Can we retrieve a clear paleoclimatic signal from the deeper part of the EPICA Dome C ice core ?

Introduction Conclusions References

reported against VSMOW values.Inter-comparison of reference waters among the 28 involved laboratories were conducted over the analysis period and the same water standard 29 was used in both LSCE and DST/DMG laboratories.30

S1.3 Total gas content, Gas Mixing ratios and isotopes 31
The measurements of the total gas content have been performed at LGGE using an 32 original barometrical method implemented with an experimental setup called STAN 33 (Lipenkov et al., 1995).This technique allows a precise evaluation of the pressure and 34 temperature of the air extracted from an ice sample having a mass of 20-30 g by its 35 melting-refreezing under vacuum in a volume-calibrated cell.After correction of the 36 measured pressure for the partial pressure of saturated water vapour and of the 37 calibrated volume for the volume occupied by refrozen bubble free ice, the gas content 38 V is calculated using the ideal gas law.The V values are then corrected for gas loss from air inclusions (i.e.gas hydrates and relaxation features such as gas cavities at 40 depths under consideration) cut at the surface of the sample (Martinerie et al., 1990).41 The absolute precision of the STAN measurements has been estimated to be within 42 ±0.6%.However the overall error of obtained V values amounts 1% because of the 43 uncertainties in the cut-bubble correction.The replication of the results estimated by 44 repeated measurements in the same horizontal slice of an ice core has been confirmed 45 to be better than 1%.46 For CO 2 measurements at the University of Bern, ice samples of about 7 g were 47 pulverized in a cooled and evacuated vacuum chamber using a needle cracker 48 principle.After this dry extraction process, the extracted air was expanded from the 49 vacuum chamber into an infrared laser spectrometer which is used to derive the CO 2 50 concentration.At the LGGE about 40g of ice were pulverized in a cooled vacuum 51 chamber using a ball mill principle.The composition of the oxygen 18 in entrapped air, which gives access to the isotopic 70 composition of the atmosphere, 18 O atm , has been measured at LSCE using the melt-71 freezing method for air extraction followed by mass-spectrometry measurement as 72 described in Landais et al. (2003) and Dreyfus (2008).The measurement precision for 73 the 2004-2005 dataset is 0.02 ‰ (pooled standard deviation).

S1.4 Crystal size 75
Crystal sizes in the Deep and Basal ice were determined in the field, using a simple 76 version of the linear intercept method (Pickering, 1976).In this method, the number (N) 77 of grain boundaries crossed by a random linear traverse of length (N) across the thin 78 section is averaged over many traverses.The mean grain diameter (d) is then 79 estimated as / .At NEEM, crystals with cross sections up to 40 cm were 80 observed, with occasional bands of smaller crystals.We therefore simply counted the 81 amount of crystals along a center line of a 3mm thick slab observed through polarized 82 light, and divided by the length of the core (55cm).83

S1.5 Dust 84
At the LGGE), a set of 24 discontinuous 7-cm long samples was selected.Ice was 85 decontaminated through 3 successive baths of ultrapure water.Insoluble dust 86 concentration and size distribution measurements were performed by Coulter Counter 87 Multisizer IIe in a clean room setting.Each data represents the average of three 88 consecutive measurements performed following ultrasonic treatment.A density of 2,5 89 g/cm 3 was assumed for all particles in agreement with earlier studies.Analytical 90 procedure followed in this study is identical to that described in Lambert et al. (2008).91 D values.To achieve this, we have selected these successive "full glacial" episodes by 96 isolating in the D data set the periods with minimal values, using locations of 97 increasing -gradients as cutting points on both sides.Table 1 shows that the D ranges 98 obtained for the various selected periods are quite similar.However, these various 99 glacial episodes obviously cover different depth and time intervals.Before we discuss 100 and compare the mean values and the variability between those groups and the Deep 101 and Basal ice layers, it is essential to show that no artificial "time smoothing" is induced 102 by the methodology.For example, one might expect that, depending on the resolution of 103 the data set, increasing the time window at a given resolution would damp the 104 variability.In figure S1, we have plotted, for each chemical species considered and for 105 each group of samples, their mean (Fig. S1a) and 1 (Fig. S1b) value as a function of 106 the duration of the period, based on the EDC-3 time scale (Parrenin, 2007, extended for 107 the Deep and Basal ice layer).Although there is some variability, no clear trend of the 108 chemical signals is seen with duration of the episodes.Furthermore, if "time smoothing" 109 was to occur, it should result in a decrease of the variability, at a given resolution, as we 110 Can we retrieve a clear paleoclimatic signal from the deeper part of the EPICA Dome C ice core? 4 go down the EDC core, which is the reverse of what is observed comparing the Deep 111 and Basal ice groups to the "full glacial" episodes above.Finally, the contrast between 112 the signature of different elements within the Deep ice or within the Basal ice concerns 113 samples that are strictly from the same time window, at the same resolution, so that it is 114 sound to discuss it.115

S2.
Validity of the comparison between previous "full glacial" periods and the Deep and 92 Basal ice layers 93 As discussed in section 3, we have chosen to compare the chemistry of the Deep and 94 Basal ice layers to the one of the previous glacial climatic episodes, showing minimum 95 In this system the extracted air is transferred into a 52 GC in order to derive the CO 2 concentration.A more detailed description of the systems 53 used in Bern and at the LGGE is found in Siegenthaler et al. (2005) and Barnola et al.