Volume 115, Issue D18
Composition and Chemistry
Free Access

Sulfate sources and oxidation chemistry over the past 230 years from sulfur and oxygen isotopes of sulfate in a West Antarctic ice core

S. A. Kunasek

S. A. Kunasek

Department of Earth and Space Sciences, University of Washington, Seattle, Washington, USA

Search for more papers by this author
B. Alexander

B. Alexander

Department of Atmospheric Sciences, University of Washington, Seattle, Washington, USA

Search for more papers by this author
E. J. Steig

E. J. Steig

Department of Earth and Space Sciences, University of Washington, Seattle, Washington, USA

Search for more papers by this author
E. D. Sofen

E. D. Sofen

Department of Atmospheric Sciences, University of Washington, Seattle, Washington, USA

Search for more papers by this author
T. L. Jackson

T. L. Jackson

Department of Chemistry and Biochemistry, University of California San Diego, La Jolla, California, USA

Search for more papers by this author
M. H. Thiemens

M. H. Thiemens

Department of Chemistry and Biochemistry, University of California San Diego, La Jolla, California, USA

Search for more papers by this author
J. R. McConnell

J. R. McConnell

Desert Research Institute, Nevada System of Higher Education, Reno, Nevada, USA

Search for more papers by this author
D. J. Gleason

D. J. Gleason

Department of Earth and Space Sciences, University of Washington, Seattle, Washington, USA

Search for more papers by this author
H. M. Amos

H. M. Amos

Department of Atmospheric Sciences, University of Washington, Seattle, Washington, USA

Now at Department of Earth and Planetary Sciences, Harvard University, Cambridge, Massachusetts, USA.

Search for more papers by this author
First published: 28 September 2010
Citations: 42

Abstract

[1] The sulfur and oxygen isotopic composition of sulfate in polar ice cores provides information about atmospheric sulfate sources and formation pathways, which have been impacted regionally by human activity over the past several centuries. We present decadal scale mean ice core measurements of Δ17O, δ34S, Δ33S, and Δ36S of sulfate over the past 230 years from the West Antarctic Ice Sheet (WAIS) Divide deep ice core drill site (WDC05-A). The low mean δ34S of non–sea-salt sulfate at WAIS Divide (6.0 ± 0.2‰) relative to East Antarctic coastal and plateau sites may reflect a combination of stronger influence of volcanogenic and/or stratospheric sulfate with low δ34S and the influence of frost flowers on the sea-salt sulfate-to-sodium ratio. Δ33S and Δ36S measurements are all within analytical uncertainty of zero but do not contradict a contribution of stratospheric sources to background sulfate deposition at WAIS Divide. Δ17O of non–sea-salt sulfate shows a small but significant increase between the late 1700s (1.8‰ ± 0.2‰) and late 1800s (2.6‰ ± 0.2‰), but the influence of stratospheric scale volcanic events on Δ17O in the early 1800s remains uncertain. An isotope mass balance model shows that the lack of change in Δ17O of non–sea-salt sulfate from the mid-1800s to early 2000s (2.4‰–2.6‰ ± 0.2‰) is consistent with previous atmospheric chemistry model estimates indicating preindustrial to industrial increases in O3 as high as 50% and decreases in OH of 20% in the southern polar troposphere, as long as H2O2 concentrations also increase by over 50%.

1. Introduction

[2] Atmospheric oxidants (e.g., O3, OH, H2O2) are the primary sink for most reduced trace gases that contribute to air pollution and climate change (e.g., CO, CH4). The abundance of atmospheric oxidants thus limits the atmospheric residence time of reduced trace gases. Global chemical transport and climate models of past atmospheric oxidation chemistry suggest that recent increases in anthropogenic emissions due to biomass burning and fossil fuel combustion have altered the global abundances of tropospheric oxidants [Martinerie et al., 1995; Wang and Jacob, 1998; Mickley et al., 1999; Lelieveld and Dentener, 2000; Grenfell et al., 2001; Hauglustaine and Brasseur, 2001; Lelieveld et al., 2002; Shindell et al., 2003; Lamarque et al., 2005]. While these models agree on the sign of preindustrial to industrial change in O3 concentration (25% to >60% increase), estimates of changes in OH concentration vary in both sign and magnitude (+6% to −33%). Improved quantitative constraints on past changes in atmospheric oxidation chemistry are necessary to improve global atmospheric chemistry and climate model predictions of future changes in response to additional increases in anthropogenic emissions.

[3] The oxygen isotope anomaly (Δ17O ≈ δ17O − (0.52 × δ18O) where δxO = ((xO/16O)sample/(xO/16O)standard) − 1, with x = 17 or 18 and Vienna Standard Mean Ocean Water (V-SMOW) is the standard) of sulfate in ice cores shows promise for empirical validation of model-calculated paleoatmospheric oxidation chemistry [Alexander et al., 2002, 2004]. The Δ17O of sulfate (Δ17O(SO42−)) reflects the relative importance of different oxidants (e.g., O3, OH, H2O2) in the production of sulfate because the oxidants transfer different Δ17O to oxidation products [Savarino et al., 2000]. Ice core records of Δ17O(SO42−) and Δ17O(NO3) over the past three centuries from Site A, Greenland, show a strong perturbation attributed to high levels of biomass burning in North America in the late 1800s/early 1900s, with a weaker perturbation due to increasing fossil fuel burning emissions during the twentieth century [Alexander et al., 2004]. Global climate and chemical transport models suggest that oxidation chemistry in the southern polar low to midtroposphere (>600 mb; 60°S–90°S) has also been perturbed on this time scale, with O3 increasing by 10%–50% and OH decreasing by 0%–20% due to increases in NOx (=NO + NO2), CO, and CH4 concentrations since the preindustrial period [Wang and Jacob, 1998; Mickley et al., 1999; Shindell et al., 2003]. However, no observational record from the Southern Hemisphere has yet been available to validate these model estimates.

[4] Sulfur isotope data provide information complementary to Δ17O(SO42−) concerning the sources of sulfate (e.g., sea-salt, marine biogenic, terrigenous and biogenic continental, and volcanogenic emissions). Sulfur isotopes are reported as δxS(‰) = (((xS/32S)sample/(xS/32S)standard) − 1) × 1000, where x = 33, 34, or 36 and Canyon Diablo Triolite (CDT) is the standard. δ34S analysis has been widely used to examine partitioning between various tropospheric sources of sulfate, which produce sulfate with different δ34S [Nielsen, 1974; Rees et al., 1978; Calhoun et al., 1991; Nielsen et al., 1991; McArdle and Liss, 1995; Patris et al., 2000a]. Measurements of Δ33S (= δ33S − 1000 × (1 + δ34S/1000)0.515 − 1(‰)) and Δ36S (= δ36S − 1000 × (1 + δ34S/1000)0.190 − 1(‰)) can be used to investigate the importance of stratospheric sources of sulfate, since only sulfate of stratospheric origin is expected to have nonzero Δ33S and Δ36S in the present day [Farquhar et al., 2001; Savarino et al., 2003b].

[5] Here we present the first ice core measurements of the multiple isotope composition of sulfate (Δ17O, δ34S, Δ33S, Δ36S) spanning the preindustrial-industrial transition (late 1700s to present) from the West Antarctic Ice Sheet (WAIS) Divide (79°28.1′S 112°5.2′W). The isotopic composition of sulfate is expected to be preserved during burial in the polar icepack, enabling ice core measurements to be used to reconstruct paleoatmospheric conditions [Lloyd, 1968; Patris et al., 2000a; Alexander et al., 2002, 2003]. We use sulfur isotopes of sulfate to estimate relative sulfate source contributions in West Antarctica and examine differences relative to East Antarctica. We interpret ice core Δ17O(SO42−) quantitatively using the partitioning between tropospheric sulfate formation pathways at WAIS Divide extracted from a global chemical transport model for the present day [Alexander et al., 2009]. We then examine whether ice core Δ17O(SO42−) changes are consistent with reported tropospheric oxidant abundance changes from model simulations of the preindustrial atmosphere.

2. Controls on Isotopic Composition of Atmospheric Sulfate

[6] Atmospheric Δ17O(SO42−) reflects the composition of its precursor SO2, which is in isotopic equilibrium with water vapor (Δ17O = 0‰), and the transfer of isotopically anomalous oxygen atoms (Δ17O > 0‰) from oxidants during SO2 oxidation to sulfate [Savarino et al., 2000]. Because of the large Δ17O of tropospheric ozone (25‰–35‰) [Schueler et al., 1990; Krankowsky et al., 1995; Johnston and Thiemens, 1997; Krankowsky et al., 2000; Lyons, 2001], SO2 oxidation by ozone on cloud water droplets or aerosols is expected to produce large tropospheric Δ17O(SO42−) of 6.3‰–8.8‰, following the observation that ozone transfers one quarter of its Δ17O signature to sulfate during SO2 oxidation [Savarino et al., 2000]. The transfer of Δ17O from O3 to sulfate found by this early laboratory study [Savarino et al., 2000] may be an underestimate since it did not examine the potential for preferential transfer of the terminal oxygen atom of ozone (i.e., O-O-Q) that is enriched in Δ17O [Bhattacharya et al., 2008], which occurs in other O3 oxidation processes (e.g., NO + O3 [Savarino et al., 2008]). In-cloud SO2 oxidation by H2O2 is expected to produce Δ17O(SO42−) of 0.5–1‰, based on the Δ17O of tropospheric H2O2 (1‰–2‰) and the observed transfer of one-half the Δ17O signature of H2O2 during SO2 oxidation [Savarino and Thiemens, 1999; Savarino et al., 2000]. By contrast, due to rapid oxygen isotope exchange between OH and water vapor [Dubey et al., 1997; Lyons, 2001], gas-phase SO2 oxidation by OH is expected to produce Δ17O(SO42−) = 0‰ throughout most of the troposphere, although it has been suggested that Δ17O of OH may be non-zero in the dry polar atmosphere [Morin et al., 2007]. Partitioning between the dominant tropospheric sulfate production pathways is controlled by oxidant abundances, cloud liquid water content, which affects the relative importance of gas and aqueous phase reactions, and pH, which affects the partitioning of dissolved SO2 as SO32− with which O3 reacts most rapidly [Chameides, 1984; Calvert et al., 1985]. Sulfate produced in the stratosphere is expected to reflect gas phase SO2 oxidation by stratospheric OH [Alexander et al., 2002; Savarino et al., 2003a], which has nonzero Δ17O (2‰–45‰) [Lyons, 2001; Liang et al., 2006; Zahn et al., 2006]. However, some episodic stratospheric scale volcanic events cause stratospheric SO2 injection large enough to titrate stratospheric OH, leading to sulfate formation via a different oxidation pathway producing Δ17O(SO42−) = 0‰ [Savarino et al., 2003a]. Atmospheric transport and deposition processes fractionate oxygen isotopes of sulfate according to mass-dependent relationships (δ17O ≈ 0.52 × δ18O) [Matsuhisa et al., 1978], such that Δ17O(SO42−) values are conserved.

[7] Atmospheric δ34S of sulfate reflects the relative contributions of different sources of sulfate [Calhoun et al., 1991; McArdle et al., 1998; Patris et al., 2000a; Alexander et al., 2003; Pruett et al., 2004; Jonsell et al., 2005] because different tropospheric sulfur sources display different ranges of δ34S (Figure 1). While δ34S of sea-salt sulfate is well constrained to 21‰ [Rees et al., 1978], and estimates of δ34S of marine biogenic sources fall within a small range (14‰–22‰) [Calhoun et al., 1991; McArdle et al., 1998; Patris et al., 2000b], the range of possible δ34S from tropospheric and stratospheric scale volcanism is comparatively large (−6 to +17‰) [Nielsen et al., 1991; Baroni et al., 2008] (Figure 1). Terrigenous continental material also displays a large range of δ34S values (0‰–20‰) [Nielsen et al., 1991]. Background (nonvolcanic) sulfate from the lower stratospheric reservoir has been estimated to have δ34S of 2.6‰ [Castleman et al., 1973], although values as low as −24‰ have been observed higher in the stratosphere [Castleman et al., 1974]. The influence of sulfur isotope fractionation in different sulfur oxidation pathways [Saltzman et al., 1983; Tanaka et al., 1994] and Rayleigh fractionation during transport must also be considered in order to deduce tropospheric sulfur source partitioning from δ34S.

Details are in the caption following the image
(a) Literature summary of reported δ34S ranges of sulfur sources including (1) sea-salt [Rees et al., 1978], (2) marine biogenic theoretical range [Calhoun et al., 1991], (3) marine biogenic at South Pole, Antarctica [Patris et al., 2000a], (4) continental [Nielsen et al., 1991], (5) volcanogenic [Nielsen et al., 1991], (6) volcanogenic at South Pole and Dome C, Antarctica [Baroni et al., 2007, 2008], (7) background stratospheric [Castleman et al., 1973]. Where source signature estimates are derived from a single measurement, the range indicates 2σ analytical uncertainties. (b) Summary of measured background δ34Snss ranges from Antarctic ice core studies including measurements from (8) South Pole [Patris et al., 2000a], (9) Vostok and Dome C [Alexander et al., 2003], (10) Dome A [Jonsell et al., 2005], (11a) coastal East Antarctica [Jonsell et al., 2005], (12) West Antarctic RIDSA core site [Pruett et al., 2004], (13a) WAIS Divide (this study). All studies report δ34Snss based on k = 0.25 (see text for k definition). Dotted bars reflect the calculation of δ34Snss based on different k, including (11b) k = 0.09 in coastal Antarctica [Jonsell et al., 2005] and (13b) k = 0.07 in this study.

[8] Sulfur isotope anomalies of sulfate reflect the influence of stratospheric sources of sulfate. Unlike δ34S, Δ33S, and Δ36S are conserved during atmospheric transport, deposition, and oxidation processes, which fractionate sulfur isotopes following mass-dependent relationships (δ33S = 1000 × (1 + δ34S/1000)0.515 − 1 and δ36S = 1000 × (1 + δ34S/1000)0.190 − 1). Nonzero sulfur isotope anomalies are produced during UV photolysis of SO2 at wavelengths less than 310 nm, which occur only above the tropopause [Farquhar et al., 2001; Savarino et al., 2003b]. Studies of ice core sulfate [Alexander et al., 2003; Savarino et al., 2003b; Baroni et al., 2008] and atmospheric sulfate aerosols [Romero and Thiemens, 2003; Mather et al., 2006] reinforce the interpretation of Δ33S as a tracer for stratospheric sulfur sources, although the utility of Δ36S as a stratospheric tracer is limited by large analytical uncertainties [Alexander et al., 2003; Baroni et al., 2008].

3. Methods

[9] Isotopic and major ion concentration measurements were made on a 70 m ice core drilled at WAIS Divide during the summer 2005 field campaign, WDC05-A (78°55′S, 114°13′W). Cores were shipped frozen to the U. S. National Ice Core Laboratory (NICL) in Colorado, where core allocations for sulfate isotope and major ion analyses were cut with a bandsaw in a clean laboratory at −22°C. To avoid analysis of contaminants, only sections of ice taken from the inner portion of the core were used for isotopic analysis of sulfate and major ions. Inner core sections were kept frozen until analysis.

[10] Analysis of major trace elements of WDC05-A was performed at the Desert Research Institute following methods adapted from McConnell [2002], in which an ice core melter is linked with a continuous flow analysis system (CFA) to provide real-time measurements of major elements with high depth resolution (centimeter scale). Sulfur, sodium, calcium, magnesium, and a range of other elements are measured with two Thermo-Finnigan Element2 high-resolution inductively coupled plasma mass spectrometers (HR-ICP-MS) operating in parallel. Analytical uncertainties of sodium, sulfur, calcium, and magnesium measurements are ±5% (1σ). The depth-age relationship for the 70 m core (“WDC05-A:1”) is derived from sodium and total sulfur concentrations and has uncertainty of ±1 year [Banta et al., 2008; Mischler et al., 2009]. Total sulfur concentrations of WDC05-A were used to determine the volume of ice needed to achieve ∼5 μmol of sulfate per sample for isotopic analysis. We assume dissolved sulfate is ∼70% of total sulfur at WAIS Divide following comparisons between ion chromatographic (IC) and ICPMS duplicate measurements of a nearby WAIS Divide ice core (WDC05-Q, 79°28.05′S, 112°5.14′W) (J. Cole-Dai, personal communication). Three ice cores drilled near WDC05-A and WDC05-Q within 200 m of elevation differ by less than 10% in average sulfate concentrations (ITASE 00-1, RA, RB) [Dixon et al., 2004]. An uncertainty of ±10% in our assumption for dissolved sulfate (i.e., ∼70% of total sulfur) impacts our correction of Δ17O values by less than analytical uncertainty (0.2‰, 2σ) and has limited influence (<2%) on the estimated minimum volcanogenic/stratospheric sulfate contribution derived from δ34S measurements in our discussion.

[11] For isotopic analysis of sulfate, sample silver sulfate was prepared following methods described by Kunasek et al. [2008] and isotope analysis of silver sulfate was performed following the methods of Savarino et al. [2001]. At the University of Washington, ice was melted and combined in clean 4 L beakers then evaporated to ∼50 mL in a class 100 clean hood in order to concentrate sulfate. Concentrated samples were then pumped through an automated system for separation of anions by ion chromatography (IonPac® AS19 separation column (4 × 250 mm); 7–15 mM KOH) and conversion of sulfate to silver form (Ag2SO4) by cation exchange (AMMS® III membrane (4 mm); 2.5 mM Ag2SO4 regenerant) [i.e., Kunasek et al., 2008]. Silver sulfate fractions were dried and shipped to University of California-San Diego, where they were transferred to and dried in quartz boats. In a continuous flow system [i.e., Savarino et al., 2001], the Ag2SO4 was pyrolyzed to release O2 and SO2, which are then trapped separately. The SO2 is further converted to SF6, following established procedures [Farquhar et al., 2001]. The three oxygen isotopologues of O2 and four sulfur isotopologues of SF6 were then analyzed using Finnigan MAT 251 and 252 isotope ratio mass spectrometers (IRMS), respectively. For oxygen isotopologues, we report only Δ17O because Δ17O values are conserved during sample processing that fractionates δ17O and δ18O according to mass-dependent relationships (δ17O ≈ 0.52 × δ18O) [Matsuhisa et al., 1978].

[12] The contribution of sea-salt (ss) and non–sea-salt (nss) sulfate were determined from Na+ and SO42− concentrations following [SO42−]total = [SO42−]nss + k[Na+], where k is the sulfate-to-sodium ratio of sea salt. (Hereafter, the subscripts “nss” and “ss” refer to non–sea-salt and sea-salt sulfate, respectively.) We calculate the sea-salt contribution using k = 0.25 to reflect the sulfate-to-sodium ratio of the open ocean [Holland, 1978], following the majority of previous sulfur isotope studies in Antarctica [Alexander et al., 2002, 2003; Patris et al., 2000a; Pruett et al., 2004]. We also estimate a minimum sea-salt contribution using the lowest estimated coastal Antarctic k value of 0.07 [Wagenbach et al., 1998], which reflects a strong contribution of frost flowers with depleted k to total sea-salt deposition. Isotopic values are corrected for the influence of sea salt by mass balance, assuming Δ17O(SO42−)ss = 0‰ [Alexander et al., 2002] and δ34Sss = 21‰ [Rees et al., 1978]. We also estimate the sulfate contribution of terrigenous continental material (“terr”)(e.g., from crustal CaSO4 and MgSO4) using non–sea-salt calcium and non–sea-salt magnesium concentrations following [SO42−]terr = m[Ca2+]nss and [SO42−]terr = n[Mg2+]nss, where m and n are the average sulfate-to-calcium and sulfate-to-magnesium ratios in crustal material (∼0.18 and 0.24, respectively) [e.g., Patris et al., 2002]. Non–sea-salt concentrations of magnesium and calcium are determined using known ratios of calcium to sodium (0.038) and magnesium to sodium (0.12) in seawater, as described in the work of Legrand et al. [1997].

[13] Ice core Δ17O(SO42−)nss is interpreted using the fractional contribution of different atmospheric sulfate production pathways (e.g., oxidation by OH, H2O2, O3) calculated using a global chemical transport model of the present day (1989–1991) described by Alexander et al. [2009] (GEOS-Chem, http://acmg.seas.harvard.edu/geos/). The model bulk cloud water pH is set to 5.0 at the high end of the range found in marine stratocumulus clouds (3.3–5.0 [Faloona, 2009, and references therein]), since the remote southern polar region is expected to be impacted minimally by anthropogenic emissions that acidify precipitation. The global model results at the WAIS Divide site include the influence of tropospheric transport. In addition to the three dominant SO2 oxidation pathways described in section 2, the global model also considers SO2 oxidation by O3 on sea-salt aerosols [Alexander et al., 2005] and by O2 catalyzed by transition metals (Fe(III), Mn(II)) in calculating sulfate oxidation pathway partitioning [Alexander et al., 2009].

4. Ice Core Observations and Model Results

[14] Ice core measurements of multidecadal mean sulfur and oxygen isotopes are shown in Table 1 and Figure 2, along with multidecadal mean sodium and sulfate concentrations. Each sample represents 27–44 years of snow accumulation. Total sulfur concentrations (Figure 2a) show a strong (56%) increase in the early 1800s sample, which spans two stratospheric scale volcanic eruptions recorded in other Antarctic ice cores: an unknown volcanic eruption in 1810 and Mt. Tambora in 1815 [Dai et al., 1991; Cole-Dai et al., 1997]. The impact of these eruptions is also shown in detail in the high-resolution non–sea-salt sulfate record in Figure 3, assuming k = 0.07 (using k = 0.25 leads to some negative non–sea-salt sulfate concentration values). The contribution of sea-salt sulfate to total sulfate (not shown) derived from sulfur and sodium measurements (Figure 2a) depends strongly on k (14%–21% for k = 0.25 and 4%–6% for k = 0.07) but varies little temporally (±1%) with the exception of the early 1800s, where the stratospheric-scale volcanic influence reduces the contribution of sea salt (5% and 2% for k = 0.25 and 0.07, respectively). Calcium and magnesium concentrations of our samples fall within 1.0–1.8 and 2.8–3.2 ppb, respectively. The calculated contribution of terrigenous continental material to total sulfate (not shown) is <2% throughout the ice core record whether non–sea-salt calcium or non–sea-salt magnesium is used as the conservative tracer. Raw δ34S measurements vary between 5.5‰ and 9.3‰ (mean 6.8‰), with a maximum in the late 1800s/early 1900s and relative minima during the early 1800s stratospheric scale volcanic influence and the late 1900s (Figure 2b). δ34Snss shows similar temporal structure (Figure 2c), but the absolute values depend strongly on k (mean 3.1‰ for k = 0.25 versus mean 5.9‰ for k = 0.07). Δ33S and Δ36S measurements (Figures 2d and 2e) are all within 2σ analytical uncertainty of zero, consistent with mass-dependent chemical processes. Because of the large analytical uncertainties of the Δ36S measurement and consequent ambiguities in its use as a stratospheric sulfur source tracer [Baroni et al., 2008], we focus on Δ33S measurements in our discussion. Both raw Δ17O(SO42−) measurements (Figure 2f) and corrected Δ17O(SO42−)nss (Figure 2g) show small but significant changes during the ice core record, with lower Δ17O(SO42−)nss in the late 1700s (2.2‰ and 1.8‰ for k = 0.25 and 0.07, respectively) and consistent higher Δ17O(SO42−)nss from the late 1800s to 2005 (mean Δ17O(SO42−)nss = 3.0 and 2.5‰ for k = 0.25 and 0.07, respectively, from 1880 to 2005). The global model of the present day finds that sulfate reaching WAIS Divide is predominantly produced by in-cloud SO2 oxidation by H2O2 (68%), with in-cloud SO2 oxidation by O3 and gas phase SO2 oxidation by OH comprising the bulk of the remaining sulfate production (16% and 12%, respectively). SO2 oxidation by O3 on sea-salt aerosols and by O2 with transition metal catalysis are minor contributors to total sulfate reaching WAIS Divide (<1% and <4%, respectively) in the model, as is primary sulfate from anthropogenic emissions (<1%).

Details are in the caption following the image
Time series of WAIS Divide ice core measurements including (a) average of high-resolution concentration measurements of sulfate and sodium; (b) δ34S (solid); (c) upper and lower estimates of δ34Snss (dash-dotted); (d) Δ33S (solid); (e) Δ36S (solid); (f) Δ17O(SO42−) (solid); (g) upper and lower estimates of Δ17O(SO42−)nss (dash-dotted). Concentrations are in ppb and isotopic ratios are shown as ‰. 2σ errors for δ34S, Δ33S, Δ36S, and Δ17O measurements are shown as dashed lines (±0.2‰, ±0.1‰, ±1.6‰, ±0.2‰, respectively). Values of k = 0.25 and 0.07 are used to estimate range of sea-salt sulfate correction (see text). See text for description of time scale WDC05A:1.
Details are in the caption following the image
Time series of non–sea-salt sulfate concentration (ppb) at WAIS Divide, calculated from high-resolution measurements of total sulfur and sodium concentration. We assume dissolved sulfate is 70% of total sulfur and use a dissolved sulfate-to-sodium ratio (k) of 0.07 here. (Assuming k = 0.25 results in some negative non–sea-salt sulfate concentrations.) Inset shows expanded view of 1940–1960.
Table 1. Raw and Non–Sea-Salt δ34S and Δ17O of Sulfate From WAIS Divide
Start Year End Year δ34S (‰) Δ17O (‰) k = 0.25 k = 0.07
fss (%) δ34Snss (‰) Δ17Onss (‰) fss (%) δ34Snss (‰) Δ17Onss (‰)
2005 1977.5 5.5 2.4 19.4 1.8 2.9 5.4 4.6 2.5
1977.5 1950.5 7.2 2.2 21.5 3.4 2.8 6.0 6.3 2.4
1923.5 1879.5 9.3 2.3 20.7 6.2 3.0 5.8 8.6 2.5
1879.5 1837.5 8.0 2.5 21.5 4.4 3.1 6.0 7.2 2.6
1837.5 1809.5 4.0 2.3 14.3 1.2 2.6 4.0 3.3 2.4
1809.5 1773.5 6.6 1.7 20.6 2.9 2.2 5.8 5.7 1.8

5. Sulfur Isotopes and Source Partitioning

5.1. δ34S and Sulfur Sources

[15] Figure 1 summarizes spatial differences in ice core δ34Snss throughout the Antarctic, including the findings of the present study. The range of ice core δ34Snss (1.2‰–6.2‰, k = 0.25) from WAIS Divide is consistent with recent measurements of another West Antarctic ice core from the RIDSA campaign (−0.7‰ to +6.8‰, k = 0.25; 78.73°S, 116.33°W) [Pruett et al., 2004] (Figure 1), confirming that West Antarctic background δ34Snss is significantly lower than reported ice core measurements from both East Antarctic coastal and plateau sites (minimum δ34Snss = 11‰) [Patris et al., 2000a; Alexander et al., 2003; Jonsell et al., 2005; Baroni et al., 2008] (Figure 1). Ice core δ34Snss therefore does not exhibit a clear trend with elevation or distance inland, unlike ice core non–sea-salt sulfate concentrations, which generally decrease with increasing altitude at least near the coast [Stenberg et al., 1998; Proposito et al., 2002; Bertler et al., 2005]. Previous authors reviewed the potential impacts on δ34Snss from spatial variability in fractionation during atmospheric transport (i.e., Rayleigh fractionation) and sulfate oxidation [Alexander et al., 2003; Jonsell et al., 2005] and suggest fractionation due to these processes relates to distance inland, elevation, and/or transport path length. These processes alone therefore cannot explain the low δ34Snss of WAIS Divide. We thus suggest that the lower δ34Snss values of West Antarctica relative to East Antarctica is due to a relatively stronger contribution of sulfur sources with low δ34S, such as regional tropospheric volcanism and/or background stratospheric influence, as also suggested by Pruett et al. [2004]. Although the influence of a terrigenous continental sulfate source may also help explain the low δ34Snss at WAIS, our estimate based on ice core non–sea-salt calcium and non–sea-salt magnesium concentrations suggests this contribution is negligible (<2%). Following previous authors, we also assume limited influence of continental biogenic sulfate sources due to the remoteness of the Antarctic continent from Southern Hemisphere landmasses [Patris et al., 2000a; Alexander et al., 2003; Pruett et al., 2004; Jonsell et al., 2005].

[16] A large difference in volcanic and/or stratospheric sulfate contributions between West and East Antarctica is possible given differences in meteorology and topography. Of the 26 Antarctic region volcanoes listed by the Smithsonian volcano data base (http://www.volcano.si.edu/world/) as active within the Holocene, all are within or bordering on the Western Hemisphere in proximity to regions of cyclogenesis that impact West Antarctica [King and Turner, 1997; Simmonds et al., 2003; Carrasco et al., 2003]. Continental volcanogenic emissions from regions such as the southern Andes have also been suggested to contribute to the Antarctic sulfate budget [Jonsell et al., 2005]. Because of the lower mean elevation of West versus East Antarctica, cyclone activity penetrates further into West Antarctica [King and Turner, 1997], potentially leading to a greater contribution of volcanogenic sources to sulfate deposition in this region. However, only a few Antarctic region volcanoes are known to be currently active (i.e., Mt. Erebus, Deception Island, and Buckle Island), and most studies suggest a minor contribution of volcanic sources to sulfate deposition in Antarctica (max 10%–30%) [Radke, 1982; Rose et al., 1985; Zreda-Gostynska et al., 1997; Minikin et al., 1998], although these studies suggest that large spatial heterogeneity is possible. The greater influence of cyclone activity on West relative to East Antarctica may also lead to greater entrainment of stratospheric air in the troposphere due to tropopause folding associated with strong cyclone activity [Danielsen, 1968; Holton et al., 1995; Stohl et al., 2003]. However, because Antarctic cyclones also entrain marine air from the oceanic regions in which they originate, the net influence of cyclone activity in West Antarctica on the relative contribution of stratospheric sources to sulfate deposition is ambiguous. Most studies of sulfur source attribution in East Antarctica are consistent with negligible influence of stratospheric sources [Patris et al., 2000a; Alexander et al., 2003], but a study of coastal East and West Antarctica could not rule out a stratospheric sulfur contribution as high as 10%–33% [Minikin et al., 1998].

[17] We can estimate whether volcanogenic and stratospheric sulfur source contributions are a reasonable explanation for the low 34Snss at WAIS Divide, using the following quantitative representation of sulfur source δ34S signature mixing [e.g., Patris et al., 2000a],
equation image
where f and δ34 are the fractional contribution and corresponding δ34S of each sulfur source, including sea-salt (ss), non–sea-salt (nss), marine biogenic (mb), volcanic (v), and stratospheric (s). Mean background δ34Snss in the WAIS ice core is 3.7‰ (k = 0.25), when the data point of the early 1800s (1810–1837) is omitted due to the known stratospheric-scale volcanic influence during this time [Dai et al., 1991; Cole-Dai et al., 1997]. Assuming marine biogenic δ34S of 18.6‰ [Patris et al., 2000a] and a low range of −2‰ to +2‰ to represent both volcanic and stratospheric source δ34S signatures [Castleman et al., 1973; Nielsen et al., 1991; Baroni et al., 2008], this mean background ice core δ34Snss value suggests that the combination of volcanic and stratospheric sulfur sources contributes 72%–90% of West Antarctic surface sulfate deposition. Thus, contributions of stratospheric and volcanic sources of sulfur higher than those estimated in previous Antarctic studies (i.e., ∼33% stratospheric, ∼30% volcanic) [Radke, 1982; Rose et al., 1985; Minikin et al., 1998] are necessary to explain the low δ34Snss of West Antarctica relative to East Antarctica through differences in sulfur source contributions alone.

[18] Another factor that may contribute to the low δ34Snss of West Antarctica is the calculation of the sea-salt fraction. It has been suggested that frost flowers formed on sea ice, not the open ocean surface, are the dominant source of sea salt across the Antarctic continent [Wolff et al., 2003]. Processing of sea salt during frost flower production depletes sulfur relative to sodium abundances, potentially leading to an underestimate of non–sea-salt sulfate concentrations if k of open ocean water is assumed (i.e., k = 0.25) [Wagenbach et al., 1998; Rankin et al., 2002]. This effect is most noticable at coastal locations, where unrealistic negative non–sea-salt concentrations are frequently found unless very low k is used (e.g., 0.07, 0.09 [Wagenbach et al., 1998; Jonsell et al., 2005]) to reflect the influence of sulfate depletion in frost flowers. Some negative non–sea-salt concentration values are also found at WAIS Divide when k = 0.25 is used (not shown), suggesting k = 0.25 is also an overestimate for the WAIS Divide site. Using a low k of 0.07 instead of the classical open ocean value (k = 0.25) increases mean background ice core δ34Snss to 6.5‰ (see Table 1, Figure 1; again, data points of 1810–1837 omitted from mean).

[19] Using the higher mean background ice core δ34Snss (6.5‰, k = 0.07) in equation (1) and following similar assumptions as above for δ34S source signatures (δ34mb = 18.6‰, δ34v = δ34s = −2‰ to +2‰), the calculated contribution of volcanic and stratospheric sulfur sources to background West Antarctic sulfate deposition is reduced to 59%–73%. Thus even when strong frost flower influence is considered (i.e., k = 0.07), the low WAIS ice core δ34Snss values are best explained by an important contribution of stratospheric and/or volcanogenic sulfate sources (i.e., 59%–73% stratospheric and volcanic combined) near the maximum of previous independent estimates of Antarctic sulfate source attribution (i.e., max 33% stratospheric, 30% volcanic) [Radke, 1982; Rose et al., 1985; Minikin et al., 1998]. This interpretation of the δ34Snss record suggests marine biogenic sources contribute a maximum of 41% to sulfate at WAIS Divide. Previous studies have suggested a strong marine biogenic influence on high-elevation West Antarctica based on the correlation of seasonal cycles in non–sea-salt sulfate and sea ice extent [Dixon et al., 2004, 2005] and strong sea-salt sulfate contributions [e.g., Pruett et al., 2004]. An approximate doubling of non–sea-salt sulfate concentration between winter and summer (see Figure 3, inset) can be reconciled with a maximum mean annual marine biogenic sulfate contribution of 41% at WAIS Divide, if we assume negligible wintertime marine biogenic sulfate input and no seasonal cycle in snow accumulation. However, observations of the seasonal cycle in ice core non–sea-salt sulfate concentration near WAIS Divide vary interannually and between studies [Dixon et al., 2004; Pruett et al., 2004; this study], and there is no evidence to support a negligible wintertime marine biogenic sulfate input.

[20] Several additional factors may lead to an underestimate (overestimate) of the marine biogenic (volcanogenic and/or stratospheric) sulfate contribution based on the δ34Snss record and help reconcile the interpretations of ice core isotope and major ion records. Assuming a lower δ34S source signature for background stratospheric and/or volcanogenic sources (δ34v = δ34s < −2‰) would reduce the calculated combined contribution of stratospheric and volcanogenic sulfur sources. While such low δ34S have been observed for background stratospheric sulfate (i.e., min −24‰ [Castleman et al., 1974]) and stratospheric scale volcanic sulfate sources (i.e., min −6‰ [Baroni et al., 2008]), mean δ34S for these sources is not expected to be so low (see Figure 1). A reduced sea-salt source δ34S signature (δ34Sss < 21‰), due to depletion of heavy isotopes during frost flower formation, would also reduce the calculated contribution of volcanic and stratospheric sources at WAIS Divide but has not yet been investigated. Refinements to δ34S source signature characterization (e.g., δ34S of frost flowers) and/or additional independent estimates of sulfate source attribution (e.g., using aerosol 10Be or 35S [Minikin et al., 1998; Lee and Thiemens, 2001]) are critical for improving quantitative sulfate source attribution at WAIS Divide. At this time, temporal changes in ice core δ34Snss at WAIS Divide (Figure 2) are thus best interpreted qualitatively, with higher (lower) δ34Snss reflecting greater marine biogenic (stratospheric and/or volcanic) influence.

5.2. Δ33S and Stratospheric Sulfur Sources

[21] Ice core Δ33S measurements of background sulfate at WAIS are all within analytical uncertainty of zero (±0.2‰) (Figures 2d and 2e), suggesting no detectable stratospheric influence on sulfate deposited at WAIS Divide following previous interpretations of Δ33S [Alexander et al., 2003; Savarino et al., 2003b]. This finding is identical to previous background Δ33S measurements in East Antarctica [Alexander et al., 2003; Savarino et al., 2003b; Baroni et al., 2007, 2008] and consistent with other Antarctic background sulfate source attribution studies that suggest dominance of tropospheric sulfate sources [Legrand and Mayewski, 1997; Bergin et al., 1998; Minikin et al., 1998]. However, the lack of significant perturbation to the WAIS ice core Δ33S record during the early 1800s period of known stratospheric scale volcanic influence necessitates explanation. Assuming that 36% of the sulfate in the early 1800s sample is from episodic stratospheric scale volcanism (see section 4), if 30% of the remaining 64% of background sulfate is also of stratospheric origin, consistent with the maximum contribution estimated by independent analyses of sulfate source attribution [Minikin et al., 1998], then stratospheric sources may contribute a total of up to 55% to sulfate of this sample (i.e., 36% episodic stratospheric-scale volcanic, 19% background stratospheric, 45% background tropospheric). In fact, despite the dominance (>60%) of stratospheric scale volcanogenic sulfate in ice core Δ33S samples in a recent East Antarctic study [Baroni et al., 2008], only half of the eight stratospheric-scale volcanic events identified displayed significant nonzero Δ33S before background sulfate contributions were corrected. This may be explained by considering that a stratospheric scale volcanic event increases both tropospheric and stratospheric SO2. Our results along with the findings of Baroni et al. [2008] suggest that background measurements of ice core Δ33S within analytical uncertainty of zero cannot yet be used to rule out significant contributions of stratospheric sources (e.g., approaching 60%). The lack of significant nonzero Δ33S throughout the WAIS ice core records, including the early 1800s sample, thus does not contradict a contribution of stratospheric sources to background sulfate deposited at WAIS Divide, as suggested by our ice core δ34Snss analysis (section 5.1).

6. Oxygen Isotopes and Oxidation Chemistry

6.1. Ice Core Variations in Δ17O of Sulfate

[22] The robustness of the increasing trend in WAIS ice core Δ17O(SO42−)nss between the late 1700s and late 1800s is weak due to the influence of stratospheric scale volcanism in the early 1800s (1810–1837 sample). Previous work indicates that stratospheric scale volcanic Δ17O(SO42−) signatures can vary greatly, with relatively small amounts of stratospheric SO2 injection resulting in high Δ17O(SO42−) (e.g., Pinatubo 1991 eruption: Δ17O(SO42−) = 4.3‰) due to the dominance of stratospheric SO2 oxidation by OH with high Δ17O, and relatively large amounts of stratospheric SO2 injection resulting in dominant stratospheric SO2 oxidation by secondary oxidation pathways with low Δ17O after OH is titrated (e.g., 1259 eruption of unknown origin: Δ17O(SO42−) = 0.7‰) [Savarino et al., 2003a]. Because only these two stratospheric scale volcanic eruptions have been measured for ice core Δ17O(SO42−) signatures, a reliable empirical relationship between stratospheric SO2 injection amount and a stratospheric scale volcanic Δ17O(SO42−) cannot yet be determined. The influence of the two early 1800s stratospheric-scale volcanic eruptions (1815 Tambora eruption and 1810 eruption of unknown origin) on ice core Δ17O(SO42−)nss of the early 1800s sample thus cannot be strongly constrained. The perceived trend of increasing Δ17O(SO42−)nss between the late 1700s and late 1800s is thus driven by only one data point (late 1700s). Several independent records, including terrestrial charcoal records [Marlon et al., 2009] and ice core δ13CH4 [Ferretti et al., 2005], suggest a Holocene minimum in global biomass burning in the late 1600s and 1700s, followed by an increase into the late 1800s. The low ice core Δ17O(SO42−)nss in the late 1700s (2.2 and 1.8‰ for k = 0.25 and 0.07) is coincident with the minimum of the global biomass burning records for the past 2000 years; however, extending the ice core record of Δ17O(SO42−)nss further back in time is necessary for clarifying and quantifying the relationship between these events. In the model-based interpretation that follows, we thus focus on the ice core Δ17O(SO42−)nss record from the mid-1800s to early 2000s, which corresponds to the period examined by most model studies of preindustrial to industrial change in atmospheric chemistry.

6.2. Model Interpretation of Ice Core Δ17O of Sulfate (1837–2005)

[23] Here we examine whether the lack of change in ice core Δ17O(SO42−)nss from 1837 to 2005 is consistent with model estimates of preindustrial to industrial changes in tropospheric oxidant concentrations in the southern polar region. Previous chemical transport and climate model studies have found preindustrial to industrial increases of 10%–50% for O3 and decreases of 0%–20% for OH (i.e., ([X]PD − [X]PI)/[X]PI for X = O3 or OH) [Wang and Jacob, 1998; Mickley et al., 1999; Shindell et al., 2003] in the lower to middle troposphere (600–1000 mb) of the southern polar region (60°S–90°S). Preliminary results of a more recent model simulation that generally agrees with these earlier literature values of preindustrial to industrial changes in O3 and OH in the lower to middle polar troposphere (20%–28% increase in O3; 8%–20% decrease in OH) suggests that H2O2 increases by 40%–52% from the preindustrial to the industrial in this region [Sofen et al., 2010]. An increase in atmospheric H2O2 in the southern polar region during the past century is qualitatively supported by ice core H2O2 concentration increases observed at high accumulation sites near the WAIS Divide, although additional work is needed to quantify an atmospheric H2O2 change from these records [Frey et al., 2006]. We calculate the change in tropospheric Δ17O(SO42−)nss associated with these model predictions of preindustrial to industrial change in tropospheric oxidant concentrations and compare this with the ice core Δ17O(SO42−)nss record from 1837 to 2005.

6.2.1. Assumptions in Calculation of Δ17O(SO42−)nss Change

[24] In our calculation of preindustrial to industrial change in tropospheric Δ17O(SO42−)nss, we assume no significant change in cloud water pH consistent with the lack of change in WAIS ice core records of sulfate (see Figure 2), the dominant control on cloud water pH and preliminary measurements of mineral acidity in East Antarctic ice cores (D. Pasteris, personal communication). We also assume negligible preindustrial to industrial changes in cloud liquid water abundance in the southern polar region. Although recent warming (since 1957) in West Antarctica has been significant (>0.1°C/decade) [Steig et al., 2009], records of snow accumulation over Antarctica do not indicate a statistically significant change during the last half century [Monaghan et al., 2006]. We do not examine the influence of stratospheric sulfate formation in our calculation of preindustrial to industrial change in Δ17O(SO42−)nss because of uncertainties in both the contribution of stratospheric sulfate sources at WAIS (see sections 5.15.2) and the Δ17O(SO42−)nss of background stratospheric sulfate sources, which has never been constrained by measurements. The potential influence of stratospheric sulfate is discussed at the end of section 6.2.2. Our analysis primarily focuses on whether the lack of change in ice core Δ17O(SO42−)nss is consistent with model estimates of changes in tropospheric oxidants alone.

6.2.2. Calculation of Preindustrial to Industrial Tropospheric Δ17O(SO42−)nss Change

[25] Results from the global model for the contributions of various tropospheric SO2 oxidation pathways to sulfate production at WAIS uphold previous interpretation of background tropospheric Δ17O(SO42−)nss in Antarctica [Alexander et al., 2002; Savarino et al., 2003a] as largely controlled by the transfer of isotopic signatures from the three dominant tropospheric SO2 oxidation pathways: gas phase oxidation by OH and in-cloud oxidation by O3 and H2O2. Other tropospheric formation pathways contribute minimally to sulfate deposited at this location (<4% total). We can thus interpret tropospheric Δ17O(SO42−)nss following a simple isotope mass balance,
equation image
where fOH, fH2O2, and fO3 represent the fractional contributions of sulfate production by each oxidant, and Δ17OOH, Δ17OH2O2, and Δ17OO3 represent the signatures of sulfate produced by each oxidant. We assume Δ17OOH, Δ17OH2O2, and Δ17OO3 values of 0‰, 0.9‰, and 8.8‰, respectively, following previous model studies [Alexander et al., 2005, 2009]. Using the normalized fractional contribution of the three dominant tropospheric oxidation pathways (fOH, fH2O2, and fO3 = 12%, 71%, and 17%) derived from the global model for WAIS Divide for the present day [Alexander et al., 2009], we calculate tropospheric Δ17O(SO42−)nss = 2.1‰ for the present-day (i.e., industrial) following equation (2). The estimated present-day tropospheric Δ17O(SO42−)nss is lower than ice core Δ17O(SO42−)nss measurements of the industrial period (3.0‰ and 2.5‰ for k = 0.25 and 0.07, respectively). The use of a low k value to reflect the potentially important influence of frost flowers on the WAIS sea-salt budget reduces but does not eliminate the difference between calculated tropospheric and ice core Δ17O(SO42−)nss in the industrial period. The influence of a background stratospheric sulfate contribution with high Δ17O(SO42−) may explain part or all of the difference between calculated tropospheric Δ17O(SO42−)nss and ice core Δ17O(SO42−)nss measurements of the industrial period. Uncertainties in the relative contributions and isotopic signatures of different sulfate formation pathways (i.e., fOH, fH2O2, fO3, Δ17OOH, Δ17OH2O2, and Δ17OO3) may also contribute to the difference between our calculated tropospheric Δ17O(SO42−)nss and ice core Δ17O(SO42−)nss measurements of the industrial period but impact our calculation of preindustrial to industrial Δ17O(SO42−)nss change minimally as detailed later in section 6.2.3.
[26] We estimate the preindustrial sulfate contribution by each oxidant (fx, x = OH, H2O2, O3) by scaling the baseline industrial (present day) fx extracted from the global model (fH2O2, fO3, fOH of 71%, 17%, 12%) by the ratio of preindustrial to industrial abundance for each oxidant and renormalizing all fx values:
equation image
where rx is the ratio of preindustrial (PI) to present day (PD) abundances of oxidants (rx = [X]PI/[X]PD). In this approach, we treat fOH, fH2O2, and fO3 as equivalent to the sulfate production rate by each oxidant normalized to the total sulfate production rate in the industrial period, thereby inherently assuming that the lifetimes of sulfate produced by different oxidants are similar, consistent with results from previous global model studies [Alexander et al., 2005, 2009]. We then make use of the fact that the sulfate production rate by each oxidant is linearly related to the oxidant abundance. We examine preindustrial to industrial oxidant concentration ratios (rx = [X]PI/[X]PD) of 0.68, 0.80, and 1.17 for rH2O2, rO3, rOH, which correspond to preindustrial to industrial oxidant changes ([X]PD − [X]PI)/[X]PI) of 45%, 25%, and −15% for H2O2, O3, and OH, respectively. These values are consistent with preliminary global model results for preindustrial to industrial changes in all three oxidants (i.e., including H2O2) in the lower to middle troposphere of the southern polar region [Sofen et al., 2010] and fall within the range of published results for model preindustrial to industrial changes in O3 and OH in this region [Wang and Jacob, 1998; Mickley et al., 1999; Shindell et al., 2003]. Using these preindustrial to industrial oxidant ratios and estimating preindustrial fx and Δ17O(SO42−)nss via equations (3) and (2) results in estimated preindustrial tropospheric Δ17O(SO42−)nss that differs negligibly (<0.01‰) from the calculated present-day value (2.1‰).

[27] The calculated tropospheric Δ17O(SO42−)nss does not differ between preindustrial and industrial, despite significant changes in tropospheric oxidant abundances, due to the counterbalancing influence of different oxidant changes on Δ17O(SO42−)nss. The contribution of H2O2 to sulfate production is lower in the preindustrial relative to the industrial, while contributions of O3 and OH to sulfate production are higher in the preindustrial relative to the industrial. The relative contribution of O3 to sulfate production in the preindustrial is slightly higher than in the industrial (18% versus 17%), despite lower preindustrial O3 abundances, due to the substantially lower contribution of H2O2 to sulfate production in the preindustrial relative to the industrial (63% versus 71%). The slightly higher sulfate production by O3 in the preindustrial would elevate preindustrial Δ17O(SO42−)nss relative to the industrial, but is offset by the combination of higher sulfate production by OH and lower production by H2O2 in the preindustrial, which would reduce preindustrial Δ17O(SO42−)nss relative to the industrial. Our model-based calculation of preindustrial to industrial tropospheric Δ17O(SO42−)nss changes thus indicates that the preindustrial to industrial model oxidant concentration changes we examined here (45%, 25%, and −15% for H2O2, O3, and OH, respectively) are independently consistent with the lack of change observed in the ice core Δ17O(SO42−)nss record. No temporal changes in other atmospheric conditions (e.g., stratospheric sulfate contribution or isotopic signature, tropospheric cloud water pH or cloud liquid water abundance) are necessary to reconcile these tropospheric model oxidant changes with the ice core Δ17O(SO42−)nss record.

[28] We can use the same calculation (equations (2) and (3)) to estimate H2O2 changes necessary for reported model preindustrial to industrial O3 and OH changes in the lower atmosphere (<600 mb) of the southern polar region [Wang and Jacob, 1998; Mickley et al., 1999; Shindell et al., 2003] to be consistent with deviations of tropospheric Δ17O(SO42−)nss within analytical uncertainty (±0.2‰) over this time period. We estimate that deviations of tropospheric Δ17O(SO42−)nss within analytical uncertainty are consistent with the minimum reported preindustrial to industrial O3 increase of 10% and OH decrease of 10% in the lower atmosphere (>600 mb) of the southern polar region [i.e., Shindell et al., 2003] only if H2O2 changes remain between −9% and +49%. The maximum reported preindustrial to industrial O3 increase of 50% and OH decrease of 20% in the southern polar region [Mickley et al., 1999] is consistent with deviations of tropospheric Δ17O(SO42−)nss within analytical uncertainty only if H2O2 increases between 51% and 170%. Tropospheric Δ17O(SO42−)nss may have changed by greater than ±0.2‰ from preindustrial to industrial and still be consistent with the ice core Δ17O(SO42−)nss record, if there is a significant background stratospheric sulfate source at WAIS Divide that does not vary temporally. A significant background stratospheric source would reduce the sensitivity of ice core Δ17O(SO42−)nss to changes in tropospheric Δ17O(SO42−)nss. The ranges of preindustrial to industrial tropospheric H2O2 changes calculated here are therefore conservative estimates for reconciling tropospheric model O3 and OH changes with deviations of ice core Δ17O(SO42−)nss within analytical uncertainty (<±0.2‰).

6.2.3. Uncertainties in Calculation of Δ17O(SO42−)nss Change

[29] In section 6.2.2, we focus on the calculated preindustrial to industrial change in tropospheric Δ17O(SO42−)nss, rather than the magnitude of Δ17O(SO42−)nss at each time, because the change in Δ17O(SO42−)nss is less sensitive to uncertainties in the boundary conditions of our calculation (i.e., fOH, fH2O2, fO3, Δ17OOH, Δ17OH2O2, and Δ17OO3 in equation (2)). Uncertainties in these boundary conditions may partially contribute to the difference between calculated tropospheric Δ17O(SO42−)nss and ice core Δ17O(SO42−)nss measurements of the industrial period. Uncertainties of ±4% in the model-derived relative contributions of sulfate formation by O3, OH, and/or H2O2 (i.e., fOH, fH2O2, and fO3) can explain the difference (0.4‰) between calculated tropospheric Δ17O(SO42−)nss (2.1‰) and ice core Δ17O(SO42−)nss of the industrial period (2.5‰, k = 0.07). Increasing the Δ17O transferred to sulfate by O3 and/or OH in our calculation (i.e., Δ17OOH = 0‰ and/or Δ17OO3 = 8.8‰) by 1‰–3‰ can also fully reconcile the difference (0.4‰–0.9‰ for k = 0.07–0.25, respectively) between calculated tropospheric Δ17O(SO42−)nss and ice core Δ17O(SO42−)nss and is consistent with previous work suggesting these values are underestimates (see section 2). Neither of these changes to our boundary conditions results in a significant (>±0.2‰) calculated preindustrial to industrial tropospheric Δ17O(SO42−)nss change as described in section 6.2.2, nor do they alter the sign of the bounds we estimate in order for preindustrial to industrial H2O2 changes associated with reported model O3 and OH changes to be consistent with tropospheric Δ17O(SO42−)nss changes of less than ±0.2‰. Estimated bounds for preindustrial to industrial H2O2 changes associated with reported model O3 and OH changes are altered by a maximum of 8% by uncertainties of ±4% in the relative contributions of sulfate formation by O3, OH, and/or H2O2. The calculation of bounds for preindustrial to industrial H2O2 changes associated with reported model O3 and OH changes should be revisited if new constraints become available concerning polar Δ17O of sulfate produced by O3 and or OH.

7. Conclusions

[30] We have presented the first ice core record of the multiple isotope composition of sulfate from WAIS Divide, Antarctica spanning the preindustrial to industrial time period (∼1775–2005). Sulfate concentrations show no trend over the WAIS Divide ice core record, although a strong spike is apparent in the early 1800s sample due to two stratospheric-scale volcanic eruptions during this time (1815 Tambora eruption and 1810 eruption of unknown origin). Mean δ34Snss values from the WAIS Divide ice core are significantly lower than both inland and coastal East Antarctic ice core measurements of late Holocene age, consistent with other West Antarctic observations [Pruett et al., 2004]. Because the observed pattern of spatial variability in Antarctic δ34Snss does not follow a spatial pattern similar to that expected for either variations in sulfate transport pathways or oxidation chemistry, we rule these out as dominant causes of the relatively low δ34Snss at WAIS Divide, following previous authors [Jonsell et al., 2005]. The sulfate contribution of terrigenous continental material at WAIS Divide is estimated to be low (<2%) and cannot account for the low ice core δ34Snss. On the basis of current available information on sulfate δ34S source signatures, we suggest the low δ34Snss in West Antarctica reflects a combination of stronger influence of volcanogenic and/or stratospheric sulfate with low δ34S in West relative to East Antarctica and the influence of frost flowers on the sea-salt sulfate-to-sodium ratio (i.e., k). Because of the importance of frost flowers on k and ultimately the calculation of δ34Snss, potential fractionation in δ34Sss during frost flowers should be investigated. Δ33S remains within analytical uncertainty of zero throughout the WAIS Divide ice core record, consistent with previous East Antarctic ice core Δ33S measurements [Alexander et al., 2003]. However, the lack of significant nonzero Δ33S throughout the WAIS ice core records does not contradict a contribution of stratospheric sources for background sulfate deposited at WAIS Divide, as suggested by our ice core δ34Snss analysis.

[31] Ice core Δ17O(SO42−)nss at WAIS Divide shows no significant changes (± 0.2‰) throughout the late 1800s to the early 2000s. This contrasts with the ice core Δ17O(SO42−)nss record from Greenland, which shows a strong perturbation during this time (2‰) [Alexander et al., 2004], consistent with a stronger influence of anthropogenic activity on northern hemisphere oxidant abundances. Data extracted from a previously published global chemical transport model study [Alexander et al., 2009] indicates that tropospheric Δ17O(SO42−)nss at WAIS Divide can be interpreted following a simple isotope mass balance of three primary tropospheric sulfate production pathways: SO2 oxidation by OH, H2O2, and O3. Using this relationship, we show that the lack of significant change in ice core Δ17O(SO42−)nss between the mid-1800s and early 2000s is consistent with preindustrial to industrial changes in tropospheric O3 and OH within the range of published model results for the southern polar region [Wang and Jacob, 1998; Mickley et al., 1999; Shindell et al., 2003] along with an increase in southern polar H2O2 concentrations that is supported by preliminary global model results [Sofen et al., 2010]. We also estimate that maximum atmospheric chemistry model estimates of preindustrial-industrial O3 concentration increases (50%) and OH decreases (20%) in the southern polar region [Mickley et al., 1999] are consistent with changes in tropospheric Δ17O(SO42−)nss within analytical uncertainty only if H2O2 also increases by over 50%. An increase in atmospheric H2O2 in the southern polar region during the past century is qualitatively supported by ice core H2O2 concentration increases in West Antarctica, but additional work is needed to quantify atmospheric H2O2 concentration changes from these records [Frey et al., 2006]. In the future, the consistency of preindustrial to industrial Antarctic oxidant concentration changes estimated from atmospheric modeling and ice core H2O2 records can be assessed using ice core Δ17O(SO42−)nss records, following the simple isotope mass balance relationship described here.

Acknowledgments

[32] We acknowledge financial support from NSF awards OPP-0538049, 0538520, and 0538427. We thank Michael Town, Joël Savarino, Samuel Morin, and Jessica Lundin for assistance interpreting the isotopic records, Peter Neff for assistance with ice core processing at the University of Washington, and Ross Edwards, Ryan Banta, and Tommy Cox for laboratory assistance at the Desert Research Institute in Reno, Nevada. We appreciate the support of the WAIS Divide Science Coordination Office at the Desert Research Institute for the collection and distribution of the WAIS Divide ice core and related tasks (Kendrick Taylor, NSF grants 0440817 and 0230396). We also acknowledge additional support provided by the NSF Office of Polar Programs, including: the Ice Drilling Program Office and Ice Drilling Design and Operations group for coring activities; Raytheon Polar Services for logistics support in Antarctica; the 109th New York Air National Guard for airlift in Antarctica; and the National Ice Core Laboratory for archiving the core and organizing core processing.