estimating the initial vapor 8 =.!!..... Fa.. -'_ I (I) isotopic composition leads to systematic differences , a, ' compared to simulations conducted with three dimensional isotopic-atmospheric general circulation where li (liD or liMO) is isotope composition models (IGeM) [54]. The drawback was corrected in expressed as ratio of heavy isotope concentmtion (mole fmction) in the sample (SA) to its concentration in [70] by replacing k by k· = k+ A(I-k), where A accounts standard water (ST), in per mil: for the involvement of outer water vapor into the area of air mass fonnation in addition to the water vapor 8 = R", - RST x 1000 (2) fonned by evaporation from ocean. Typical value of k* RST is 0.022 (for oxygen IS), which means stronger apparent kinetic effect during the air mass formation. One of the most suitable indexes of the kinetic effect intensity is "deuterium excess" which has been defined [15]as for IPO, and 0: is the fractionat ion coefficient at a drs = liD - Sli /80. (5) certain moment of time, O:Q is the same coefficient in the beginning of the condensation process and am is the

-300- Deuterium excess value in the air mass is proportional the condensation temperature (Tc), while experimental to the effective coefficient of isotopic fractionation isotopic data are ploned against 10 m fim temperature during the air mass fo rmation. Because du only slightly (TIO) which is conventionally assumed [51] to changes in the course of equilibrium condensation correspond to the mean annual surface snow process it must bear quantitative information about the temperature (Ts) and near-surface (measured at 2 m conditions in the moisture source [42]. height) air temperature (T2) , i.e., Tlo ::::: Ts::::: T2• In this The isotope model based on the above equations work we generally follow this assumption mainly due to explains satisfactorily the global relationships between the lack of the data, keeping in mind possible systematic mean annual values of isotope composition of offset between them [53). For Vostok, the related precipitation and air temperature [15]; uncertllinty may be as large as 2.5°C [50). ;J'BO = 0.7T - 13.6, (6a) According to Robin [67], the condensation oD = 5.6T - 100, (6b) temperature in Antarctica corresponds to the as well as between concentrations of JD and 15 '80 in temperature at the upper boundary of the inversion layer low and middle latitudes [9, 15]: (T,) within ± 4"C, i.e.: JD = 815 /80 _ 10. (7) Teo T,. (9) However, after perfonning the measurements of isotopic composition of snow samples collected along o~-L~ __L-~-L~ __~~-+ the inland Antarctic traverses [51] it became clear that 00, %0 the model does not work well for the snow falling in polar regions. Isotope models of Rayleigh type (for -100 equi librium conditions) gives too high values of drs thus pointing to another unknown kinetic effect during the formation of solid precipitation. The issue was resolved in 1984 by Jouzel and Merlivat [41] who -200 theoretically explained and empirically confirmed a kinetic effect during sublimation of water vapor on the surface of ice crystals in the air supersaturated with -300 water vapor. According to their RMK model (Rayleigh Model taking into account Kinetic effect) the effective coefficient of fractionation in this case equals to; -400 a.. = (l x a., where Cl is the coefficient of fractionation 1>I'r~ ,· K_ . _.I,!;.,.. • p""., Will. So"'" Pol • • V . . ..k for water vapor and ice in equilibrium, and tlk is the o o...... ,·41J,,"U • • n.... c kinetic fractionation coefficient: S. -60 +50 -40 +30 -20 -10 0 10 20 a " = ' (always < I), (8) I +a(S, -I)DI D' Temperature,OC where D and D' are the diffusion coefficients for light and heavy molecules, respectively, S, is supersaturation Figure 2. Isolope+lemperatllre diagram Jor preCipitation of air with water vapor with respeclto ice. and deposiled snow ill AlI/arClica. Dashed lilies are The uncertainty of the model prediction is mostly simple is%pe model simulatiolls (OOfTcJ [70] Jor /11'0 associated with value of SI which is generally unknown. inilial ocean slllface temperatures (TocJ . 10 alld 20"C. As an optimal solution of this problem the authors Symbols represelll isotopic vallles Jor the upper 1-2 m oj suggested to approximate S, as a function of An/arctic S IIOIV, collected along the three meridiollal condensation temperature choosing the coefficient of traverses (as indicated in the legend [J 4. 20. 51]) and the function by adjusting the model results to the plolted against JO-m jim temperature (Tw::::: TzJ . Bold observed distribution of isotope composition of snow, lilies are OOfT1 cllrves obtailledfrom correj,pondillg while keeping S, values in reasonable limits [41]. modeled OOfTc Cllrves /Ising a dTd dT) slope oJO.6. Within temperature range from about -5 to about -20"C all the three water phases coexist in clouds In tum, temperature inversion intensity becomes [68], which considerably complicates the isotopic weaker as one approllches the coast of Antarctica, which processes during precipitation fomlation. This means the T, T] difference is not constant [8, 64] (see complexity is not taken into account in the RMK. This also Fig. 2). Geographical relationship between the lack was filled up in so-called "mixed cloud isotope near-surface and 1I1verSlOn air temperatures was models" [5, 70]. obtained in [41] using the data from a number of The results obtained using a simple isotope model Antarctic sites with Tl ranging from -IS to -58"C: [70] are compared in Figure 2 with isotopic composition T,= C,r, - 1.2, (10) of surface snow samples collected II long the traverse where C1 = 0.67. routes Dumont-d'Urvilte - Dome C [51), Mirny - However, the condensation temperature in central KomsomolskaYIl [14, 20] and Patriot Hills - South Pole Antarctica may differ from T;. It has been shown [20] - Vostok [14]. Note that the model simulations refer to that weighted condensation temperature is considerably

-301- lower than that on the upper boundary of the inversion Several physical reasons may be invoked to explain layer, simply because the "clear-sky precipitation" these biases. State-of-the-art atmospheric GCMs do not (prevailing here) forms throughout the inversion layer. resolve very shallow inversion layers due to their Our calculations (10 be published elsewhere), confirmed representation of boundary layer processes; they usually by simple isotopic modeling [70], predict mean have no specific simulation of diamond dust fonnation weighted condensation temperature for Vostok Station and effects; the simulated atmospheric dynamics may be equal to _43 °C, which is 5°C lower than temperature on affected by grid point singularities at South Pole; their the top of the inversion layer [20]. [n the case of the representation of polar cloud distribution and coastal areas, inversion is weak and most of the microphysics may be problematic [54]. These precipitation is fomled III [ower clouds under drawbacks might be eliminated by incorporating water temperature which is not necessarily equal to mean­ isotopic transfomlations into the meso-scaled annual air temperature in the inversion layer. According atmospheric models with better geographical and to our estimations (to be discussed elsewhere), the vertical resolution (e.g., [31,48]). geographical near-surface/condensation air temperature In our study we lise a simple Rayleigh-type isotope slope for Antarctica is equal to 0.6. model developed by Salamatin et al. [70] on the basis of In Fig. 2 the relationship between the isotopic the theory of isotopic fractionation along the entire composition (JD) of surface snow in Antarctica and the trajectory from the air mass fonnation above the tro pical 10 m fim temperature (TIO::: T2) is also shown. Despite ocean to central Antarctica [6, 9, II, 13 , 15,27,34,36, systematic difference in isotopic composition of snow in 41, 47, 57. 80]. The model has some improvements various sectors of Antarctica. the slope 6.oD/6.T/Q in all compared to the previous studies. e.g., the involvement i cases is nearly the same and equals 6%0 0C- . Taking of the outer water vapor into the area of ai r mass into account above established 6.Tcl6.T2 ratio (0.6), this formation, optimization of microphysical processes value is consistent with the theoretical isotope­ approximations for mixed and solid clouds, as well as condensation temperature slope (Cr = 6.OD/6.Tc) of multi-trajectory scheme of simulations of isotope 10%0 0C- i produced by simple isotopic model (Fig. 2). If depletion in precipitation. The model was tuned using all model parameters except for the initial ocean surface experimental data on isotopic distribution in the whole temperature (Tod are kept constant. the difference in Antarctic continent (Fig. 3), along the Patriot-Hills - snow isotopic composition in different sectors of the Vostok traverse (Fig. 2), as well as on the seasonal Antarctic ice sheet implies variations of Toc from about isotope-temperature changes at Vostok Station (Fig. 5, 10 to 20°C. Section 4). The good coincidence between the experimental data A fler tuning, the model parameters are fixed (except and model predictions encouraged the use of the for those describing the source conditions, which could present-day geographical relationship between isotopic have changed in the past) and model is used for paleo­ composition of surface snow in Antarctica and mean climate reconstructions as presented in Section 5. annual near-surface air temperature for paleo­ temperature interpretation of the isotopic profiles from deep ice cores, which will be discussed in more details in Section 5. 6 ... 15 In spite of finn physical basis of "simple" Rayleigh­ ~ type isotope models, it can be argued that they do not }. ~ •.~ always adequately reproduce past changes in isotope ~ .', " ~ . composition of snow during such global climatic 10 o ' , "Q , transitions as that from LGM to Holocene [39]. This is X • • , , mainly due to the fact that simple models do not take , " • , , into account changes 111 atmospheric trajectories 5 .-..: t ·, ...... ~ • A A · ~rI. A (circulation), physical conditions in clouds etc. This is S why, since the late 1980s, a number of attempts have ~ • • been made to incorporate isotopic transfonnations in the A" course of the global water cycle into the General 0 Circulation Models (GCM), as well as into hydrological -400 -300 -200 -100 models of intennediate complexity (see review in [54]). Though GCMs basically well represent global SD (0/00 ) distribution of water isotopes in precipitation, as well as their seasonal cycle, they sometimes poorly describe Fig. 3. Comparisoll 0/drs /OJ) curve simulatiolls/or meteorology and isotopic distribution 111 inland Alllarctica with available experimelllal data (from [70}) Antarctica [54]. [n particular, two systematic biases are typical for GCMs: I) a lack of isotopic depletion fo r In this section we have briefly considered the central Antarctica, even in the models providing a theoretical basis of the isotope-temperature method of correct range of near-surface temperatures, and 2) an paleo-reconstructions and presented simple Rayleigh­ underestimation of moisture supply to inland Antarctica. type isotopic model adjusted to simulate the isotopic

-302- composition of precipitation in the polar regions. The In Fig. 4 the profile of isotopic composition of the isotope·condensation temperature geographical slope of upper 10-cm layer of snow is also shown. The negative 10%0 °Cl has been deduced for Antarctica, which seems correlation between this profile and that of to be in agreement with the experimental data. Next accumulation is remarkable (r = -0.68), with the slope l sections (3 and 4) will be devoted to the discussion of (regression coefficient) of -3.45 %0 cm· . Such

-303- processes, the stronger are post-deposition changes. It is varied, too. At present, it is not possible to quantify the expected that such changes in central Antarctica lead to innuence of this factor, but we assume it IS the enrichment of snow isotope composition [82], which comparatively small, because the past variations of is in agreement with the observed negative correlation temperature and accumulation rate caused opposite between isotopic content and snow build-up (Fig. 4). effects on the post-depositional isotopic changes [82]. It has been shown [22, 23] that the spatial variations In this section we discussed the local spatial of r5D and accumulatjon rate are reflected in time series variability of snow isotopic composition in the vicinity of these parameters due to the drift of the involved of Vostok Station, which appears to be very high and dunes. For example. the dunes shown in Fig. 4 are likely related to uneven snow re-distribution by wind. associated with the accumulation wave in temporal Corresponding "relief-related" variability in isotopic series with the length of about 20 years. We call such time-series accounts for roughly 80010 of total variance variations, not directly attributed to climate variability and therefore a 20-year running mean smoothing is of the studied snow characteristics, "relief-related" needed to isolate the climatic signal. Alternatively, if variations [21]. The overall noise in isotopic time-series several temporal series are available fo r the studied area due to these variations is estimated to represent about (as it is the case for Vostok), constructing stack series is 80010 of the total variance (which corresponds to a the best solution. If an ice core drilling site is located in signal-to-noise ratio of 0.25). a mega-dunes area, "relief-related" oscillation may be as In [21] several non-direct indications of presence of long as about 10J years. However, in the case of Vostok mega-dunes in the area of Vostok Station are given. The Station the resolution of deep ice core isotopic record corresponding "relief-related" temporal variations (500 years) is low enough to exclude the impact of attributed to mega-dunes that can be found in the time­ mega-dunes. series of accumulation rate and snow isotope composition are of the order of 1000 years in 4. Recent temporal variability of isotopic wavelength. This conclusion highlights the importance co ntent in the vicinity of Vostok Station of the results presented in this section for the ice core record interpretation. How to extract the climatic signal In this section we discuss the temporal vHriability of from the relief-related variations? In the case of a single the Vostok snow isotopic composition on the time-scale time-series available for an area, the smoothing of the from seasonal to centennial. Available isotopic data are series may be applied with the averaging window wide calibrated against instrumental temperature data, on enough to erase possible relief noise. For example, the seasonal (Section 4.1) and multi-year scale (Section 4.2). period of smoothing needed to suppress the influence of The calibration coefficients are then used to reconstruct meso-dunes (Fig. 4) at Vostok Station is 20 years [25]. temperature variations in the vicinity of Vostok Station Another method to separate climatic and relief-related over the past 200 years (Section 4.3), using the data variations is based on using band-pass filtering from deep snow pits, with the aim to demonstrate that techniques. Also, non-climatic oscillations can be isotopic composition may indeed be successfully used suppressed by constructing stacked series of studied for paleo-temperature reconstructions. parameter using the data from several pits (stakes, cores) due to the fact that relief-related temporal 4.1. Seasonal changes of snow precipitation isotopic variations are not generally correlated in adjacent sites composition located in the same area. See an example of such study In Figure 5 the mean monthly values of precipitation in Section 4.2. isotopic composition are shown as measured in Ihe Distinguishing between climatic and relief-related samples collected during the period from December temporal oscillations is straightforward and is based on 1999 to December 2000. The oD values reveal clear the following principles: I) relief-related temporal seasonal cycle (with an annual average of -453 %0) variations (unlike climatic ones) always have their consistent with the annual cycle of local air temperature. counterparts in the surface (spatial) profiles of studied The corresponding annual mean of 0/80 (not shown) parameter; 2) climatic variations are synchronous at all is -58.4 %0. the sites located in the area and thus they should be well Also shown in Figure 5 are seasonal changes of drs. preserved in the stacked series of studied parameter; 3) This parameter reveals minimum values in summer and additional confinnation of the climatic origin of maximum in winter, being in anti-phase with JD and air observed temporal changes in the climate proxy may be temperature. Such behavior of drs is mainly caused by obtHined through its comparison with aVHilable opposite trends of r5D and drs in ice clouds precipitation instrumentHI data (e.g., data on air temperature). (Fig. 3), but may also be related to seasonal changes of As for the post-depositional changes of snow isotopic the moisture source conditions (Tad. It is assumed that composition, the influence of this fHctor on the isotope­ summer minimum of deuterium excess in precipitation temperature paleo-reconstructions would be zero is due to the southward shift of the major moisture provided thHt those chHnges remained constant in the source [7,17,18, 54,81). This assumption is supported past. However, since both involved parameters by our simple isotope model [70], which suggests IhHt (temperature and accumulation rate) varied with time, the Hverage weighted summer tempemture for the the effect of the post-deposition changes may have

-304- source region is about J .s"C lower than the winter one, for the JDITe slope with our simple isotopic model [70]. likely due to reducing sea icc extent around Antarctica. We attribute this difference to the significant deviation The correlation coefficient between the mean monthly of the temperature on the top of the inversion layer from values of oD and near-surface air temperature (TA for the weighted condensation temperature. Our model the period December 1999 - December 2000 is simulation suggests that the observed winter-summer signi ficant and equals 0.89±0.14, whereas the regression oD changes in precipitation from -490%0 to -410%0, l slope equals 2.12±0.35 %0 "C . This value is accompanied by dxs shift from 22%0 to 7%0. may be considerably lower than corresponding geographical interpreted as Tc change from -4S"C to -4O"C, and TOC ' slope of 6 %0 °el discussed in Section 2. The difference change from 19.5"C to I soc. is partly explained by significant change of moisture source between summer and winter periods as pointed 4,2. Isotope-temperature calibration on multi-yea r above. But the main reason for the observed scale discrepancy is considerable seasonal changes of local In this subsection we di scuss the results of detailed conditions. in particular, the temperature inversion isotopic studies performed in snow pits dug in the strength (due to changes in radiation and heat balance), vicin iti es of Vostok Station. Understanding the which governs relationship between condensation and fonnation of isotopic composition of the upper snow near-surface temperatures. layers should give a clue for interpreting the isotopic profiles measured along the deep ice core. -3 0 The data discussed in this Section have been obtained u as a result of detailed snow thickness study in 6 shallow 0 -40 !! and 3 deep pits dug in 19S0-2000. In addition to isotope snow sampl ing, stratigraphic (Fig. 6) and geochemical , ->0 "Ie observations have been carried out, which allowed accurate dating of snow. as described in [20]. § -00 .. The main feature of the vertical profiles of snow · 400 -10 isotopic composition is the regular osci ll ations (Figs. 6 -420 and 8) with the total magnitude of about 60-S0 %0 for Of) (that is, about 70-90% of the seasonal change of .440 ~ isotopic composition of precipitation at Vostok, see ci -460 Section 4.1). Note that this value is larger than the '" magnitude of Of) change during LGM-Holocene -480 transition (about 50-55 %0 [63]). 30 -500 oJ ,., .." .; 20 0 0 "- 0 10 § ·0 £ 0 il ~ -10

o t 2 3 4 5 6 7 8 9 to \I 12 Momhs

Fig. 5. Seasonal variations o/isotopic composition (oD and drs) o/precipitation (so lid lines), blowing SIIOW ~ ~ (dashed lilies) al/d air temperature at Vostok Station. ...N " .. ...N " " ~-- c:J"'- ___ ~-.. Error bars represent slal/dard deviation (I 0') of isotope composition in the precipitation samples collected ill the same month (from [20)). Fig. 6. Straligraphic studies in 4 shallow pils in the vicinity of Vostok Station( (20). We cannot determine relationship between oD in prec ipitation and Tc or Ti, in 2000 because the balloon­ As shown in Secti on 3 (see also [20. 2 L 23]), these sounding observations have not been carried oul. To oscillations are mainly related to the drift of different estimate the isotope-temperature slope we took the types of snow surface relief fonns (micro-relief and mean monthly values of Ti for the period 1963- 1991 meso-dunes) mther than to climatic variations. The keeping in mind comparatively low inter-annual signal-to-noise ratio [28], detennined from the mean variability of this temperature [20]. Using these data and correlation coefficient of the J time-series from S JD values measured in 2000 leads to a JDITi slope of individual shallow pits studied in the vicinity of Vostok 6.2± 1.1 %0 oe ', which is lower than 10 %0 °C l obtained Station [23], is 0.25, which suggests that "stratigraphic

-305- noise" accounts for about 80 % of the total variance of than to the mean an nual temperature (Fig. 7). The snow isotope composition in a single point. This noise, correlation coefficient between OD and summer T2 is l typ ical for low-accumulation sites, was previously 0.92 with the slope of 10 .2±0.8 %0 °C . The latter value found when investi gating the snow accumulation at the corresponds to the predicted model slope between Vostok snow-stake network [22]. To reduce the noise, precipitation isotopic composition and condensation we constructed the stacked liD ti me-series using the data temperature (see Section 2). This cou ld be explained if from all the eight pits. In Fig. 7 we show this series for most of precipitation at Vostok occurs during the the period 1946- 1995 togethe r with the mean annual and summer, which is not the case [20, 25]. The other summer temperatures at Vostok Stati on, as obtained possibl e reason fo r this phenomenon is possibly strong from instrumental observations. post-deposition effects in this low-accumulation region In Section 3, we have established that the sensible of Antarctica, which considerably changes the initial resolution of the climatic 0 time-seri es obtained in a isotopic composition of snow precipitation and occurs single point at Vostok should not be better than 20 years. mostly in the warm part of the year. This assumpt ion is This period is needed to sufficientl y reduce the relief­ consistent wi th the fact that the Vostok snow pit related variations linked wi th the drifting meso-dunes isotopic record closely covariates wi th the "effective (Fig, 4). For the stacked series the corresponding period snow surface temperature" derived from the shallow was estimated to be 7 years [23]. Further smoothing borehole temperature analysis [50]. decreases the variance of the series insignificantly but does not increase the signal-to-noise ratio. 4,3. Vostok temperature history over th e past 200 According to Fig. 7, the isotopic composition of sno w years based on the isotope data from deep pits changed significan tl y during the last 60 years with the In this sub-section we use the isotope-temperature minimum in 1953- 1964 and ma ximum in the 1980s calibration established above to reconstruct the Vostok followed by a decrease during 19905 with assumed temperature history on centennial scale based on isotope minimum around 2000. The correlation (r) between data from deep pits. snow isotope composition and near-surface air In 25 'h and 45 'h summer seasons of Russian Antarctic temperature (for smoothed series) is statistically expedition snow stratigraphy was studied in three deep significant and eqw.lls nearly 0.6. Thus, the local snow pits dug at Vostok Station, vklO (1 980, 10 m temperature at Vostok accounts for about 40 % of the deep), vk99 and s/30 (1999/2000, both 12 m). Aside isotope inter-annual variability. The corresponding from stratigraphic observations, detai led (2-5 cm) regression coefficient is 17±4 %0 "C '. This value is isotope sampling was carried out in the two latter pits. almost 3 times larger than the geographical JDIT] slope. Snow thickness was dated based on layer counting Similar value (20.3 %0 "C ') was obtained by Jouzel and corrected by absolute age markers of 1955 , 1965 (high others [43] for the South Pote precipitation. beta-activity) and 181 6 (Tambora layer of high acidity) and by correlation to snow build-up at the accumulati on Slimmer stakes [20, 25]. The results of studies carried out in vk99 clIm. temp . 'f!mp . -, 4.4 and s130 are presented ill Fig. 8. -425 ·3 1.2 Based on the results of the deep pits studies, temporal series of sno w isotopic composition and acculllulation -410 _54.S ·31.6 rate have been re constructed for the last 225 years (1774-1999). Stacked oD and acculllulat ion (a) series -H 5 -S5.2 ·n are presented in Figure 9. One should note a high level ~ -.140 of stmtigraphic noise in the variability of both d parameters. The required smoothing to sufficiently • -55 .6 ·32.4 .44 5 suppress this noise is estimated to be II years for both oD and o. Note that for oD series the contribu ti on of ,4'0 ." ·32.8 high-frequency noise is less than for accumulation due to diffusive isotope smoothing (see also fading of the .455 -'6.4 ·n2 isotopic oscillations in the deeper parts of the profiles ..NO 1980 shown in Fig. 8) [28, 37]. ''''time (yun AD) '''' The reconstructed series of tempemture and accumu lation rate (Fig. 9) correlate with each other (r = th Fig. 7. Slacked record of liD ofsl/oll' il/ Ih e vicilli!y of 0.2 but statistically significant; for the 20 century /" = Vostok Slalioll based 011 the dalafrom 8 shallow pits 0.6), and also wi th the data presented in Fig. 7. (red) sholl'lI along with meall ollllllal (greell) alld mean In general, Vostok deep pit isotopic data suggest that slimmer (blue) remperalllres. All series are 7-year I I-year means of T2 have changed over the past 200 rulll/ing means. Diamolld depicts meall is%pe years within the range of loe (using the T.JOD slope composition ofsill/a ce SIIOW ill 2000. established above). The magnitude of corresponding changes of effective weighted condensation temperature Interestingly, the snow isotopic composition ap pears is about 2"C (according to the slope of 10%0 °C'), which to be much closer related to the summer temperature may likely correspond to variation of summer

-306- tempcerature rather than mean annual temperature (see temperature and accumulation rate. It has been shown also Fig. 7). [25] that these variations correlate with the so-called The pattern of small-scale climatic changes presented Pacific Decadal Oscillation (e.g., [77]), which suggests in Figs. 7 and 9 is likely common for the whole East a te1e-connection between centml Antarctica and Antarctica (see, for example, [59, 66]). According to tropical Pacific. M. Pourchet with co ~ autho r s [66], who obtained snow accumulation values in 14 sites by snow p-radioactivity 410 +----L~~----L-~~--~ measurements, the mean snow accumulation mte in East -54.5 -. -420 ~ Antarctica in 1965- 1977 was 30 % higher than during t -430 -55.0 ~ ~ the previous decade. -440 o -55.5 :-'" K) 450 l'k99 sr30 460 -56.0 3 -~ , ... 2 E o '" 1.:9 f-.-LT---~-'--r--'---r-~-+ 0 " '" 1750 t800 1850 1900 1950 2000 Y ears '" ..., Fig. 9, Stacked records ofJD alld accllmulatioll rate from deep pits. Thin lines are origillaitillle- series. whereas the thick olles represel1/ Ii-year mnllillg e '" meall.\". ([20). '"~15. '" In general, the comparison of the series of isotopic 8 composition and snow accumulation rate from deep pits "" with other Vostok and Antarctic data on JD, a and T allows to conclude that the reconstructed 200-year '" series (Fig. 9) reliably represent climatic variability in the studied area of Antarctica. For the present study this .. implies that isotopic composition of snow thickness can, indeed, be used for paleo-tempemture reconstructions. ''''' In this Section we summarized the data on temporal variability of snow isotopic composition at Vostok " .. Station and discussed its relationship with instrumentally obtained air temperature. On both '''' seasonal and multi-year time scales the strong d (""l '0-, correlation between OD and Tl is observed. The regression coefficient (slope) of this relationship is, Fig. 8. The results ofSIIOW rhicklless sllldies (allllllal however, significantly different from that of layer bOllndaries. isotope composition (oD alld dxs). geographical correlation (Section 2), which likely roral beta-radioactivity alld SIIOW liquid cOllductivity) ill reveals different near-surface/condensation temperature pits vk99 alld srJO. Values of (fllllllal snow build-lip at relationships for various temporal and spatial scales. stake 30 are corrected for SIIOW sel/ling. The hOl'izolltal This fac t implies that the present-day geogmphical lilies represent reference levels of 1955. 1965 alld 1816 OD/T} slope in Antarctica can hardly be directly used for paleo·temperHture reconstructions based on the deep ice (f1'01ll [20]). core data. The problem is additionally complicated by During the 1990s all the series re veal a clear decrease possible offset between fim temperature, that is used as of their values, especially well marked for air a proxi of near-surface air temperature T}, and T} itself. temperature. This agrees with cooling observed at the as mentioned in Section 2. On the other hand, we have most Antarctic stations (except for Antarctic Peninsula) not found any evidence that the theoretical oD/Tc slope during this period [19]. (10%0 "'cl, Section 2) does not hold true for the The observed changes are likely related to the temporal (seasonal to multi-year) isotopic variability, variations of cyclonic activity in Antarctic [26, 59, 73], which encourages using it for paleo-reconstructions. It since cyclones bring both moisture and heat to the has also been demonstrated that snow isotopic interior of the continent. composition al Vostok is ralher related to summer, than Another interesting feature of the 200-year Vostok to the mean annual temperature, which likely reflects climatic data is an apparent SO-year cycle of both not iceable posI·deposit ional changes of initial snow

-307- isotopic content. Finally, in this Section we used the l!. T. ~ l!.oD - pl!.o"O~ isotopic data from deep snow pits in order to reconstruct , C (" > local paleo-temperature over the past 225 years. The , correlation of this record with avai lable climatic data for where 6J" O« is correction for the changes of mean East Antarctica suggests that the isotopic profiles of isotopic composition of sea water in the past due to Antarctic snow can be used for paleo-climatological changes of water volume tmpped in Earth's glaciers. T/ studies. is related to the near-surface air temperature T} by equation (10) and is supposed to be a proxy of Tc (Eq. 5. Palco-temperature interpretation of 9). The coefficient p was first taken equal to 8 [63J Vostok deep ice core isotopic data (following Eq. (7». However, it was shown later that p has smaller value [46J due to the fact that the infl uence In this Section we usc the results obtai ned in the of isotopic change at the ocean surface weakens as an previous Sections to reconstnlct long-tenn temperature lIi r mass becomes isotopically depleted, which can changes using isotope record from the deep Vostok ice easi ly be demonstrated wi th a Rayleigh model. core. According to this approach, during LGM (00 is First, we no te that, as described in Section 3, a Hbo ut -483 %0, see fig. 10) T/ and T} were, respectively, de tai led isotopic time-series obtai ned in u single point 5.5 and 8.5°C lower than at present (00 IS (pit or ice core) refl ects "relicf-related" noise about -410 S\So). (represent ing 80% of the total variance) rather than The comparison of isotope method with the results of climatic signal. The wllvelength of this noise may be as independent palco-temperature approaches has shown large as 10J years in the case of mega-dunes influence. that the former likely underestimates the amplitude of For shallow cores, and for intennediate and deep cores past climatic chunges in the polar regions (see the drilled in the high-accumulation areas, where a high­ review in [20. 39, 46J). Among these independent resolution record is intended, the effect of relief-related approaches. the most elaborated one is the deep variations will be a critical issue. The results of shallow borehole thennometry method that allows extracting coring at Dronning Maud Land clearly iIIustmte this paleo-tempemture history from the ice sheet notion [61]. Individual cores located here in climatically tempemture fie ld (see the me thod overview in [69J). In unifonn area reveal quite different trends over the last part icular. for Greenland this me thod shows that LGM­ 200-year period. It underlines the fact that only Holocene surface snow temperature shift was about two constnlcting stack record on the base of the data from times larger than pred icted from the isotopic approach several cores may produce reliable climatic series for [1 2. 38]. This discrepancy has been explained by the area. considerable change of atmospheric circulat ion between The described effect however will unlikely be visible glacial and present time, accompanied by strong change in the low-resolut ion (500 years) Vostok isotopic record. of precipitation seasonality [83]. In part icular, in Another source of noise in isotopic series is related to Holocene the contribution of winter prec ipitation has the post-depositional changes of initial snow isotopic increllsed, whic h reduced apparent LGM-Holocene content (Section 3). The contribution of th is noise to the isotope shin recorded in ice core. deep ice core isotopic series is supposed to be relatively As for Antarctica, the difference between the two low because the past changes of temperature and methods shou ld be less (up to 30%) [69, 71, 78), likely accumulation mte te nd to compensate their influence on due to a wellker chnnge in precipitation seasonality [46]. the post-depositional processes. As it has been shown in Also, for Antarctica the borehole method is less Section 4, the values of the isotope/near-surface straightforward and requires very high precision of temperature slopes for the geogrdphical, seasonal and borehole tempemture data, which leads to a lower inter-annual variations arc different. In contrast, the accuracy of the obtained results [69]. Together with isotope-condensation temperdture slope appears to be uncertainty of the isotopic method, th is means that constant (about 10%0 ''(:"1) in all cases. The latter. paleo-temperature estimations produced by both however, is difficult to can finn empiricall y due to the methods for central Antarctica do not differ lack of the data on the condensation temperature. significantly [46]. The first attempt to quantitatively estimate past Another important factor responsible for the changes in ncar-surface air temperature using the ice difference between isotopic and bore-hole Ihennomelry core isotopic data was made in 1979 by C. Lorius and methods (or. in other words. between present-day others [52] on the basis of the present-day geographical geographic:11 and long-term temporal isotope­ temperature slopes) is changing conditions in the OITto slope. Twenty years after, the same approach with onl y minor corrections was used to reconstruct the air moisture source region [3, 20. 46]. temperature changes over the last 420 ka at Vostok As shown in Section 2 (see Eq. ( I», the precipi tation Station [63]. In the laller paper the ch.mges of inversion isotope composition is related to the source­ temperature in the past (compared to its present-day condensation temperature difference, mther than to the Tc alone. Thus, simultaneous climatic change in the va lue), 6Th are calculated fro m changes of isotopic composition of ice, 6JD: tropics and polar region reduces the isotopic content of the polar precipitu tion.

-308- The relationsh ip between geographical and te mporal The paleo-temperature reconstruction obtained by isotope·ternperature slopes can be expressed as: improved isotopic method (Eq. 13). as presented in Fig. (d/jldTd wmporol = (Cr + Cr.x!m), (12) 10. is consistent with the results of recently refined where Cr and Cue are geographical slopes between borehole themlometry method [78] (see also paper by precipitation isotope content (0), and condensation (Td Salamatin and others in thi s volume). We thus consider and source temperature (Toc ), while 11/ - '"mnplification these values (LGM and Holocene optimum IlT! factor" describing the ratio of the amplitudes of paleo.­ of -10 and 2 "C compared to present day) as the most climate change in polar regions and tropics [46]. For Cr re liable paleoclimatic estimations for the Vostok Station. l and Coe equal 10 and +2. 8 %0 "C , respectively, as The same LGM TJ has been recently deduced from the follows from the isotope model [70], the 15 % Dome C deep isotope record [40] using similar refined difference between geographical and temporal slopes isotopic method (Eq. 13). corresponds to 11/ value of 1.9, which seems to be This result implies that the moisture source influence reasonable [80]. is the most (and probably the only) important factor In order to account for the moisture source effect affect ing isotope-temperature paleo-reconstructions in using only ice core isotopic data, K. Cuffey and F. central Antarcti ca. Vimeux developed a method combining the data from We should note that direct comparison of the the both isotopes, OJ) and 8 80 [13 , 80]. The method is "isotope" and "borehole thermometry" methods of based on the fac l that both OJ) and deuterium excess paleo-reconstructions is somewhat problematic, since (drs) are dependent on both condensation and source the first deals with condensation temperature Tc (which temperatures. Solving th e system of two linear is further transformed to near~surface air temperature TJ, equations one can come to the following formulas [80]: as demonstrated in Section 2), while the second does 18 6Toc = 0600 + Mdxs + c65 0"" (13a) with "effective snow surface temperature" , which 6Tc"" d65D + e6drs + f65 I ~ O "", (l3b) roughl y corresponds to Ts. We should also note that where the linear coefficients can be derived from the effective condensation temperature (Tc) may not be simple isotopic mode l. The application of this method to eq ulli to mean annual air temperature at any atmospheric the Vostok isotopic data gives 20 % stronger amplitude level. As an example, in Section 4 it has been of LGM-present day temperature shift than "classical demonstrated that snow isotopic composition closer isoto pic method", which on ly uses geographical olT re lated to summer temperature, which likely reflects the slope (Eq. 11 )[63]. post-depositional changes of initial snow isotopic As a result, using the method described by Eq. (13) content. In this section we have demonstrated that "classical and applying coeffi cients obtained from our simple isotopic method" (based on the present-day isotope model [70], we estimate that Tc and TJ during LGM time at Vostok Stat ion were, respectively, about 6 geographical isotope-temperature slope) is not able to and IO"C lower than their present-day va lues. correctly reproduce the amplitude of past temperature changes in Greenland and Antarctica, which is evident The isotope-paleotempemture reconstruction based on from the results of the a lternative approach, borehole Vostok ice core data is presented in Fig. 10. thermometry. For Antarctica, the mai n reason for this 410 6 discrepancy is the past changes in the moisture source conditions. The refined isotopic method has been -420 4 appli ed instead, wi th the moisture-source correction based on the use of drs data. Assuming that Vostok 2 -430 deep ice core isotopic record reliably represents climatic 0 signal (i. e., relief+related or post-deposit iona l noise is 440 IlTJ -2 negligible), we obtain the following estimations of ..:' in the past (comparing to present-day value): _lOuC for 0' -4>0 '" -4

-480 -10 6. Isotopic co mposition of the basal part of the Vostok ice core -490 -12 ISO 100 >0 0 Below 3538 m the Vostok ice core comprises frozen Time OP, ka wa ter of subglacial lake Vostok [44). A number of studies have been devoted to the isotopic [24. 62, 75] Fig. 10. DellferiUIII COlllellt ill Vostok deep ice core microbiological [4). gas [49, 56] and chemical [16, 55, (red) alld il/fen·ed paleo-temperawre (Ty (black) Jo/" 72, 76) content of the lake ice, as well as to its physical the last 150 ka. properties [58]. The scheme of water balance in Lake Vostok and lake ice formation is supposed to be as follows [45, 74, 75].

-309- The lake has two main water sources, glacier melt and low drs of Vostok lake ice is considered to be one of the hydrothennal water. Melt water comes from the evidences for the presence of the hydrothennal water northern part of the lake where energy balance allows sources in the lake [62], as hydrothermal water is glacier ice to fuse. Fresh melt water is lighter than usually enriched in oxygen 18 [10, 35]. sli ghtl y salt y deeper lake water, so it seeps southward The whole thickness of the lake ice may be divided along the tilted glacier sole, mixing with lake water. [n into 2 layers, lake ice I and 2, with the boundary at the southern part of the lake (where Vostok Station is 3609 m. For lake ice I the high concentration of visible situated) the freezing takes place involving two main mineral inclusions is typical, while ice 2 is extremely mechanisms, fruzil crystal formation and slow freezing clean [44]. It is supposed that ice I was fonned in the in equilibrium with ice. shallow part of Lake Vostok, probably in a strait In [24) we developed a simple model describing the between western shore and small island upstream of isotopic composition of lake ice and taking into account Vostok Station (65 ], whereas ice 2 likely originates possible non-steady-state of lake Vostok, not complete from the deep central-southern part of the lake. mixing of glacier melt water with the deeper lake water, At first glance, from isotopic point of view, the ices I and the presence of the second lake water source and 2 are very similar in temlS of mean isotopic content (hydrothennal) in addition to glacier melt [4, 62]. and variability (Fig. II). The only difference between At present, the depth of the deep 5G-l borehole at them seen by now is comparatively poor OD_8 80 Vostok is 3668 m. Ice core down to 3650 m is being correlation in ice 2. This phenomenon is still awaiting now investigated, and here we present isotopic profile for ex planation. down to the depth of3623 m (Fig. II). Using lake Vostok isotopic model (24] it is possible to estimate the isotopic content of the lake water. The 3D, lV,o d,W. latter depends on several factors. but for the most likely -44~ -444 -443 -412 -441 ~ 6 7 8 \I scenario of lake ice formation (50% of fruzil crystals 3'" fonned under reasonable super-cooling conditions and sufficient mixing of melt-water with residence lake water) it is characterized by JD of -450 W>o and 8 80 3m ._ ....._ ... of -58 W>o. , 7. Conclusions ! 3580 In this paper we discussed the main factors responsible for the formation of the isotopic composition of an ice core. The overview of theory and a range of experimental dllta strongly suggests that the iet' I snow (ice) isotopic content is a proxy of the temperature icc} conditions that occurred during the precipitation and

3620 deposition of this snow in the remote past. On the other hand, the quantitative temperature interpretation of the ·56.6 -~6.4 ·.s 6. 2·56.0-~~.8 isotopic record is not straightforward. On the short time 0"0, $0 scales a number of factors (local meteorology, wind­ driven snow-redistribution, post-depositional processes, Fig. II. Isotopic compositioll oJrlle Lake Vostok ice as etc.) may alter or weaken the isotope-temperature measllred 011 rile Vostok ice core (3538-3623 /II). relationship. On the longer time-scale, the change of moisture source conditions and atmospheric circulati on The main feature of the lake ice isotopic profile is its may affect the isotopic composition of precipitation in extremely low va riability, about 1-2 orders of the polar regions. In particular, for Vostok Station the magnitude less than for meteoric snow and ice, which difference between present-day geographical isotope­ requires very high measurement prec ision. The highest temperature slope and corresponding long-tenn variability is characteristic for upper section of lake ice temporal slope is as much as about 20% and mlly be profile, which was fonned in the conditions of strong explained by the past changes of the temperature in the water super-cooli ng, due to contact of cold glacier with tropical oceans where the precipitating air mllss has relatively wann lake water [78]. been fonned. Another feature of the lake ice is very low deuterium In the last section of the paper we have briefly excess va lue: 7-8W>o instead of 15W>o typical for the reviewed the formation of the isotopic composition of Vostok meteoric ice. However, if the lake is in steady­ the lower part of Vostok ice core that comprises frozen state, the isotopic composition of income water (glacier water of subglacial lake Vostok . melt) and outcome water (lake ice) has to be the same. Thus, the low deuterium excess in lake ice suggests non-equilibrium state of the lake Vostok, or, more likely, the presence of an additional water source. Actually,

-310- Acknowledgements Stable isotopes in oceanographic studies and paleotemperatures, Pisa, 1965, Consiglio Nazionale We highly appreciate valuable comments from an della Ricerche, Laboratorio di Geologia Nucleare, anonymous reviewer that allowed us to improve the pp. 9~130. manuscript. [12] Currey K.M. , Clow G.D. , Alley R.B. , Stuiver M., This work is the part of Project 2 "Vostok ice core, Waddington E.D., Saltus R.W., "Large Arctic paleo-climate and subglacial lake Vostok" of the temperature change at the Wisconsin-Holocene Russian Federal Targeted Program "Antarctica" and glacial transition", Sc ience, 270, 1995, pp. 455-458. supported by Russian Foundation for Basic Research gran ts 05-05-66803 and 05-05-65248. The A.E.'s stay [13] Cuffey K.M., Vimeux F. , "Covariation of carbon in IL TS is supported by Japan Society for Promotion of dioxide and temperature from the Vostok ice core Science (JSPS). after deuterium-exess correction", Nature, 412, 200 I, pp.523-527. References [14] Dahe Q. , Petit J.R., Jouzel J. , Stievenard M., "Distribution of stable isotopes in surface snow along the route of the 1990 International Trans­ [ I] Black H.P. , Budd W., "Accumulation in the region Antarctic Expedition", J Glaciol., 40{l34), 1994, pp. of Wilkes, Wilkes Land, Antarctica", J. Glaciol., 107-118. 5(37), 1964, pp. 3-15. [15] Dansgaard W. , "Stable isotopes in precipitation", [2] Blunier T., Schwander J., Chappellaz 1. , Parrenin F. , Tellus, 16, 1964, pp. 436-468. Bamola 1M., "What was the surface temperature in central Antarctica during the last glacial maximum?", [16] de Angelis M., Morel-Fourcade M.-C., Bamola Earth and Planetary Science Letters, 218, 2004, pp. J.M., Susini J., Duval P., "Brine micro-droplets and 379-388. solid inclusions in accreted ice from Lake Vostok (East Antarctica)", Geophys. Res. Lett., 32(L 1250 I), [3] Boyle E.A., "Cool tropical temperature shift the 2005 , pp. 1-4. global 8 1~ 0-T relationship: An explanation for the ice core 180 - borehole thermometry conflict", [17] Delmolle M. , Enregistrements climatiques a Law Geophys. Res. Lett., 24(3), 1997, pp. 273-276. Dome: Variabilite pour les periodes recentes et pour la deglaciation, These Doctorat de l'Universite [4] Bulat SA, Alekhina LA. , Blot M., "DNA signature Joseph Fourier, 1997, Grenoble, 115 p. of thermophilic bacteria from the aged accretion ice of Lake Vostok, Antarctica: implications fo r [18] DelmOlle M., Masson V., Jouzel J., Morgan V. L, searching life in extreme icy environments", Intern. "A seasonal deuterium excess at Law Dome, coastal Astrobiology, 3( 1), 2004, pp. 1-12. eastern Antarctica: A southern ocean signature", 1. Geophys. Res., 105(D6). 2000, pp. 7187-7197. [5] Ciais P., Jouzel J., "Deuterium and oxygen 18 in precipitation: Isotopic model, including mixed cloud [19] Doran P.T., Priscu J., Lyons W.B., Walsh J. E., processes", J. Geophys. Res., 99(08), 1994, pp. Fountain A.G., McKnight D.M., Moorhead D.L., 16,793-16,803. Virginia R.A. , Wall D.H. , Clow G.D .. Fritsen C.H., McKay c.P., Parsons A.N. , "Antarctic climate [6] Ciais P., Jouzel 1., Petit 1 R., Lipenkov V., White cooling and terrestrial ecosystem response", Nature, J.W.c., "Holocene temperature variations inITered 415, 2002 , pp. 517-520. from six Antarctic ice cores", Ann. Glaciol., 20, 1994, pp. 427-436. [20] Ekaykin A.A. , Meteorological regime of central Antarctica and its role in the formation of isotope [7] C iais P. , White 1C.W., Jouzel J , Petit JR .. "The composition of snow thickness, These Doctorat de origin of present day Antarctic precipitation from l' Universite Joseph Fourier, 2003 , Grenoble, 136 p. surface snow deuterium excess data", J. Geophys. Res. , 100(9), 1995, pp. 18917-18927. [21] Ekaykin A.A., Lipenkov V.Y. , ""Relief-related" oscillations of snow thickness characteristics in [8] CO ll nolley W.M. , "The Antarctic temperature Antarctica [Rel'efo-obuslovlennye kolebaniya inversion", Int. Journal Clim, 16, 1996, pp. 1333- harakteristik snezhnoy tolschi v Antarktide], in 1342. Russian" , Materialy Glyatsiologicheskih [9] Craig H. , "Isotopic variations in meteoric waters", Issledovaniy, 97 , 2004, pp. 35-43. Science, 133 , 1961 , pp. 1702-1703. [22] Ekaykin A.A., Lipenkov V.Y. , Barkov N. L. [10] Craig H. , "The isotopic geochemistry of water and "Spatial and temporal stmcture of snow carbon in geothennal areas", Nucelar Geology in accumulation field in the vicinity of Vostok Station, Geothennal Areas, E. Tongiorgi. Editor, Spoleto, central Antarctica [Prostranstvenno-vremennaya 1963, pp. 17-53. struktura polya snegonakopleniya v rayone stantsii Vostok, tsentral'naya Antarktida). in Russian", [II) Craig H. , Gordon L.l., "Deuterium and oxygen-I 8 Vestnik SPbGU, ser. 7, 4(28), 1998, pp. 38-50. variations in the ocean and the marine atmosphere",

-311- [23] Ekaykin A.A ., Lipenkov V.Y. , Barkov N.L, Petit [36] Johnsen S., Dansgaard W. , White 1 W.C, "The 1 R., Masson-Oelmolte V., "Spatial and temporal origin of Arctic precipitation under present and variability in isotope composition of recent snow in glacial conditions", Tellus, 41 B, 1989, pp. 452-468. the vicinity of Vostok Station; Implications for ice· [37] Johnsen S.1. , Clausen H.B., Currey K.M., core interpretation", Ann. Glaciol., 35, 2002, pp. Hoffmann G. , Schwander 1 , Creyts T., "Diffusion 181 · 186. of stable isotopes in polar tirn and ice: the isotope [24J Ekaykin A.A. , Lipenkov V.V., Johnsen S., effect in firn diffusion", Physics of Ice Core Records, Prilepsky E.B., Shibaev Y.A., "Isotopic composition Hondoh T., Editor, Sapporo, 2000, Hokkaido of subglacial lake Vostok water [lzotopnyi sostav University Press, pp. 121- 140. vody podlednikovogo ozera VostokJ, in Russian", [38] Johnsen S.l, Dahl-Jensen D., Dansggard W., Materialy Glyatsiologicheskih Issledovan iy, 98, Gundestmp N. , "Greenland palaeotemperatures 2004, pp. 92-99. derived from GRIP bore hole temperature and ice [25] Ekaykin A.A., Lipenkov V.Y., Kuzmina LN., Petit core isotope profiles", Tellus, 47B, 1995 , pp. 624- 1 R., Masson-Delmotte V., Johnsen S. , "The changes 629. in isotope composition and accumulation of snow at [39] Jouzel J., Alley R.B. , Cuffey K.M. , Dansgaard W., Vostok Station over the past 200 years", Ann. Grootes P. , Hoffmann G., Johnsen S.l, Koster R.O., GlacioL , 39, 2004, pp. 569-575. Peel D. , Shuman CA., Stievenard M. , Stuiver M., [26] Enomoto H., "Fluc tuations of snow accumulation White 1 , "Validity of the temperature reconstruction in the Antarctic and sea level pressure in the from water isotopes in ice cores", J. Geophys. Res., southern hemisphere in the last 100 years", CH m. 102(CI2), 1997, pp. 26471-26487. Change, 18 , 1991 , pp. 67-87. [40] Jouzel J., Masson-Delmolle V., Callani 0., Dreyfus [27J Fisher O.A. , "Remarks on the deuterium excess in G., Falourd S. , Hoffmann G. , Minster 8., Nouet J" precipitation in cold regions", Tellus, 43B, 1991. pp. Bamola 1 M .. Chappellaz 1 , Fischer H., Gallet J.C, 401-407. Johnsen S., Leuenberger M. , Loulergue L , Luethi D., Oerter H.. Parrenin F., Raisbeck G .. Raynaud D" [28J Fisher O.A., Reeh N., Clausen H.B., "Stratigraphic noise in time series derived from ice cores", Ann. Schilt A., Schwander J. , Selmo E., Souchez R., Glaciol., 7, 1985, pp. 76-83. Spahni R., Stauffer B., Steffensen 1 P. , Stenni B., Stocker TF., Tison J.L, Wemer M., Wolff E.W., [29] Frezzotti M., Gandolfi S., La Marca F., Urbini S., "Orbital and millenia I Antarctic climate variability "Snow dunes and glazed surfaces in Antarctica: new over the past 800,000 years", Science, 317, 2007, pp. field and remote-sensing date" , Ann. Glaciol., 34, 793-796. 2002, pp. 81-87. [41] Jouzel J., Merlivat L , "Deuterium and oxygen 18 [30] Frezzotti M., Gandolfi S., Urbini S., "Snow in precipitation: modeling of the isotopic effects megadunes in Antarctica; Sedimentary structure and during snow formation", 1. Geophys. Res., 89(D7), genesis", J. Geophys. Res., 107(DI8), 2002, pp. 1984, pp.11749- 11757. ACL 1-12. [42] Jouzel J. , Merlivat L , Lorius C , "Deuterium [31] Gallee H., Guyomarc'h G., Brun E., "Impact of excess in an East Antarctic ice core suggests higher snow drift on the Antarctic ice sheet surface mass relative humidity at the oceanic surface during the balance; Possible sensitivity to snow-surface last glacial maximum" , Nature, 299(5885), 1982, pp. properties", Boundary Layer Meteorology, 99, 2001 , 688-591. pp. 1~19. [43] Jouzel J., Merlivat L , Petit J.R., Lorius C., [32] Gow AJ., Rowland R., "On the relationship of "Climatic infonnation over the last century deduced snow accumulation to surface topography at "Byrd from a detailed isotopic record in the South Pole Station", Antarctica" , J. GlacioL, 5(42), 1965, pp. snow", J. Geophys. Res., 88 , 1983 , pp. 2693-2703. 843-847. [44] Jouzel 1 , Petit J.R. , Souchez R. , "More than 200 m [33) Helsen M.M., On the intemretation of stable of lake ice above subglacial Lake Vostok, isotopes in Antarctic precipitation, 2006, University Antarctica", Science, 286, 1999, pp. 2138-2 141. of Utrecht, Utrecht, 152 p. [45] Jouzel J. , Souchez R" "Melting-refreezing at the [34] Hendricks M.B.. Paolo 0.10., Cohen R.C., "Space glacier sole and the isotopic composition of the ice", and time variations of S180 and SO in precipitation: J. GlacioL, 28(98), 1982, pp. 35-42. can paleotemperature be estimated from ice cores'?", [46] Jouzel J., Vimeux F., Caillon N., Delaygue G., Global Biogeochem. Cycles, 14(3), 2000, pp. 851 - Hoffmann G. , Masson-Delmotte V. , Parrenin F., 861. "Magnitude of isotope/temperature scaling for [35] Hoefs J., Stable Isotope Geochemistry, 2004, interpritation of central Antarctic ice cores", J. Springer, 244 p. Geophys. Res., 108(012), ACL 6, 2003, pp. 1-10.

-312- [471 Kavanaugh J.L. , Cuffey K.M., "Generalized view [57] Merlivat L. , Jouzel J.. "Global climatic of source-region effects on 0 and deuterium exess interpretation of the deuterium-oxygen 18

of ice 4 sheet precipitation", Ann. Glaciol., 35, 2002, relationship for precipitation", J. Geophys. Res., pp. 111 4117. 84(C8), 1979. pp. 5029-5033. [48] Krinner G., Magand 0., Simmonds I., Genthon C , [58] Montagnat M., Duval P., Bastie P. , "High Dufresne J.L., "Simulated Antarctic precipitation crystalline quality of large single crystals of and surface mass balance at the end of the twentieth subglacial ice above Lake Vostok (Antarctica) century", Clim. Oyn., 28,2007, pp. 2 I 5-230. revealed by hard X-ray diffraction", Earth Plane!. Sci. Lett. , 333, 2001 , pp. 419-425. [49] Lipenkov V., lstomin VA, "On the stability of air clathrate-hydrate crystals in subglacial lake Vostok, [59J Morgan V.l., Goodwin 1. 0., Etheridge D.M., Antarctica", Mater. Glyatsiol. lssled, 91 , 2001, pp. Wookey C.W., "Evidence from Antarctic ice cores 138-149. for recent increases in snow accumulation", Nature, 354,1991, pp. 58-60. [50J Lipenkov V.Y ., Shibaev Y.A., Salamatin A.N. , Ekaykin A.A., Vostretsov R.N., Preobrazhenskaya [60] NOI1hGRlP, "High-resolution record of Northern A.V., "Present-day climatic changes recorded in Hemisphere climate extending into the last glacial temperature variations in the upper 80-m ice period", Nature, 431, 2004, pp. 147-15 L thickness at Vostok Station [Sovremennye [61] Gerter H., Wilnelms F., Jung-Rothenhausler F., kJimaticheskie izmeneniya, zaregestrirovannye v Goktas F.. Miller H., Graf W., Sommer S., variatsiyah temperatury verhnego 80-metrovogo "Accumulation rates in Dronning Maud Land, sloya lednikovoy tolschi na stantsii Vostok], in Antarctica, as revealed by dielectric-profiling Russian", Materialy Glyatsiologicheskih measurements of shallow firn cores", Ann. GlacioL, Issledovaniy, 97, 2004, pp. 44-56. 30,2000, pp. 27-34. [51 Lorius C, Merlivat L., "Distribution of mean J [62] Petit J.R., Alekhina LA., Bulat S.A .. "Lake Vostok, surface stable isotope values in East Antarctica: Antarctica: exploring a subglacial lake and observed changes with depth in the coastal area", searching for life in an extreme environment", IAHS publications, 118, 1977, pp. 127-137. Lectures in Astrobiol0G:Y, M. Gargaud, et aI. , Editors, [52} Lorius C, Merlivat L. , Jouzel J., Pourchet M. , "A 2005, Springer, pp. 227·288. 30,000 yr isotope climatic record from Antarctic [63] Petit J.R., JouzeJ J. , Raynaud D., Barkov N. L, ice", Nature, 280, 1979, pp. 644-648. Bamola J.M., Basile I., Bender M., Chappellaz J., [53] Mann M.E., Schmidt G.A., "Ground vs. surface air Davis M., Delaygue G., Delmolle M., Kotlyakov temperature trends: Lmplications for borehole V.M., Legrand M. , Lipenkov V.Y., Lorius C, Pepin surface temperature reconstructions", Geophys. Res. L. , Ritz C, Saltzman E., Stievenard M. , "Climate Let!., 30( 12), 2003, pp. 9 1-4. and atmospheric history of the past 420,000 years from the Vostok ice core, Antarctica", NalUre, [54J Masson-DeLmOlle V., HOll S., Ekaykin A.A., Jouzel 399(6735), 1999, pp. 429-436. J., Aristarain A., Bernardo R.T. , Bromwich D., Cattani 0., Delmotte M., Falourd S., Frezzolli M., [64] Phillpot H.R. , Zillman J.W., "The surface Gallee H., Genoni L. , Isaksson E., Landais A., temperature inversion over the Antarctic continent", Helsen M., Hoffmann G., Lopez J., Morgan V., J. Geophys. Res. , 75(2 I), 1970, pp. 4161-4169. Motoyama H., Noone D., Oerter H., Petit J.R., [65] Popov S.V., Masolov V.N., "Forty-seven new Royer A., Uemura R. , Schmidt G.A., Schlosser E., subglacial lakes in the 0-110* E sector of East Simoes lC, Steig E., Stenni B., Stievenard M., van Antarctica", J. Glacio1. , 53(181),2007, pp. 289-297. den Broeke M., van de Wal R. , van den Berg W.-l, Vimeux F., White J.W.C., "A review of Antarctic [66] ['ourchet M., Pinglot F., Lorius C., "Some surface snow isotopic composition: observations, meteorological applications of radioactive fallout atmospheric circulation and isotopic modelling", J. measurements in Antarctic snows", J. Geophys. Res. , Clim., 21(13), 2008, pp. 3359-3387. 88(CIO), 1983, pp. 6013-6020. [55] Mayer C., Grosfeld K. , Siegert MJ., "Salinity [67] Robin G.d.Q., "Ice cores and climatic change", impact on water flow and lake ice in Lake Vostok, Philosoph. Transact. Royal. Soc, B 280, 1977, pp. Antarctica", Geophys. Res. Let!., 30(14), 2003, pp. 143-168. OCE 8 1-4. [68] Rogers R.R. , A short course in cloud physics, New [56] McKay CP., Hand K.P. , Doran P.T., Andersen York, 1979, Pergamon, 253 p. D.T., Priscu J.C, "Clathrate formation and the fate [69] Salamatin A.N., "Paleoclimatic reconstructions of noble and biologically useful gases in Lake based on borehole temperature measurements in ice Vostok, Antarctica", Geophys. Res. Lett. , 30(13), sheets", Physics of Ice Core Records, T. Hondoh, 2003, pp. 35 1-4. Editor, Sapporo, 2000, Hokkaido University Press, pp.243-282.

-313- [70] Salamatin A.N., Ekaykin A.A. , Lipenkov V.Y. , [78J Tsyganova E.A. , Salamatin A.N., "Non-stational)' "Modelling isotopic composition in precipitation in temperature field simulation along the ice flow line Central Antarctica", Mater. Glyatsiol. lssled, 97, "ridge B - Vostok Station", East Antarctica", 2004, pp. 24-34. Materialy Glyatsiologicheskih Issledovaniy, 97, 2004, pp. 57-70. [71J Salamatin A.N., Lipenkov V.Y., Barkov N. J. , Jouzel 1. , Petit J.R. , Raynaud D., "Ice-core age [79J Van der Veen C,J., Mosley-Thompson E., Gow A., dating and paleolhennomeler calibration on the Mark B.G., "Accumulation at Soulh Pole: basis of isotope and temperature profiles from deep Comparison of two 900-year records", 1. Geophys. boreholes at Vostok Station (East Antarctica)", 1. Res., 104(D24), 1999, pp. 31067-31076. Geophys. Res., 103(D8), 1998, pp. 8963-8977. [80] Vimeux F., Currey K.M., Jouzel 1. , "New insights [72] Siegert M.l, Tranter M., Ellis-Evans J.C., Priscu l, into Southern Hemisphere temperature changes Lyons \V.B., "The hydrochemistry of Lake Vostok from Vostok ice cores using deuterium exess and the potential for life in Antarctic subglacial correction", Earth Planet. Sci. Lett. , 303, 2002, pp. lakes", Hydro!. Process, 17,2003, pp. 795-814. 829-843. [73] Simmonds I., Keay K., "Variability of Southern [8 1] Yimeux F. , Masson Y., Jouzel J., Slievenard M., Hemisphere eXl ratropical cyc lone behavior, 1958- Petit l.R., "Glacial-interglacial changes in ocean 97", l CHm., 13,2000, pp. 550-561. surface conditions in the Southern Hemisphere", Naturc,389,1999,pp.410413. [74] Souchez R., Jouzel J., "On the isotopic composition in dD and dl80 of waler and ice during freezing", J. [82] Waddington E.D. , Steig E.1., Neumann T.A., Glaciol., 30(106), 1984, pp. 369-372. "Using characteristic times to assess whether stable isotopes in polar snow can be reversibly deposited", [75] Souchez R., Petit J.R., Tison J.L., Jouzel l, Ve rbeke V., "Ice fonnation in subglacial Lake Ann. Glaciol., 35, 2002, pp. 118-124. Vostok, Central Antarctica", Earth and Planetary [83] Werner M., U. Mikolajewicz, Heimann. M., Science Leiters, 181 ,2000, pp. 529-538. Hoffmann. G., "Borehole Versus Isotope Temperatures on Greenland: Seasonality Does [76] Tranter M., Royston-Bishop G.W., Siegert M.J., Matter", Geophys. Res. Lett., 27(5), 2000, pp. 723- Bates P.D. , Lee V .. "Is Lake Vostok in chemical and 726. physical steady-state?", Geophysical Research Abstracts, 5(02706), 2003, p. 2. [84] Whillans I.M ., "Effect of inversion winds on [771 Trenberth K.E., Hurre ll J.W., "Decadal topographic detail and mass balance on inland ice atmosphere+ocean variations in the Pacific", Clim. sheets", J. Glaciol., 14(70), 1975, pp. 85-90. Dyn. , 9, 1994, pp. 303·319.

-314-