Interactive comment on “ Radar stratigraphy connecting Lake Vostok and Dome C , East Antarctica , constrains the EPICA / DMC ice core time scale ”

Cavitte et al. tracked isochronous englacial radar reflectors between EPICA Dome C and Vostok ice cores, in order to better constrain the depth-age relationship of the Dome C ice core. In general, this manuscript is not well organized so hard to follow. The title says that radar constraints of ice-core time scales are reported, but the abstract says no specific results how this method improved the ice-core time scales. In section 2 (data and methods), it is said that the propagation speed is assumed to be 169 m/usec citing Carter et al. but without any further justifications of this assumption. Nothing is said that this propagation speed is further examined later. Suddenly, uncertainty of this assumption is discussed very end of Section 3.1.


Introduction
Ice cores retrieved from East Antarctica provide the longest record of direct greenhouse gas concentrations and are key to understand late Quaternary climate forcings.EPICA Dome C (EDC) (75 • 28 S 106 • 48 E, (Bender et al., 1994)) and Vostok (75 • 06 S 123 • 21 E, (Landais et al., 2006)) provide dated records down to 3300 m and 3189 m depths, corresponding to 411 ka and 801 ka, respectively (Parrenin et al., 2007;Bender and Suwa, 2008).Modelling uncertainties at such depths become significant: the EDC3 chronology has a 6 ka confidence interval for ice older than 100 ka.The O 2 /N 2 dating method applied at Vostok gives an improved accuracy of 2 ka throughout (Kawamura, 2009).The combined age uncertainties limit temporal and spatial resolution of climate change.We argue that modern radio-echo sounding (RES) surveys of the ice sheet provide an accurate way of validating ice cores age-depth relationship Introduction

Conclusions References
Tables Figures

Back Close
Full and extending these chronologies to spatially extensive areas of the ice sheet where no cores exist.Internal RES layering can be related to (a) density changes, (b) ice chemistry variation or (c) ice fabrics (Dowdeswell and Evans, 2004).For the depths considered, ice chemistry variation (b) is thought to be the dominant source of reflecting horizons.These horizons result from the deposition of discrete acidic aerosols as laterally extensive sheets on the ice surface and are preserved by later accumulation (Siegert et al., 1998a).The RES layering represents an independent method for ice core correlation as it is related to discrete volcanic events and not solely climatic events.Use of continuous RES interpretation as an alternative method for ice core correlation is advantageous over other techniques: (1) for contributing negligible errors to layer ages with respect to core dating uncertainties, (2) for providing a fast correlation method, (3) for providing an independent signal for palaeoclimate correlation and (4) for imaging spatially large areas to map englacial flow (Siegert, 1999).

Data and methods
We use RES lines acquired over several seasons by the University of Texas Institute of Geophysics (UTIG) aerogeophysical program (Fig. 1).The radar system operates with a centre frequency of 60 MHz, (Blankenship et al., 2001).Pre-2008 radar data (Vostok site coverage) were collected using a 250 ns pulse width; signals were digitised at 16 ns intervals and stacked to along track records every 10-12 m along track (Carter et al., 2009).Post-2008 data (Dome C site coverage) were acquired using a 1 µs chirp width and 6400 Hz pulse repetition frequency, with ≈ 100 ns pulse width after range compression (Peters et al., 2005); signals were digitised at 20 ns intervals and stacked yielding records every ≈ 22 m along-track (Young et al., 2011).Data interpretation was performed by tracking continuous horizons in ice, following peaks or troughs in processed amplitude, using an industry standard layer interpretation package (Schlumberger's GeoFrame).Introduction

Conclusions References
Tables Figures

Back Close
Full Horizon depths relative to the surface are computed assuming a constant electromagnetic velocity of 169 m µs −1 and 300 m µs −1 in ice and air, respectively (Carter et al., 2009).Complex internal layer geometry in the area of buried megadune structures requires manual interpretation to track layers.
3 Correlating Dome C and Vostok

Horizon correlation
Eleven horizons (see Fig. 2) spanning a 209 ka period, from the last glacial to the penultimate interglacial, were successfully picked between the Vostok and EPICA Dome C ice core sites, providing a direct stratigraphic correlation between the two sites.Ten to twelve radar lines were used for each horizon correlation; intersecting lines with crossover points ensure that the same horizon is being tracked throughout.Two alternative routes lead to Dome C ice core sites from Vostok providing a means of doublechecking the correlation.
An initial correlation was given by Siegert et al. (1998b), but involved a 150 km data gap close to Vostok, over which horizontal layer geometry was assumed.More recent data suggests that this approximation is not valid, due to the sloping bedrock and ice surface geometry of the area (Tabacco et al., 2006).Similarly, we find that layers are deeper at Vostok than in the Siegert et al. (1998b) study, for the same depth interval, which is more consistent with the local topography.
The eleven horizons were chosen on the basis of brightness and continuity, some dimmer horizons showing strong interference patterns were rejected.For all successful horizons, depths were measured at both ends of the correlation, where the radar lines passed closest to the ice core site location, using diffraction hyperbolae off station buildings in the radar data as a reference location (see Fig. 3 for radar lines).Horizontal continuity was assumed over the minor data gaps between the ice core sites and the radar lines of closest approach, corresponding to 1.2 km and 0.4 km at Introduction

Conclusions References
Tables Figures

Back Close
Full Vostok and Dome C, respectively.Depths are measured from the surface, to which we apply firn corrections (z f ) of 14.71 m and 9.40 ± 0.94 m at Vostok and Dome C, respectively; these corrections were computed using the Eq. ( 1), following (Dowdeswell and Evans, 2004) and published vertical density profiles for each site (Dowdeswell and Evans, 2004;Barnes et al., 2002, respectively) (firn corrections confirmed at Dome C by two independent studies: radar reflection study, 9.48 ± 2.7 m, E. Le Meur, personal communication, 2012; and seismic refraction study, 9.23 ± 2.7 m, R. Gasset, 1982).
n i refractive index of solid ice, 1.78.
Vertical resolution of RES layer interpretations is obtained from the measured radar pulse width (Millar, 1982); it represents 30 m at Vostok and 8 m at Dome C, respectively.
The picking accuracy is estimated to be ±1/10th of the pulse width, except for layers traversing megadune disturbances where it decreases to ±1 pulse width.This is the case of the five shallowest layers (excluding the top-most) that span the last glacial period from 113 ka to 41 ka (see Table 1).

Age-depth stratigraphy
Horizons are dated at Dome C and Vostok sites using published age-depth chronologies (Parrenin et al., 2007;Suwa and Bender, 2008).We linearly interpolate bagged ice core depths to fit our picked radar depths, and the same is done with the corresponding ice-sample age data to date the layers.Introduction

Conclusions References
Tables Figures

Back Close
Full The age uncertainty associated with dating the RES layers is computed independently at each horizon; radar depth uncertainties combine radar depth picking accuracy, core location uncertainty, and a firn correction uncertainty.RES depth uncertainty intervals are then projected to age uncertainty intervals, which vary with depth due to strain thinning and glacial/interglacial accumulation rate variations.All uncertainties can be found in Table 1.Published ice core dating errors are also reported at each horizon for comparison.

Discussion
A distinction must be made between layers belonging to the last glacial period and layers from the penultimate glacial period.The last glacial cycle is characterized by pervasive aeolian reworking of the ice sheet surface and translates to zones of buried megadunes through subsequent accumulation.These have been described as actively present at the surface of the East Antarctic Ice Sheet (Arcone et al., 2012a,b).Megadune facies are clearly visible in the post-2008 RES data sets collected with a modern "coherent" radar sounder (Peters et al., 2005(Peters et al., , 2007)).In these profiles, erosional surfaces are easily identifiable and some diagonal cross-bedding is visible in areas (see Fig. 4).Because of widespread aeolian reworking, radar-dating uncertainties (Table 1) in this region of the ice sheet are larger than traditional ice core dating uncertainties at shallow layer depths corresponding to the last glacial cycle.
Layers pertaining to the penultimate glacial cycle (from the MIS5e interglacial at Introduction

Conclusions References
Tables Figures

Back Close
Full the age error extrapolated from Vostok remains below 1 ka including for layer depths at Dome C in close proximity to the echo-free-zone.
Considering our extrapolated estimate of age-depth resolution at Dome C, we would expect a horizon to show a geochemical age difference between Dome C and Vostok no greater than the order of the radar resolution i.e. less than 700 yr.Differences obtained here are of the order of ka.Reasons for this discrepancy, if we do not consider layers from the last glacial cycle, are that (1) layers are not as laterally continuous as assumed and volcanic deposits from different layers merge, (2) interpretation errors are underestimated where radar vertical resolution is not sufficient to identify breaks in the stratigraphy or where layer roughness induces a jump to a contiguous shallower/deeper layer, giving an erroneous correlation, and (3) errors in ice core dating of ± ka are too large and the age differences observed between the two sites are a result of the lower dating resolution of the EDC3 timescale.The magnitude of our radar uncertainties lead us to believe that errors in ice core dating (3) is the most likely cause of the age differences.This implies that integration of an ice core site with relatively small age uncertainties in a network of radar surveys and ice core sites would allow refinement of age-depth estimates for other ice core stratigraphies.
For layers in the penultimate glacial cycle, we note that six out of the seven horizons considered are deeper at Vostok than at Dome C. We hypothesize that this can be explained by the presence of Lake Vostok (Kapitsa et al., 1996).Subglacial water eliminates strain thinning of the layers.In addition, melting of the bottom layers accommodates surface accumulation, thereby reducing compaction thinning in the Vostok area.(Petit et al., 1999) uses this argument for derivation of the GT4 Vostok timescale.The MIS6a glacial period is 324 m (≈ 50 %) thicker at Vostok than Dome C. (Siegert et al., 1998b) show the reverse of what we observe but we argue that the 150 km gap between the older RES data and Vostok, over which the authors assumed layers remained horizontal is the reason behind the discrepancy.
Our results show that radar layer tracing over larger areas of the central East Antarctic Ice Sheet would allow propagation of ice core stratigraphies to anywhere layers can Introduction

Conclusions References
Tables Figures

Back Close
Full be reliably traced.This tracing would strongly benefit inverse model studies of agedepth over any area.In addition, modern coherent radar measurements over cored sites would allow direct inter-core stratigraphic correlations and provide a verification of the quality of layer interpretations.Identification of future deep "old ice" coring sites is strongly reliant on such radar studies.

Recalibration
Using the (Suwa and Bender, 2008) Vostok timescale and our RES correlation, we are able to carry ages from Vostok Station to the Dome C ice core.We only do so for the seven layers of the penultimate glaciation that are unaffected by megadune reworking.Each layer at Dome C is in turn assigned the corresponding age obtained at Vostok.For depths between 1590 m and 2320 m at Dome C, radar uncertainties are negligible with respect to ice core dating errors in the computation of our associated age uncertainty and we give the combined rms error.Table 2 gives

Discussion
The EDC3-radar chronology is older than the Parrenin EDC3 chronology for the intervals 120-140 ka and post 220 ka, while it is younger than the ice core-based EDC3 between 140 ka and 220 ka.The 140-220 ka interval corresponds to the penultimate Introduction

Conclusions References
Tables Figures

Back Close
Full glacial period, while the other two intervals correspond, respectively to the last and penultimate interglacials.The disagreements between the two trends could reflect (1) incorrect assumptions used for the EDC3 ice core chronology reconstruction or (2) incorrect interpolation of the "EDC3-radar timescale".Various parameters used in the modeling of Parrenin et al. (i.e., initial accumulation rate temporal variation as well as the vertical thinning function) are poorly known and determined by inverse methods (Parrenin et al., 2007).They further indicate that the resulting ages do not match age markers perfectly.Accumulation rate reconstructions could easily be responsible for the change in sign of the discrepancies.We must also keep in mind that this study uses deep horizons; with the deepest layer (layer 13) reaching 7/10th of the total ice thickness, where high rates of shear thinning are experienced in the ice sheet.Interpolation of the new EDC3 timescale would benefit from the use of a thinning model.We believe the disagreement between the two timescales reflects a combination of errors ( 1) and (2).
Our results do agree with comparative dust flux variations in the two ice cores.Delmonte et al. ( 2004a) detailed dust fluxes in the Vostok and Dome C cores and found a strong coherence between the respective dust profiles; both show highest dust concentrations at glacials while low concentrations reflect interstadial or interglacial conditions.They were able to correlate the two cores through the matching of 10 outstanding dust events, which show the same stratigraphic relationship as this radar study.Their correlation shows a common dust marker to be shallower at Vostok between the surface and a depth of 1120 m, and an inverse relationship between 1120 m and 2200 m.
Our eleven layers match this stratigraphic relationship extraordinarily well.The fact that dust markers in the ice would show the same stratigraphic relationship confirms the strong link between our radar reflectors and discrete dust/acidic depositional events, and supports the stratigraphic correlation obtained in this study (see Fig. 2).Introduction

Conclusions References
Tables Figures

Back Close
Full

Conclusions
Thirteen strong and continuous radio-echo layers were identified in UTIG airborne surveys from 2000 to 2011 and eleven of these layers were tracked between the EPICA Dome C and Vostok ice core sites.The correlation provides the first continuous direct chrono-stratigraphic link between the two ice core timescales.The location of this study along the Byrd-Totten ice divide over deep bedrock basins makes this area a very important one in our search for old ice.We map layer geometries over the last two glacial cycles, which can strongly benefit ice core modeling as well as studies of spatial patterns of accumulation over time using inverse layer modeling (Leysinger Vieli et al., 2011).The strong advantage of the technique used is that it does not require traditional modeling and inter-core marker comparisons with the uncertainties involved.
Through the last glacial cycle, where ice core chemistry is very reliable, radar dating techniques are compromised by buried megadune fields, whilst the penultimate glacial cycle, where larger uncertainties prevail in the ice core chemistry and timescales, RES dating uncertainties are of the order of several hundred years, quite small in comparison to current ice core dating uncertainties.By linking Vostok and Dome C ice cores, age uncertainties in the EDC3 timescale are reduced from ±3-6 ka to ±2 ka, a major chronology improvement for precise climate change and forcing studies (Kawamura, 2009).Our results indicate that an improved EDC3 chronology will be required to understand spatial and temporal climate variation at Dome C and, in turn, more accurate age-depth model there will be useful as a final validation of our techniques.We emphasize the importance of extending these techniques over wider areas of Antarctica as well as carefully connecting interpreted radar surveys (and new surveys) to existing ice cores; this will be especially important where a known age depth stratigraphy will be needed to constrain modeling in search of deeper "old ice".

TCD Figures Back Close
Full Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Screen / Esc Printer-friendly Version Interactive Discussion Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | our recalibrated EDC3 timescale (termed hereafter EDC3-radar).A simple linear interpolation is used to reconstruct EDC3 between horizon pairs and should be taken as an initial result as no thinning function or accumulation model has been used in the reconstruction.The EDC3-radar timescale is plotted with the EDC3 Parrenin et al. (2007) δO 18 timescale for comparison (Fig. 5): differences are within the Parrenin et al. (2007) error range but significantly different if the EDC3-radar timescale uncertainty bounds are considered.
Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper |

Fig. 1 .Fig. 2 .
Fig. 1.Map of the East Antarctic Plateau focused on the study area.The inset locates the ice cores in a map of Antarctica; a white rectangle locates the blown-up study area.Radar transects used are shown, overlaid on MODIS ice surface velocity mosaic (Rignot et al., 2011); black contours are ice surface elevation in meters(Bamber et al., 2009).

Fig. 3 .
Fig. 3. Internal layering along UTIG RES transects, over Vostok (A-A ) and EPICA Dome C (B-B ) ice core sites.Ice core locations are indicated.Note the bright bedrock and Lake Vostok reflections, as well as the thick accreted ice layer, well resolved in the Vostok section.Internal layering is clearly visible and continuous throughout.