PROJECT DESCRIPTION: COLLABORATIVE RESEARCH: RESOLVING AMBIGUOUS EXPOSUREAGE DEGLACIATION CHRONOLOGIES WITH MEASUREMENTS OF IN-SITU-PRODUCED COSMOGENIC CARBON-14 This proposal aims to apply measurements of in-situ-produced cosmogenic 14C in Antarctic glacial deposits to address important ambiguities in reconstructing past Antarctic ice sheet change that have not been resolved by existing exposure-age data sets. This is important because much of our understanding of Antarctic ice sheet change between the Last Glacial Maximum (LGM; ca. 15,000-25,000 years ago) and the present stems from cosmogenic-nuclide exposure-dating of glacial deposits in currently-ice-free areas. Ice sheets transport rock and sediment eroded from their bed, where it has not been exposed to the surface cosmic-ray flux. They deposit it at ice margins, at which point exposure to the cosmic-ray flux begins. Thus, the concentration of trace cosmicray-produced radionuclides in glacially transported clasts in ice-free areas is directly proportional to the deglaciation age of a site. Exposure-age deglaciation chronologies are simple to interpret, powerful, and effective in simple situations where all the glacial sediment present on a nunatak was deposited during the most recent deglaciation (e.g., Stone et al., 2003). However, many ice-free areas of Antarctica are covered by glacial deposits that were emplaced during many previous glacial-interglacial cycles, and preserved during periods of ice cover by frozen-based ice that did not transport or erode existing deposits (Sugden et al., 2005; Balco, 2011). These areas display surficial glacial deposits with a wide range of exposure ages that often do not directly represent the emplacement age of the deposits. Exposure-age data are extremely complex to interpret in this situation, in particular when no samples with exposure ages postdating the LGM are found. In this case, there is no way to distinguish between two incompatible possibilities: First, the site was never covered by ice during the LGM; second, the site was ice-covered during the LGM, but researchers failed to find a few glacially transported clasts deposited at or after the LGM among a large abundance of similar-appearing but older clasts. In other words, absence of evidence is not evidence of absence. A number of past exposure-dating studies, many in important locations from the perspective of ice sheet reconstruction, are subject to this ambiguity. It presents a significant obstacle to accurate reconstruction of LGM-to-present ice sheet change in Antarctica. Here we describe a means of resolving this ambiguity by measuring the concentration of in-situ-produced cosmogenic carbon-14 in quartz in samples collected during these previous studies. This is possible because 14C has a two-order-of-magnitude shorter half-life (5730 yr) than other cosmic-ray-produced radionuclides that were used in these past studies, specifically 10Be (1.5 Ma), 26Al (0.7 Ma), and 3He (stable). A short half-life acts to i) limit or negate the importance of inherited nuclide concentrations produced during previous glacial-interglacial cycles, and ii) maximize the effect of short periods of ice cover on nuclide concentrations. Because of these features, the two scenarios described above -- no ice cover during the LGM, or ice cover but no emplacement of glacial deposits -- can be resolved by 14C measurements on either bedrock surfaces or erratic clasts that predate the LGM. Furthermore, in cases where the LGM-to-present deglaciation history is already known from an unambiguous exposure-age chronology, additional 14C measurements can also potentially provide a chronology for ice sheet thickening prior to the LGM, information which cannot be obtained at all from existing data sets and would be extremely valuable in attributing sea level change prior to the LGM. Very few previous Antarctic exposure-dating studies (three that we are aware of) have taken advantage of this possibility. The reason for this is that so far, in-situ-produced 14C measurements have been 1 difficult, time-consuming, and only possible at a very small number of laboratories worldwide. Thus, these studies have only included a handful of analyses. In this proposal we aim to take advantage of a new, automated system for 14C extraction from quartz, currently under construction by PI Brent Goehring at Tulane University, to apply this technique at the scale needed to make significant progress on constraining LGM-to-present, and also potentially pre-LGM, ice thickness changes in important regions of Antarctica where existing exposure-age data cannot provide these constraints. In the rest of this proposal we will: i) outline the importance of accurate reconstructions of LGM-topresent Antarctic ice sheet change and highlight areas where past exposure-dating studies lead to ambiguous reconstructions; ii) describe how these ambiguities could be resolved by targeted 14C measurements; and iii) detail a number of sites where such measurements, carried out on existing samples collected in previous field projects, would significantly improve reconstructions of past ice sheet change. This proposal does not involve new fieldwork in Antarctica. Figure 1. Map of Antarctica showing all sites mentioned in text. Importance of reconstructing LGM-to-present ice sheet change. Reconstructing past Antarctic ice sheet change is important because an accurate understanding of past ice sheet changes is necessary for determining the causes of past and future sea level changes. Specifically, a significant body of research over the past several decades has sought to reconcile i) estimates of changes in global ice volume over time inferred from inversion of a global data set of paleosea-level proxy data, with ii) direct reconstructions of changes in the size of glaciers and ice sheets derived from geological evidence. To date, these have not yet been satisfactorily reconciled (e.g., 2 Lambeck et al., 2014; Clark and Tarasov, 2014, and references therein). In Antarctica, estimates of the total excess LGM ice volume relative to present range from ca. 30 m (from relative sea level inversions, see discussion in Lambeck et al., 2014) to ca. 10 m (from model interpolation of geological data; e.g., see Whitehouse et al., 2012 and Briggs et al., 2014). A possible reason for this discrepancy, proposed by several authors (e.g., see discussion in Carlson and Clark, 2012), is that the most significant changes in ice volume in Antarctica between the LGM and present took place in the large Ross and Weddell marine embayments, where a large expanse of continental shelf permits large changes in grounding line position and, in addition, few nearby geological constraints exist. Although researchers involved in collecting geological and geochronological constraints on past ice sheet change in Antarctica have for the most part argued that the geological evidence is not consistent with large changes in ice volume in the Ross and Weddell embayments (e.g., Bentley et al., 2011), this evidence, as we discuss below, is sparse and ambiguous in several important locations. Thus, other authors have argued that geological and geochronological data in Antarctica are at present not adequate to determine with high confidence whether geologically-based LGM ice sheet reconstructions are or are not consistent with sea-level inversion estimates of ice volume (e.g., Clark, 2011; Clark and Tarasov, 2014 and references therein). To summarize, the primary reason the work we propose here is important is that it will contribute to resolving ambiguities regarding the LGM configuration of the Antarctic ice sheets. In addition, although we will focus primarily on LGM ice extent in this proposal, the results of this project will also contribute to creating accurate reconstructions of transient LGM-to-present ice sheet change from geological data. Such reconstructions, in turn, are critically important to interpreting satellite gravity data used to estimate present annual- and decadal scale rates of Antarctic ice volume change (e.g., Ivins et al., 2013; Shepherd et al., 2012). Why frozen-based ice cover and cosmogenic-nuclide inheritance lead to ambiguous ice sheet reconstructions from exposure-age data. Elevation (m) 1400 1300 1200 Present ice surface 0 5 10 15 Exposure age (ka) 20 25 1k 10k 100k 1M Exposure age (yr) 10M Figure 2. 10Be exposure ages from the Quartz Hills (Todd et al., 2010, plotted on linear (left) and log (right) axes. The upper limit of the stratigraphically youngest glacial drift mapped at this site is 1400 m. Although this drift displays a large range of exposure ages, the majority are 7-20 ka. All exposure ages from stratigraphically older drifts, exposed above 1400 m elevation, exceed 100 ka. Thus, geologic mapping of the youngest drift combined with the exposure-age data allows unambiguous reconstruction of LGM ice thickness at this site. Most age-elevation plots in this proposal were generated using the ICED database of cosmogenic-nuclide exposure ages for Antarctica (http://hess.ess.washington.edu/iced/). 3 Figure 2 shows cosmogenic-nuclide exposure-age data (in this case, measured using 10Be) from the youngest glacial deposit mapped at a site adjacent to the Reedy Glacier in the southernmost Transantarctic Mountains (Todd et al., 2010; see Figure 1 for location). The mapped extent of the drift (below 1400 m elevation) and a large number of exposure ages between 7-20 ka indicate that i) this drift is LGM in age; ii) the elevation of the LGM ice surface was 1400 m at 16-19 ka; and iii) glacier thinning took place between the LGM and ca. 7 ka. Thus, this is an example of a glacial-geological and exposuredating study where, although a particular drift unit contains a few pre-exposed or recycled clasts with preLGM apparent exposure ages, the preponderance of LGM and Holocene ages allows unambigous reconstruction of the LGM ice surface elevation. In contrast, Figure 3 shows exposure-age data for erratics from glacial drift on Mt. Skidmore in the Shackleton Range, adjacent to the eastern Weddell Sea (Hein et al. 2010; see Figure 1 for location). At this site, the youngest erratics, lying on the present ice surface, are 40 ka. All erratics on the nunatak itself have apparent exposure ages > 100 ka. Hein et al. proposed that the best explanation for these data was that the LGM ice surface elevation was at or below the present ice surface, that is, this portion of the ice sheet did not thicken during the LGM. However, as they noted, it is also possible that the LGM ice surface was higher than present, but either no glacial drift was deposited by LGM ice, or that they did not identify any clasts emplaced at the LGM among a much larger number of older clasts emplaced during previous glacial-interglacial cycles and covered by frozen-based ice at the LGM. 900 Figure 3. Apparent 10Be (red) and 26Al (green) exposure ages from Mt. Skidmore, Shackleton Range (Hein et al., 2010). Ages < 100 ka are only observed in erratics on the present ice surface. Thus, it is unclear whether i) the site was not covered by ice during the LGM or ii) the site was covered, but no erratics were deposited and/or found. Elevation (m) 800 700 600 500 400 300 200 10k Present ice surface 100k 1M Exposure age (yr) 10M Sites such as this one where a large number of erratics with pre-LGM ages were found, but erratics with LGM to Holocene ages were either completely absent or only found very close to the present ice surface, are common at ice-free areas surrounding the Weddell Sea embayment (Figure 1). Besides the Shackleton Range, these include several sites in the Pensacola Mountains (specifically, the Schmidt Hills, discussed below), nunataks in the Lassiter Coast region (also discussed below), and some sites in the Ellsworth Mountains (Fogwill et al., 2014 and references therein). The inherent ambiguity of these exposure-age data sets leads to significant uncertainty in reconstructing LGM-to-present ice sheet change in the Weddell embayment, as detailed by Hillenbrand and others (2014). They showed that it was not possible on the basis of terrestrial or marine geological data to distinguish between a scenario in which i) the Weddell Sea grounding line uniformly advanced to the outer continental shelf (in which case the sites in the Shackleton Range would have been covered by thicker LGM ice), or ii) significant grounding line advance only occurred in the central Weddell Sea, and only minimal grounding line 4 advance took place in the eastern and western sides of the embayment (in which case they would not have been covered by ice). Although it is important to note that the difference in these reconstructions in terms of total ice volume could not by itself account for the difference between geological and sea-levelinversion estimates of Antarctic ice volume discussed above, the inability to distinguish between these two possible scenarios for the Weddell embayment highlights the fact that the challenge in unambiguously determining LGM ice thickness from existing exposure-age data presents a fundamental obstacle to accurately reconstructing LGM Antarctic ice volume. To summarize, the ambiguous interpretation of exposure-age data created by the fact that erratics with apparent exposure ages older than the LGM do not exclude cover by frozen-based ice at the LGM, both in the specific example of the Weddell Sea and in some other locations, leads to a large uncertainty in the LGM ice sheet configuration over a significant fraction of the continent. In the next section we describe how in-situ 14C measurements at specific sites can resolve the ambiguity presented by existing exposure-age data sets as to whether the sites were covered by ice at the LGM or not. We then propose to apply this technique synoptically at a number of sites in Antarctica, mainly focused on the Weddell embayment. In-situ-produced carbon-14 in quartz. Like the commonly used cosmic-ray-produced radionuclides 10Be and 26Al, 14C is produced in-situ in the quartz mineral lattice by high-energy neutron spallation (Lifton et al., 2001 and references therein). However, its half-life is two orders of magnitude shorter (5730 years compared to 0.7 and 1.4 Ma for 26Al and 10Be respectively). This is important because cosmogenic 14C concentrations in bedrock surfaces retain less “memory” of past periods of exposure and burial than do longer-lived nuclides. In many cases, the majority of the 10Be inventory in bedrock surfaces or erratics that have been repeatedly exposed and covered by frozen-based ice reflects periods of exposure prior to the current one. Because of its short half-life, this is not true for the 14C inventory in nearly all cases (e.g., Miller et al., 2006; Briner et al., 2014). Why LGM ice cover can be clearly detected or excluded by 14C measurements where it cannot using existing data. Figure 4 shows how measurements of cosmogenic 14C in quartz in samples, from either bedrock or erratic clasts that were emplaced prior to the LGM and were not disturbed by LGM ice cover, spanning a range of elevations, can be used to determine to what extent a site was covered by ice at the LGM. Given continuous exposure of a rock surface for several times the 14C half-life, cosmogenic 14C concentrations in quartz in this surface will reach an equilibrium concentration where production is balanced by decay and loss to surface erosion (specifically, after 30,000 years the 14C concentration reaches 98% of saturation). At the low rock surface erosion rates observed in many inland regions of Antarctica (ca. 0.2-2 m/Myr; see Balco and Shuster, 2009), this equilibrium concentration is only weakly dependent on the erosion rate (varying the erosion rate within these bounds has only a 3% effect on the concentration). In this project, we are interested in determining whether ice-free sites were ice-covered at the LGM. If they were not, they would have experienced continuous exposure for at least the last ~30,000 years, so quartz in surfaces that have been stable for at least this amount of time -- bedrock or pre-LGM erratics that were not disturbed by LGM ice cover -- would display cosmogenic 14C concentrations at steady state with respect to the production rate at the sample elevation (Figure 4). Note that the elevation dependence of the steady state 14C concentration in quartz in Antarctica has been directly measured in the Antarctic Dry Valleys (N. Lifton and CRONUS-Earth project, personal communication; see data on Figure 4), so 5 potentially uncertain scaling extrapolations to estimate this value are minimized. If the samples did experience ice cover for a time at the LGM, even if no subglacial erosion took place, ice cover exceeding several meters thickness would halt nuclide production, and the 14C concentration would be reduced by radioactive decay. If deglaciation occurred at ca. 15 ka, the subsequent time would not be sufficient to reestablish steady state, so an elevation transect of 14C concentrations in quartz would show a discontinuity at the maximum elevation reached by the LGM ice surface (Figure 5). White et al. (2011) described a similar approach, although it is important to note that our approach involves identifying a discontinuity in a 14C elevation transect rather than comparing a measured 14C / 10Be ratio to the (relatively uncertain) production ratio. Thus, their statement that one cannot detect periods of ice cover prior to ~15 ka does not apply to our proposed approach. 800 Hypothetical ice surface elevation change Resulting predicted C−14 concentration 700 Elevation (m) 600 Discontinuity at LGM ice surface elevation 500 400 300 200 Steady-state concentration in absence of LGM ice cover Data from Antarctic rock surfaces that were not ice-covered at the LGM 100 0 35 30 25 20 15 10 Time (thousands of years BP) 5 0 0 0.5 1 1.5 2 2.5 3 3.5 4 Cosmogenic 14C concentration in quartz (10 5 atoms/g) Figure 4. Predicted elevation dependence of the cosmogenic 14C-in-quartz concentration in surface samples from a hypothetical nunatak that was partially ice covered at the LGM. The ice thickness change history at left implies the 14C concentrations at the present time given by the solid line in right panel. A discontinuity between saturated and sub-saturated concentrations marks the upper limit of LGM ice cover. Data plotted in right-hand figure are 14C concentrations measured at surfaces in Antarctica that are independently known to have not been covered by ice at the LGM (unpublished CRONUS-Earth calibration data analyzed by N. Lifton); these show typical measurement uncertainties for low-elevation samples. Uncertainties would be lower at higher elevations due to the higher production rate. In summary, 14C measurements on either bedrock surfaces or surface clasts that were emplaced prior to the LGM allow one to unambiguously determine whether or not a site was covered by ice during the LGM. While there is a lower limit on the duration of ice cover that would be detectable using this approach (approximately 1-3 ka, depending on site elevation), existing data from Antarctica that pertain to the duration of LGM ice cover show that the Antarctic ice sheets were most likely near their maximum extent for more than 10 ka (summarized in Carlson and Clark, 2012, p. 25-27). Theoretically, this approach to detecting periods of ice sheet cover could also be addressed with 10Be and 26Al concentrations (Lal, 1991). The much longer half-lives of 10Be and 26Al, however, require orders-ofmagnitude longer periods of ice cover to detectably affect concentrations of these nuclides. For example, 173,000 years is required for the 26Al/10Be ratio to decrease to 92% of its equilibrium value (that is, to be detectably different from equilibrium given 8% measurement uncertainty on the 26Al/10Be ratio), whereas 6 only 700 years is required for the 14C concentration to decrease to 8% below its saturation value. Measurement of 14C, therefore, has enormously greater sensitivity than any other cosmogenic-nuclide system for detecting whether ice-free sites were covered by ice during the LGM, even if only for a short period of time (< 1 kyr). New information obtainable from 14C measurements: an ice advance and retreat chronology. 14C Figure 5. Predicted concentrations at the present time given a surface that begins with a saturated 14C concentration and then experiences ice cover at the LGM and subsequent deglaciation. If the deglaciation age is known, the 14C concentration is uniquely related to the duration of ice cover. Note that saturation with respect to cosmogenic 14C production is ~98% reached after only 30 kyr of exposure. Similarly, 14C concentrations after 30 kyr of ice cover are nearly zero due to radioactive decay. [C−14] at present time (105 atoms/g) An additional capability of 14C measurements is that they permit one to determine the duration of ice cover at a site where the deglaciation age is already unambiguously known from 10Be exposure ages of glacial erratics (or from other independent chronological evidence). This approach is essentially a form of cosmogenic-nuclide burial dating (e.g., Lal, 1991; Granger, 2006), first applied using the 14C/10Be pair on proglacial bedrock at the Rhone Glacier, Switzerland (Goehring et al., 2011). In the Rhone Glacier example, deglaciation following the LGM is known from nearby erratics with 10Be exposure ages, and the present 14C concentration in bedrock reflects burial and erosion of the bedrock surfaces during a late Holocene glacier advance-retreat cycle. We propose that this approach can be applied to Antarctic sites because i) it is likely that a bedrock or pre-LGM erratic surface is at or near its saturation concentration with respect to 14C at initiation of LGM ice cover; and ii) the potential complicating factor of subglacial erosion of the bedrock surfaces is not relevant at Antarctic sites that were covered by cold-based, nonerosive ice at the LGM. These assumptions leave the 14C concentration measured at the present time a function of two parameters of interest, the age of first ice cover at the LGM and the deglaciation age. At many sites in Antarctica (such as the Ford Ranges discussed below), the latter is known from 10Be exposure ages of erratics (see Figure 8), so the former can be inferred from the present 14C concentration. Figure 5 shows this relationship: given a known deglaciation age, there is a unique relationship between the measured 14C concentration and the onset (or duration) of LGM ice cover. 1.6 Individual lines correspond to deglaciation age of site: 1.4 12 ka 11 ka 10 ka 9 ka 1.2 8 ka 1 7 ka 6 ka 0.8 15 ka 5 ka 20 ka 25 ka 30 ka Time LGM ice cover began (yr BP) 35 ka There are two potential difficulties with this approach. First, it is difficult to verify the assumption that bedrock surfaces displayed 14C saturation at the time LGM ice cover began. If this assumption is relaxed, however, even though the measured 14C concentration no longer implies a unique age for ice advance, it continues to provide a limiting age, which would still be more information than we have at present about the duration of LGM ice cover in much of Antarctica. Second, as shown in Figure 5, because of the short half-life of 14C, the longer the duration of LGM ice cover, the less sensitive the relationship between ice 7 cover duration and 14C concentration. This is evident in the decreasing slope with LGM onset age in Figure 5; in this example, LGM onset ages older than ca. 30-35 ka would not be uniquely resolvable. However, we emphasize that even establishing limits on the duration of LGM ice cover would, at nearly all locations in Antarctica, provide new information that we do not now have. Thus, we argue that exploring this potential use of cosmogenic 14C measurements, in addition to our primary goal, described above, of better establishing the LGM configuration of the Antarctic ice sheets, is likely to yield potentially valuable information. Research plan and specific goals. The technical aspects of extracting and measuring in-situ 14C have developed relatively recently when compared to other cosmogenic nuclides such as 10Be, 26Al, or 3He. Recent progress on technical issues has overcome many of the early extraction challenges (Lifton et al., 2001; Pigati et al., 2010; Hippe et al., 2013; Goehring et al., 2014; Lifton et al., in press); the main limitation on its use is now the timeconsuming nature of the analysis. Cosmogenic 14C uses a pure quartz separate prepared in the same way as for 10Be or 26Al measurements; the sample is heated at high temperature in-vacuo, evolved CO2 purified, and its 14C/12C ratio determined by accelerator mass spectrometry (AMS; Lifton et al., 2001). Sample extraction in this project will take place in PI Goehring’s newly constructed laboratory at Tulane University. All aspects of 14C extraction, purification, and graphitization will be automated, thereby increasing both sample throughput (~1 per day) and repeatability, as evidenced by results from a similar laboratory at Purdue University’s PRIME Laboratory (Lifton et al., in press). AMS analysis for 14C will be carried out at the Purdue Rare Isotope Measurement Laboratory. Predicted 14C concentrations in the proposed work are similar to concentrations routinely measured (e.g., Lifton et al., 2001; Miller et al., 2006; Goehring et al., 2011; White et al., 2011), and therefore should yield typical measurement precision of 3-5%. The research plan is directed at two main goals. The first is is to better constrain the LGM ice sheet configuration at sites where existing 10Be data are ambiguous. As discussed above, many such sites occur at the margins of the Weddell embayment, so we are focusing on these sites with the goal of improving the overall understanding of LGM ice sheet configuration in this region. This will involve analysing samples from the Schmidt Hills and Thomas Hills (collected and already analyzed for 10Be by co-PI Balco and his collaborator Claire Todd), the Shackleton Range (collected and previously analyzed for 10Be by Andy Hein, University of Edinburgh; see attached letter of collaboration), and the Lassiter Coast (collected in recent fieldwork and already analysed for 10Be by Joanne Johnson, British Antarctic Survey; see attached letter of collaboration). The second goal is to attempt to apply 14C measurements at sites where the Holocene deglaciation history is already independently well constrained by a LGM-toHolocene 10Be exposure age chronology; the aim here is to complement these deglaciation chronologies with new information about i) advance of the Antarctic ice sheets to their LGM configuration, and ii) a potential middle Holocene ice thickness minimum and late Holocene thickening in the Antarctic Peninsula region. Research goal 1: ambiguous 10Be data sets around the Weddell embayment. Shackleton Range. Figure 3 and associated discussion above, as well as a review by Hillenbrand et al. (2014), highlight the importance of determining unambiguously whether or not nunataks in the Shackleton Range sampled by Hein et al. (2010) were or were not covered by ice at the LGM. To determine this, we will measure 14C concentrations in samples, including pre-LGM erratics and bedrock, from the full range of elevations available at several sites in the western Shackleton Range. At present, the samples are held 8 by Andy Hein (University of Edinburgh), who has agreed to make them available for this project (see attached letter of collaboration). Lassiter Coast. 10Be ages from erratics and bedrock at sites on the Bowman Peninsula of the Lassiter Coast are all in the range 100-800 ka, much older than the LGM, even when just 20 m above the modern ice surface (Figure 7; unpublished data supplied by Joanne Johnson, British Antarctic Survey). These sites are located at the present western edge of the Ronne-Filchner ice shelf edge. At present, the only constraint on timing of LGM-to-present grounding line retreat along the Lassiter Coast comes from a radiocarbon age (5398 yr BP) from till collected from a marine core on the inner shelf of the Ronne Trough near the sample sites (Hillebrand et al., 2014). Thus, the 10Be dataset as it stands at present does not allow us to either confirm or rule out that the LGM ice surface was any higher during the LGM than at the present day. Thus, we propose to measure 14C in bedrock and/or erratic samples with pre-LGM apparent 10Be exposure ages from the full elevation range available at this site. These samples are held by Joanne Johnson (British Antarctic Survey), who has agreed to make them available for this project (see attached letter of collaboration). Figure 7. Apparent Be-10 exposure ages on erratics from nunataks at the Lassiter Coast, adjacent to the western edge of the Ronne-Filchner ice shelf. Unpublished data collected by Joanne Johnson (BAS). Elevation (m) 1000 800 600 Present ice surface 400 10k 100k 1M 10M Exposure age (yr) Schmidt and Thomas Hills, Pensacola Mountains. These sites are located in the southernmost Weddell embayment (see Figure 1), adjacent to the Foundation Ice Stream. co-PI of this proposal Greg Balco and collaborator Claire Todd collected exposure-age data from these sites in 2011 and 2012. The Schmidt Hills are near the grounding line of the Foundation Ice Stream; the Thomas Hills are 100 km upstream. Samples from the Schmidt and Thomas Hills are currently held by Balco at BGC. At the Schmidt Hills, a large number of 10Be exposure ages are uniformly greater than 100 ka, providing no evidence for ice thickening, or by implication grounding line advance, at the LGM (Figure 6, left panel). However, data from the nearby Williams Hills, adjacent to the Academy Glacier, a tributary of the Foundation Ice Stream, show that the LGM ice surface was several hundred meters higher than present in the early Holocene (Balco and Todd, unpublished data; Tremblay et al., 2014). Significant thickening at this site is difficult to reconcile glaciologically with the hypothesis that no ice thickening took place at the Schmidt Hills and thus that the grounding line of the Foundation Ice Stream did not advance. 14C measurements from the Schmidt Hills would resolve this conflict and, we propose, provide unambiguous evidence for whether the grounding line in the inner Weddell Sea significantly advanced at the LGM. 9 In preparation for this proposal, we (Goehring) measured the 14C concentration in quartz in three erratics with apparent 10Be ages > 100 ka from elevations above 500 m at the Schmidt Hills. Measured 14C concentrations in all three were equal to or greater than predicted saturation concentrations at their respective elevations, which would tend to indicate that the LGM ice surface at the Schmidt Hills was not above 500 m. However, measured concentrations in two samples were significantly greater than expected saturation concentrations. Although there exist complicated geomorphic scenarios, involving transport of pre-exposed erratics from much higher elevations by LGM ice, that could in part account for this observation, these are not consistent with other geomorphic evidence, which suggests the possibility of unrecognized measurement errors. Thus, these measurements require independent replication, in this case by Goehring in his Tulane laboratory. In addition, they are from relatively high elevations at the site and provide no information about LGM ice cover at lower elevations. Thus, at the Schmidt Hills, we propose to make 14C measurements on samples from the full elevation range available at this site (220950 m). In addition, because bedrock at this site (a friable phyllite-grade metapelite with metabasalt intrusions) is not suitable for cosmogenic-nuclide measurements, we will make replicate measurements on multiple erratics at several elevations to ensure that LGM-to-present disturbance of erratics is not a possible explanation for anomalous results. Finally, 10Be results from the Schmidt Hills are sparse at middle elevations (Fig. 6). Samples from these elevations were collected, so we will make several additional 10Be measurements from poorly represented elevations to identify additional samples with preLGM 10Be exposure ages for 14C analysis and archive similar data density across the elevation range available at this site. 1000 Schmidt Hills Thomas Hills Elevation (m) 800 600 Present ice surface 400 200 10k Present ice surface 100k 1M Exposure age (yr) 10M 1k 10k 100k 1M Exposure age (yr) Figure 6. Apparent 10Be exposure ages at the Schmidt (left) and Thomas Hills (right), Pensacola Mountains, Antarctica. Unpublished data collected by Greg Balco and Claire Todd; also available via the ICE-D database (http://hess.ess.washington.edu/iced/). Apparent 10Be exposure ages from the Thomas Hills are, for the most part, well in excess of LGM age in the range 50-200 ka. However, six samples between 680-820 m elevation display early Holocene apparent exposure ages. This would appear to suggest LGM ice cover of the site followed by Holocene deglaciation. However, because these samples occur in a limited elevation range and do not form a clear age-elevation relationship, these data could potentially be explained by other geomorphic processes and the conclusion that they represent early Holocene deglaciation is weak. In addition, even if these data do record early Holocene deglaciation, they do not by themselves constrain the maximum elevation reached by LGM ice. Thus, at this site, as at the Schmidt Hills, we propose to measure 14C concentrations in erratics with pre-LGM exposure ages (again, metapelite and metabasalt bedrock is unsuitable for analysis) across the full range of elevations available at this site. We have samples at elevations higher 10 than represented by the data in Figure 6, so we will make additional 10Be as well as 14C measurements as needed to extend the elevation range available. Research goal 2: a potential chronology of ice sheet advances as well as retreats. A secondary goal of this project is to attempt to apply the concept of “burial dating” with in-situ-produced 14C in quartz, as described above, to quantify the duration of past periods of ice cover in addition to when they ended. We propose two trial applications of this approach: one to determine the duration of LGM ice cover at sites in the Ford Ranges of West Antarctica where a detailed and unambiguous exposure-age chronology of LGM-to-present deglaciation already exists (Stone et al., 2003); another to potentially investigate the timing of repeated Holocene ice shelf formation and breakup in the Larsen embayment of the Antarctic Peninsula. Ford Ranges of West Antarctica. Stone et al. (2003) collected 10Be exposure ages from several nunataks in this region (see Figure 1 for location; Figure 8 shows an example from one of these nunataks, Mt. Darling). All these data showed clear and consistent Holocene age-elevation relationships indicating ice sheet thinning between 10 ka and the present. At most elevations, erratic clasts with Holocene exposure ages lie on weathered bedrock with apparent exposure ages well in excess of the LGM and typically > 100 ka (Sugden et al., 2005). This juxtaposition indicates that we can use the relationship shown in Figure 5 above to relate 14C concentrations in bedrock samples to the duration of LGM ice cover. This, in turn, would potentially provide information about the growth of the West Antarctic Ice Sheet prior to the LGM, information that we do not now have. We note that our approach is not applicable to some bedrock samples at low elevations that have Holocene apparent 10Be exposure ages as well as geomorphic evidence that significant subglacial erosion took place during LGM ice cover. Thus, in this part of the project we propose to measure 14C in bedrock samples from relatively high elevations, with the goal of determining whether it is possible to reconstruct ice advance to its LGM position. These samples are currently held by John Stone, who has agreed to provide them to this project (see attached letter of collaboration). Figure 8. 10Be data from Mt. Darling in the Ford Ranges (Stone et al., 2003). Adjacent nunataks show similar ageelevation relationships, clearly defining early and middle Holocene thinning of this part of the West Antarctic Ice Sheet. Elevation (m) 1100 1000 900 Present ice surface 800 0 2000 4000 6000 8000 10000 12000 Exposure age (yr) Larsen embayment, northern Antarctic Peninsula. Here we propose to measure in-situ 14C in a set of samples collected from nunataks adjacent to major glaciers draining the east side of the northern 11 Antarctic Peninsula into the Larsen embayment. The 10Be exposure-age data set from this site (Balco et al., 2013; Figure 9) is unusual in Antarctica because at elevations below 150 m above sea level, erratics with middle Holocene (4-7 ka) apparent exposure ages occur adjacent to erratics of similar appearance but with much younger apparent exposure ages of 100-500 years. Balco et al. (2013), based on evidence from nearby marine sediment cores that the Larsen Ice Shelf collapsed in the early Holocene and reformed in the late Holocene prior to its final collapse in recent decades, hypothesized that this juxtaposition reflects early Holocene ice surface lowering due to overall regional deglaciation, followed by glacier thickening caused by ice shelf formation and growth in the late Holocene and then complete deglaciation of the sites in recent decades. This hypothesis would imply that i) the low-elevation erratics with mid-Holocene exposure ages were emplaced in the early Holocene and then covered by thin, frozenbased ice during late Holocene glacier thickening; and ii) the erratics with very young exposure ages were emplaced during final deglaciation of the sites in recent decades. This, in turn, is relevant to the present proposal because it would imply that erratics with early Holocene ages above 150 m elevation will display concordant 14C and 10Be ages, but erratics with early Holocene ages below 150 m elevation will display significantly lower 14C concentrations due to a period of burial during the middle to late Holocene. This situation is closely analogous to the Rhone Glacier example of Goehring et al. (2012), and comparison of 14C and 10Be concentrations in low-elevation samples would permit estimating the timing of initial thinning caused by early Holocene ice shelf collapse as well as thickening caused by subsequent ice shelf formation. Thus, we propose to measure 14C concentrations in samples with early to middle Holocene apparent exposure ages at the two sites shown in Figure 9. These samples are held by co-PI Balco at BGC. Figure 9. 10Be data from the Sjogren-Boydell fjord, northeastern Antarctic Peninsula (left) and Drygalski Glacier, same general area (right), from Balco et al. (2013). These data are unusual in that erratics with both early-middle Holocene (5-7 ka) and extremely young (<500 yr) apparent exposure ages co-occur at low elevations. This observation, with independent evidence for a mid-Holocene collapse and later regrowth of adjacent parts of the Larsen Ice Shelf (see Balco et al. 2013 and references therein), leads to the hypothesis that these data record complete deglaciation of the sites in the early Holocene followed by glacier thickening due to ice shelf formation. Thus, lowest-elevation samples with early Holocene ages should show disequilibrium 14C/10Be, and 14C measurements should enable an estimate of the duration of the ice-free period. 12 Research plan - summary. We propose to measure 14C in 15 samples from each of the six target areas described above. The 15 sample total includes replicate measurements of selected samples. Replicate measurements with 14C are particularly critical due to its slightly higher level of sample to sample scatter (6% vs 4%), compared to more common cosmogenic nuclides (e.g, 10Be, 3He), based on replicate measurements of the same material as part of CRONUS-Earth (Jull et al., 2013). We note however, that advances in the automation of 14C extraction and increases in the amount of quartz used has lowered intra-laboratory variability to ~1% (Lifton et al., in press). We expect similar performance by the Tulane laboratory; however, we propose to critically evaluate whether this is true by rigorous measurement of replicate samples. In addition, because at some sites we propose to analyse glacially transported erratics that were covered by ice at the LGM rather than bedrock samples, analysis of different erratics from the same locations is needed to test the hypothesis that the erratics were not disturbed by LGM ice cover. If this hypothesis is true, erratics with different pre-LGM apparent 10Be exposure ages will show indistinguishable 14C concentrations, indicating that they have experienced the same exposure history during the ca. 30,000year lifetime of the 14C inventory. Measurement of samples will be spread out over the three year project duration, 30 samples per year. While the main thrust of the proposed work is the measurement of in situ 14C, additional 10Be measurements are needed, as discussed above, at some of the sites, in particular the Thomas and Schmidt Hills. We propose measurement of 10Be in approximately 20 samples over the three year project period. Division of responsibilities. Responsibilities for this research will be divided between PI Goehring, co-PI Balco, and graduate students supervised by Goehring at Tulane University. Both Balco and Goehring are specialists in geologic applications of cosmogenic-nuclide geochemistry and have extensive field and analytical experience in this area. Goehring, in particular, is a specialist in analysis of in-situ-produced 14C in quartz, expertise which is only held by several researchers worldwide. Although this project does not involve Antarctic fieldwork, both Balco and Goehring are experienced in glacial-geologic mapping and exposure-dating in Antarctica; Balco has seven field seasons experience and Goehring one. Goehring, with assistance from Balco, will lead the project and will be responsible for supervising and/or carrying out all 14C measurements. Student participation in this research will be critical and students at Tulane will perform much of the analytical work under Goehring’s instruction and supervision. This is expected to form the basis of a PhD at Tulane. Balco will be primarily responsible for supplying samples from his own collections and acquiring them from other contacts and collaborators (detailed above in the site descriptions), and in addition will carry out the small number of additional 10Be measurements described above. Goehring, Goehring’s students, and Balco will collaborate on data management, analysis, and synthesis. In addition, we will collaborate closely with other specialists in Antarctic exposure-dating research Andy Hein (U. of Edinburgh) and Joanne Johnson (British Antarctic Survey). As described above and in attached letters of support, we rely on this collaboration to obtain samples from sites in the Weddell Sea embayment collected by these researchers. We will also collaborate closely with them during subsequent data analysis and synthesis. 13 Summary: intellectual merit. The primary goal of this proposal is to use in-situ cosmogenic 14C measurements on existing samples collected from several sites in Antarctica to resolve major ambiguities in present LGM-to-present ice sheet reconstructions. This, in turn, is important because of the critical nature of accurate reconstructions of ice sheet change in constraining reconstructions of past sea level change. We aim to achieve this goal by targeted 14C measurements in the region of Antarctica where the greatest ambiguity in existing exposureage reconstructions of past ice sheet change exist, specifically the Weddell Sea embayment. In addition, radiocarbon measurements applied where independently constrained deglaciation chronologies already exist can potentially yield new chronological information for past periods of ice sheet thickening, which is important in sea-level reconstructions and ice sheet-climate model validation. We propose to realize this potential by targeted measurements at sites chosen for the quality of the existing deglaciation chronology. Summary: broader impacts. The proposed work will greatly clarify the behavior of the West Antarctic ice sheet, particularly in the Weddell Sea sector, between the LGM and present by resolving ambiguous exposure age chronologies of ice sheet thickness changes using in-situ-produced carbon-14. Resolving ambiguities in Antarctic ice sheet behavior are key to fully understand reconstructions of paleo-sea level change, particularly the ability to identify the ice sheet that contributed towards a change in sea level. Furthermore, constraints on ice sheet thickness from the LGM-to-present allow one to place modern changes in ice sheet dimensions within the context of past changes and refine predictions of future ice sheet behavior. The proposed work will form the basis of graduate student research at Tulane University (one PhD). The students will be integrated into the larger Antarctic ice sheet research community through attendance of national and/or specialized conferences, such as the West Antarctic Ice Sheet meeting. During the course of the proposed work, the student will gain valuable experience in extraction of in situ 14C from quartz, making them him/her one of the few with experience in in situ 14C techniques, in addition to other laboratory procedures and data processing. The students will be expected to participate and lead manuscript development. In addition to training of a PhD student, the proposed work represents a cornerstone of new faculty Goehring’s early career, including development of mentoring skills. During completion of the proposed work, Goehring will develop a new course on sea level change, both past and present, at Tulane University as part of Tides Program (http://tulane.edu/college/tides/). A course of this nature is particularly germane in Louisiana and New Orleans. In the course students will be exposed to both modern sea level changes and its consequences, as well as perspectives that paleo sea level reconstructions provide. An understanding of the vital role that ice sheets play in controlling sea level will be a cornerstone of the course, with knowledge gained in the course of the proposed work continuously presented by Goehring and his student, given the student valuable teaching experience. Results of previous work. COLLABORATIVE RESEARCH: Terrestrial Exposure-Age Constraints on the last Glacial Maximum Extent of the Antarctic Ice Sheet in the Western Ross Sea. PIs Balco, Goehring, and Claire Todd (Pacific Lutheran University). $420,708 among three institutions; 8/1/2014 - 7/31/2016. 14 This project aims to develop a cosmogenic-nuclide exposure-age chronology for changes in ice thickness at the Last Glacial Maximum in the outer Ross Sea embayment. This is important because changes in ice sheet thickness and grounding line position have been hypothesized to be an important contributor to rapid global sea level change during so-called Meltwater Pulse 1A 14,500 years ago. Thus, the purpose of this project is to obtain geological evidence from the closest available ice-free areas to the central and outer Ross Sea that would permit evaluating this hypothesis. We completed fieldwork for this project in the 2014-15 Antarctic field season. We identified and mapped a range of geological evidence, including moraines and glacial drift, for past higher-than-present ice surface elevations in northern Victoria Land, adjacent to the hypothesized LGM position of the grounding line of the West Antarctic Ice Sheet in the Ross Sea (Figure 10). Geomorphic characteristics of these deposits are consistent with an LGM age, although this cannot be established without exposure-age data. At present, we are beginning sample preparation for cosmogenic-nuclide measurements planned to be conducted in summer and fall 2015. Figure 10. Results of fieldwork in the Tucker Glacier region of northern Victoria Land in the 2014-15 Antarctic field season. The Tucker Glacier flows from right (W) to left (E) in this image; its calving margin is several km to the left (E) of the image extent. A tributary, the Whitehall Glacier, flows in from the north. Green circles on the west side of the confluence represent granite bedrock samples collected from a range of elevations for the purpose of establishing the LGM ice surface elevation in this region using the approach described in Figure 4 above. On the other side of the Whitehall Glacier, a series of moraines and erratics composed of the same granite record westward transport of granitic erratics at a time when the surface of the Tucker Glacier was up to 200 m higher than present. Red circles denote locations of samples collected to establish an exposure-age chronology for these deposits. 15