Evidence of Recent Active Volcanism in the Balleny Islands (Antarctica) From Ice Core Records

Volcanism can play a key role in modulating climate; however, a lack of historical records has limited our comprehension of Antarctic volcanism and its role on the cryosphere. Remote sensing can provide insight into active volcanism in Antarctica during the satellite era, although the evidence is often inconclusive. Here, we use independent evidence from ice cores to validate one such potential volcanic eruption from the sub‐Antarctic Balleny Islands in 2001 CE. Multiple ice cores from downwind of the eruption site, record elevated input of sulfate, microparticles, and the presence of tephra, coincident with the eruption. In‐phase deposition of volcanic products confirmed a rapid tropospheric transport of volcanic emissions from a small‐to‐moderate, local eruption during 2001. Air mass trajectories demonstrated some air parcels were transported over the West Antarctic ice sheet from the Balleny Islands to ice core sites at the time of the potential eruption, establishing a route for transport and deposition of volcanic products over the ice sheet. The data presented here validate previous remote sensing observations and confirms a volcanic event in the Balleny Islands during 2001 CE. This newly identified eruption provides a case study of recent Antarctic volcanism.


Introduction
Antarctica is considered one of the least volcanically active regions on Earth, with the highest number of volcanoes listed as uncertainly active and many others hidden beneath the ice sheet (Hund, 2014). Over 100 volcanoes have been identified in the Antarctic continent and sub-Antarctic Islands (de Vries et al., 2018;LeMasurier et al., 1990) with more than 20 documented in the historical records (Patrick & Smellie, 2013). Among the historically active, just two are frequently monitored by ground-based instruments (Mount Erebus and Deception Island;LeMasurier et al., 1990;Patrick & Smellie, 2013;Smellie et al., 2021), while the others are rarely surveyed due to their extreme isolation. Remote sensing techniques have helped to monitor regional volcanism; however, high detection thresholds and coarse spatial resolution have hindered the capacity of some sensors to accurately identify volcanic activity (Patrick & Smellie, 2013). Moreover, the effects of volcanism can be rapidly obscured in the Antarctic and sub-Antarctic environment due to frequent snowfall and cloud coverage (LeMasurier et al., 1990). Even though Antarctic volcanoes do not present significant direct hazards, their study is important for many areas of research. Mainly, geothermal heat flux estimates (Vogel et al., 2006), ice flow dynamics models (Bingham & Siegert, 2009), and the volcanic effects on the polar climate (Cole-Dai, 2010;Robock, 2000;Sigl et al., 2014). Altogether, the remoteness and inaccessibility of most of the Antarctic volcanoes have strongly limited our knowledge of the Antarctic volcanic activity.
An alternative way to study volcanic activity in Antarctica is the analyses of volcanic tephra (assortment of fragments, from blocks of material to ash, ejected into the air during a volcanic eruption; Kittleman, 1979) preserved in ice core layers. Volcanic eruptions may also emit large amounts of particulate matter and sulfur compounds into the atmosphere. Sulfur compounds are oxidized to sulfuric acid (H 2 SO 4 2 ) and travel as particulate aerosols in the atmosphere. Volcanic sulfate aerosols and tephra in the atmosphere can be transported thousands of kilometers from the volcanic source, to be deposited and preserved on polar ice sheets (Koffman et al., 2017). Measurements of sulfate concentrations and electric conductivity (EC) in the ice strata help to detect and quantify past volcanic activity over thousands of years (Cole-Dai, 2010;Cole-Dai et al., 1997). Similarly, the physical and chemical characterization of tephra and cryptotephra (micrometer-sized tephra) embedded in ice layers can record past volcanic activity and fingerprint the source of the volcanic eruptions (Dunbar & Kurbatov, 2011;Narcisi et al., 2019). These methods have been applied to several Antarctic ice cores, providing evidence of past volcanic activity at regional, hemispheric, and global scales (Basile et al., 2001;Fujita et al., 2015;Jiang et  Abstract Volcanism can play a key role in modulating climate; however, a lack of historical records has limited our comprehension of Antarctic volcanism and its role on the cryosphere. Remote sensing can provide insight into active volcanism in Antarctica during the satellite era, although the evidence is often inconclusive. Here, we use independent evidence from ice cores to validate one such potential volcanic eruption from the sub-Antarctic Balleny Islands in 2001 CE. Multiple ice cores from downwind of the eruption site, record elevated input of sulfate, microparticles, and the presence of tephra, coincident with the eruption. In-phase deposition of volcanic products confirmed a rapid tropospheric transport of volcanic emissions from a small-to-moderate, local eruption during 2001. Air mass trajectories demonstrated some air parcels were transported over the West Antarctic ice sheet from the Balleny Islands to ice core sites at the time of the potential eruption, establishing a route for transport and deposition of volcanic products over the ice sheet. The data presented here validate previous remote sensing observations and confirms a volcanic event in the Balleny Islands during 2001 CE. This newly identified eruption provides a case study of recent Antarctic volcanism.
Here, we aim to use ice cores to independently verify the Sturge Island eruption in 2001. We present a detailed study of five ice core glaciochemical and microparticle records from the southern Antarctic Peninsula, Ellsworth Land, and Marie Byrd Land, to validate the occurrence of this eruption. We include the analysis of insoluble particulate material and cryptotephra in the ice cores as evidence for volcanic activity. Forward air mass trajectories are used to track the air masses originating from the Balleny Islands at the time of the potential 2001 eruption. The aim of this study is to provide independent ice core evidence for a 2001 volcanic eruption on Sturge Island. This study is focused on validating a recent volcanic event in the Balleny Islands and does not intend to present a standardized method to detect the occurrence of small volcanic eruptions in Antarctic ice core records.

Study Sites
Five ice cores from the southern Antarctic Peninsula, Ellsworth Land, and Marie Byrd Land were included in this study ( Figure 1; Table 1). The cores were selected due to their location downwind from the Balleny Islands, their retrieval after the 2001 austral winter, their temporal resolution (>10 samples in the youngest year), their regional distribution, and their data availability. Among the five ice cores used in this study, four (GOM, BC, 01-4, WAIS) have been previously published (Table 1). The 140 m Jurassic ice core (JUR) was drilled by the British Antarctic Survey on the English Coast, Southern Antarctic Peninsula during the austral summer 2012/2013. Ice core samples were cut using a bandsaw with a steel blade and then melted using a Continuous Flow Analysis (CFA) system (Rothlisberger et al., 2000) in the ice chemistry lab at the British Antarctic Survey, UK. An ice core chronology was established based on the hydrogen peroxide annual cycle that is assumed to peak during the summer solstice. The top 53.5 m included in this work date back to 1977 CE, with an estimated dating error for the 1977-2013 interval of ±3 months for each year and with no accumulated error. Discrete ice core samples were cut at 5 cm resolution for major ion analysis (including Methanesulphonic Acid (MSA), sodium, and sulfate) with ion chromatography, using a reagent-free Dionex ICS-2500 anion and IC 2000 cation system.
Four ice cores (GOM, JUR, BC, and WAIS) were evaluated over a 30 yr overlapping period (1977( -2007 for evidence of a volcanic eruption in the 2001 ice layer (hereafter referred to as 2001L) The ITASE 01-4 ice core, drilled in 2002, was evaluated over a 25 yr overlapping period (1977. Additionally, two previously identified ice core horizons were targeted as examples of well-dated volcanic events recorded in ice core layers. The 1994The -1992 Hoffmann et al., 2020;Jiang et al., 2012;Osipov et al., 2014;Plummer et al., 2012;Schwanck et al., 2017;Thoen et al., 2018;Zhang et al., 2002) (Jiang et al., 2012;Plummer et al., 2012;Thoen et al., 2018;Traufetter et al., 2004). Both eruptions are observed in younger ice core horizons because of the lagged deposition of volcanic sulfates over the Antarctic ice sheet after large low-latitude eruptions (Cole-Dai et al., 1997). All ages presented in this work are based on the ice chronologies reported for each core (Table1).

Sulfate Concentration Analyses
Sulfate from volcanic eruptions is superimposed over the background sulfate. This background signal includes organic sulfur compounds, such as dimethyl sulfide (DMS), from marine biogenic emissions (Castellano et al., 2004;Cole-Dai et al., 2000;Dixon et al., 2004;Maupetit & Delmas, 1992;Nardin et al., 2020), with a smaller contribution from sea salt aerosols. Even though the background sulfate is temporally variable, it can be assumed as relatively constant in the last centuries (Castellano et al., 2004;Kreutz et al., 1999Kreutz et al., , 2000Traversi et al., 2002). While volcanic sulfate can clearly be identified in continental Antarctic ice cores, the high biogenic background at low elevation coastal sites makes detection more difficult. In particular, the accurate detection of small and moderate volcanic events depends on how the background sulfate is quantified and what volcanic detection threshold is set (Budner & Cole-Dai, 2003;Castellano et al., 2004;Cole-Dai et al., 1997). Several methods have been proposed (Castellano et al., 2004;Cole-Dai et al., 1997;Gautier et al., 2016;Traufetter et al., 2004) based on the evaluation of a background sulfate representative value (m) and its standard deviation (σ) to establish a threshold (m + 2σ). Sulfate peaks above this threshold are considered indicative of volcanic activity.
In this work, the background signal is evaluated in the total sulfate concentration (SO 4 2− ) and the non-sea salt sulfate flux (nssSO 4 2− -flux; Cole-Dai et al., 1997). The nssSO 4 2− -flux was calculated using Equations 1 and 2 (Wagenbach et al., 1998), using sodium (Na + ) as the reference ion (Castellano et al., 2004;Dixon et al., 2004;Jiang et al., 2012;Li et al., 2012;Osipov et al., 2014;Ren et al., 2010). The analysis of the nssSO 4 2− -flux was incorporated because it facilitates the detection of small and moderate volcanic signals (Cole-Dai et al., 1997;Zhang et al., 2002). However, we note that the occurrence of extreme precipitation events at coastal sites (Turner et al., 2019) may reduce the reliability of the flux calculations. In the absence of sulfate data from the GOM ice core, the total sulfur (S tot ) and non-sea salt sulfur flux (nssS-flux) were used for calculations: To detect potential volcanic eruptions from elevated SO 4 2− and nssSO 4 2− -flux, we applied the method originally proposed by Castellano et al. (2004). This method was selected over other methods (Cole-Dai et al., 1997;Traufetter et al., 2004) because it considers the log-normal distribution of the sulfate data. The use of log-normal statistics in sulfate analyses has been proven to clearly differentiate between volcanic sulfate and background sulfate (Castellano et al., 2004). To calculate the background sulfate, new data sets were generated after excluding individual ice core horizons from well-known volcanic eruptions between 2007and 1977CE (e.g., Pinatubo/ Hudson (1991CE) and El Chichón (1982). After excluding these horizons, the background and its variability were estimated at each sample point by calculating in the log domain, the mean and standard deviation of a 20% weighted curve fit centered on each sample point (10% weighted curve fit for the shorter 01-4 ice core). The applied curve fit roughly corresponds to a 6-year running mean (approximately 3-year span for 01-4 ice core), ensuring sulfate annual cycle is effectively removed from the background sulfate signal. To identify samples with a potential volcanic influence, the SO 4 2− and nssSO 4 2− -flux had to exceed the background signal (m) by two times the standard deviation (>2σ-peak). This threshold ensured 95.5% of the random background variability was excluded. SO 4 2− and nssSO 4 2− -flux > 2σ-peaks were classified based on the number of data points exceeding the threshold (single point (=1) or multiple points (>1)). The method applied in this study assumes that in the  absence of inputs from large volcanic and anthropogenic sources (negligible in Antarctica), the sulfate concentration in the snow comprises inputs from regional background sulfate emissions, not controlled by a dominant source region or transport and deposition processes (Cole-Dai et al., 1997). S tot and nssS-flux from GOM were analyzed using the same method applied for SO 4 2− and nssSO 4 2− -flux detection analyses, respectively.
As previously stated, one of the main sources of background sulfate is DMS from marine biogenic emissions. The oxidation of DMS in the atmosphere produces MSA, a chemical compound widely studied in ice core records because of its link to marine biogenic emissions in the Southern Ocean (Abram et al., 2010(Abram et al., , 2013Criscitiello et al., 2013;Curran et al., 2003;Thomas et al., 2016). The MSA records available for GOM, JUR, and BC ice cores are included to assess whether the SO 4 2− >2σ-peaks identified in the 2001L were influenced by increased marine biogenic emissions. MSA records for WAIS and 01-4 ice cores were not available.
Volcanic sulfate fluxes (VSF) were calculated for two of the targeted periods (2001L and 1992(2001L and -1994.  , 1997). Due to the different parameters measured in GOM (S tot and nssS-flux), this ice core was excluded from the VSF calculations.

Physical Properties Analyses
The EC signal recorded in ice is controlled by soluble impurities that originate mostly from sea salt, biomass burning, and volcanic eruptions, and is strongly correlated with acidity (Mulvaney, 2013). EC measurements from four ice cores were included in this study (only available at GOM, JUR, BC, and WAIS) as an additional data set to test the presence of volcanic products. EC was analyzed in different labs using an Amber Science flow-through meter connected to a continuous ice core melter system. EC data presented a log-normal distribution. Therefore, data treatment and calculations were performed using log-normal statistics. The background conductivity, its variability and the establishment of an excess conductivity detection threshold were calculated following the same method presented in Section 2.2 for SO 4 2− and nssSO 4 2− -flux.

Forward Trajectory Analyses
Forward trajectory analysis is used to examine the pathways of air masses passing over the Balleny Islands and their potential transit over the ice core sites during the deposition of the 2001L. The National Oceanic and Atmospheric Administration's Hybrid Single-Particle Lagrangian Integrated Trajectory model (Draxler & Hess, 1998;Stein et al., 2015) was used to calculate three-dimensional air parcel pathways under isobaric conditions (2.5° latitude-longitude resolution). Trajectories were calculated starting from the Balleny Islands for a 30 hr interval, 15 hr before, and 15 hr after the first evidence of possible volcanic activity (13:52 UTC on June 12, 2001). Forward trajectories were initiated every hour for up to 10 days using the NCEP/NCAR Reanalysis archives (1948present) with three starting elevations: 1,000, 1,500, and 2,000 m above the sea level (a.s.l.). These elevations were selected because of their proximity to the summit of Sturge Island (1,167 m a.s.l), from where a potential volcanic plume could have been emitted. For comparison, trajectories were classified into three groups based on their starting time relative to the first image (Im) evidence of the unusual cloud formation over Sturge Island: pre-Im evidence, first-Im evidence, and post-Im evidence.

Microparticle Analyses
Microparticle concentration (MPC) and Particle size distributions (PSD) were measured in the JUR and WAIS ice cores. The latter corresponds to the WAIS Divide deep ice core, WDC06A (Kreutz et al., 2011(Kreutz et al., , 2015. MPC from the WDC06A was measured using a flow-through Klotz Abakus laser particle counter connected to a continuous ice core melter system at the University of Maine (Breton et al., 2012). Particles were measured in 31 size channels, spanning 1-15 μm diameter. Similarly, MPC from the JUR ice core was measured using a flow-through Klotz Abakus laser particle counter connected to a continuous ice core melter system at the British Antarctic Survey. Particles were measured in 23 size channels spanning 0.9-12 μm diameter. MPC data sets presented a log-normal distribution. Therefore, data treatment and calculations were performed using log-normal statistics. To assess whether a MPC peak in the data set could be influenced by volcanic activity, the microparticle background concentration and its variability were calculated. A detection threshold was set following the same guidelines presented in Section 2.2 for sulfate analyses. The JUR dust record is presented with a 4 m gap, due to problems in the data acquisition between 43.6 and 47.6 m deep.
PSD were obtained by calculating the ratio of total volume of insoluble dust contained within each size bin and the derivative of the volume with respect to the natural logarithm of the particle diameter for each bin (dV/dlnD), as presented in Koffman et al. (2014). Three individual ice core horizons were targeted to characterize their PSD: The 2001 ice core horizon (2001L), Pinatubo/Hudson eruption (1991 CE), and El Chichón eruption (1982 CE).
To determine if the targeted horizons differ from the background particle size of atmospheric dust, the PSD for the 1977-2007 CE period was calculated as the average PSD after removing the three targeted horizons. For PSD analyses, the mode particle diameter was used as a representative statistic of the volume distribution (Koffman et al., 2014;Ruth et al., 2003).
Microscopy analyses of microparticles were included to determine whether any of the particles present in the 2001 CE horizon were cryptotephra shards. For this, ice samples of 200 mL, from the 2001L of the JUR and GOM ice cores, were melted and filtered. Samples were melted using a CFA system (Rothlisberger et al., 2000) in the ice chemistry lab at the British Antarctic Survey, UK. Meltwater from the CFA waste lines was collected in new bottles then filtered through 13 mm diameter, 1.0 μm pore size Whatman™ Polycarbonate membrane filters, inside clean polypropylene Swinnnex™ filter holders. Filters were mounted onto aluminum stubs for analyses on a scanning electron microscope (SEM) at the Earth Sciences Department of the University of Cambridge. Filters were imaged on a Quanta-650F using back scattered electrons on a low-pressure mode. Each filter was imaged at x800 magnification for cryptotephra identification and physical characterization, following the analysis strategy presented in Tetzner et al. (2021). Two additional samples from the 2001L of the JUR ice core were melted in a class-100 clean room, then centrifuged (6 min at 1,200/1,600 rpm) and decanted successively until samples were concentrated in 2-5 mL fluid. The 2-5 mL sample liquid was homogenized, pipetted onto a single coverslip (22 × 40 mm), dried in an isolated drying cupboard, and then mounted onto a single microscope slide using Norland optical adhesive 61 (refractive index 1.56). Each microscope slide was scanned for the presence of cryptotephra shards.

Results
We structure our evidence in the following order, based on their merit as an indicator for this volcanic eruption. (a) Sulfate, a commonly used proxy for past volcanism in continental Antarctic ice cores but less suitable at low elevation and coastal sites where background biogenic sulfur is high. (b) EC, a secondary parameter commonly used to identify volcanic activity. (c) Microparticle concentration (MPC) and size distribution dependent on distance from the source. (d) Cryptotephra, only produced during volcanic eruptions. Each indicator is supported by air-mass trajectory analysis, demonstrating the feasibility that their source corresponds to the Sturge Island eruption in 2001.

Methanesulphonic Acid (MSA) Measurements
MSA profiles for JUR and BC from 2001L presented prominent peaks of >35 ppb indicative of spring/summer periods and a minimum <5 ppb, denoting the autumn/winter interval. The GOM MSA profile exhibits peaks of >15 ppb indicative of spring/summer and slightly smaller, isolated MSA events that correlate with autumn/winter timing for example, at 11.71 and 12.20 m.

Particle Size Distribution (PSD)
The volume concentration parameter (dV/dlnD) of insoluble dust was calculated in WAIS and in JUR for several particle diameters to obtain the particle size distribution (PSD) (Figures 5c and 5d). In the WAIS core, the background PSD (1977( -2007 presented a steady increase in concentration with increasing particle diameter (1.3-5.7 μm), followed by a constant decrease for the coarser particles (5.7-15 μm   ( , 1992( -1994( , 1982( -1984  observed in the three target horizons, except for a notable increase in the coarsest particles (10-12 μm). Despite the similarities, the PSD for Pinatubo/ Hudson (1991 CE) and 2001L exhibited considerably higher volume concentration values, up to four-times higher than the background, with PSDs that were completely detached from the background PSD. Likewise, the PSD for El Chichón exhibited higher than background volume concentration values in the finer particles (3.0-5.1 μm). Additional discrepancies were observed in the mode particle diameter. Whilst similar mode diameters were obtained in the background (5.7 μm), Pinatubo/Hudson (5.7 μm) and El Chichón (5.1 μm), the 2001L presented a considerably higher mode particle diameter (10 μm) (Figure 5c).
In the JUR core, the background PSD (1977( -2007 presented a steady increase in the volume concentration with increasing particle diameter (1.0-6.0 μm), followed by a slight decrease for the coarser particles (8-12 μm). The three targeted horizons exhibited slightly different distributions. The PSD for El Chichón exhibited a similar distribution and volume concentration to the background. However, the El Chichón PSD presented a comparatively higher volume concentration in the finer particles (3.0-6.0 μm), not identified in the background PSD. The PSD for Pinatubo/Hudson and 2001L present similar distributions with volume concentrations up to an order of magnitude higher than the background. However, the PSD for Pinatubo/Hudson displays a sharp decrease after reaching its highest particle diameter (D) value at D = 6 μm, while the PSD for 2001L reaches a steady volume concentration value above D = 8 μm. The mode particle diameters for Pinatubo/Hudson (1991 CE) and El Chichón (1982 CE) matched the mode particle diameter of the background (6 μm). Unlike the other targeted horizons, the 2001L presented a higher mode particle diameter (8 μm).

Microscopy Analyses
Two filters containing insoluble particulate material from the 2001L of JUR were analyzed for microparticle characterization (Figure 5e). Seventy particles were identified as cryptotephra shards with a mean size of 19.3 ± 8.7 μm (sizes ranging from 7 to 47 μm). Similarly, two filters containing insoluble particulate matter from the 2001L of GOM were analyzed. Ten particles were identified as cryptotephra shards (Figure 5f) with a mean size of 12.9 ± 3.9 μm (sizes ranging from 7 to 20 μm). Most of the cryptotephra shards identified in both sites (JUR and GOM) were characterized by angular morphologies and concave features (vesicles) without evidence of alteration or corrosion. The cryptotephra shards are cuspate, platy with sharp edges and few with open vesicles and some with butterfly shape. Microscope slides from the 2001L of JUR contained 10 cryptotephra shards with a mean size of 21 μm and exhibiting platy and cuspate textures with round vesicles (Figure 5g).

Air Mass Trajectories
Forward trajectory analyses showed most of the air masses passing over the Balleny Islands on the June 12, 2011 remained within the Southern Ocean for several days, mainly over the Somov, Ross, Amundsen, and Bellingshausen Seas (Figure 6). Trajectories show air masses were predominantly traveling over the June sea-ice zone, with short periods (<72 h) of transit either over the Antarctic coast (<1,000 m a.s.l) or over the ice sheet (>1,000 m a.s.l.).
After leaving the Balleny Islands, all air masses moving at 1,000 m a.s.l. (Figure 6a) transited over the Oates Coast, some of them reaching Saunders Coast in Marie Byrd Land (See Figure 1 for geographical references). A similar pattern is present in air masses moving at 1,500 m a.s.l. (Figure 6b), where most trajectories transit over the same regions, but some also reach the Walgreen Coast, next to the Amundsen Sea Embayment. Air masses over 2,000 m a.s.l. (Figure 6c) depict a different path with most trajectories moving north toward lower latitudes and only some of them traveling back south, over the ice sheet in Marie Byrd Land, Ellsworth Land, and the southern Antarctic Peninsula. These trajectories passing over Ellsworth Land and the southern Antarctic Peninsula describe the only air masses reaching four of the ice core sites (GOM, JUR, BC, 01-4). These five trajectories represent the paths followed by air masses passing over the Balleny Islands at: 12:00, 13:00, 14:00, 23:00 UTC on the June 12, 2011 and at 02:00 UTC on the June 13, 2001. After leaving the Balleny Islands, four of these trajectories (12:00, 13:00, 14:00, and 23:00 UTC) were transported E-NE and then North, leaving the Southern Ocean and reaching ∼43°S. Then they were transported back over the Southern Ocean southwards to the continent. The   02:00 UTC trajectory remained within the June sea-ice zone, passing over the ice sheet in Ellsworth Land and into the southern Antarctic Peninsula. An additional group of five trajectories was identified passing near WAIS and 01-4 ice core sites at 1,500 m a.s.l. These trajectories were passing over the Balleny Islands between 08:00 and 13:00 UTC on the June 12, 2001. After leaving the Balleny Islands they were transported over the June seaice zone of the Amundsen Sea before reaching the Walgreen Coast in the Amundsen Sea embayment (Figure 6b).

Pinatubo/Hudson (1991 CE) and El Chichón (1982 CE) Eruptions
The evaluation of previously identified volcanic eruptions, especially Pinatubo/Hudson (1991 CE) andEl Chichón (1982 CE), is intended to verify the robustness of volcanic identification in the ice core sites used in this study. The 1994-1992CE and 1984-1982 ice layers are distinctive periods in the ice core record. These intervals show strong similarities in the presence and consistency of prominent summer peaks above the background variability in total sulfate concentration (SO 4 2− ), EC, and MPC. The levels of SO 4 2− and MPC in the two intervals are above the detection threshold, suggesting an additional input of SO 4 2− and microparticles during these periods. In Antarctica, the 1994-1992CE and 1984-1982 CE ice core layers have been linked to the well-documented eruptions of Mount Pinatubo/Cerro Hudson (1991 CE;Cole-Dai & Mosley-Thompson, 1999;Hoffmann et al., 2020;Jiang et al., 2012;Osipov et al., 2014;Plummer et al., 2012;Schwanck et al., 2017;Thoen et al., 2018;Zhang et al., 2002) andEl Chichón (1982 CE;Inoue et al., 2017;Jiang et al., 2012;Kohno et al., 1999;Plummer et al., 2012;Thoen et al., 2018;Traufetter et al., 2004), the two volcanic eruptions with the greatest SO 2 emissions worldwide of the 1977-2007 CE period (Shinohara, 2008). Therefore, we propose the excess of SO 4 2− and MPC identified during these periods (1994-1992CE & 1984-1982 were derived from the large low-latitude (mid-latitude) Pinatubo (Cerro Hudson) and El Chichón eruptions.
Subtle differences were identified in the signals of SO 4 2− , nss SO 4 2− , EC and MPC between the 1994-1992 CE and 1984-1982 CE ice core layers. The principal difference was associated with the magnitude of the SO 4 2− , nss SO 4 2− , EC and MPC peak(s) in each layer. Peaks in the 1994-1992 CE ice core layer were higher and more persistent, while peaks from the 1984-1982 CE ice layer were smaller or absent (nssSO 4 2− -flux). These discrepancies are explained by differences in the volumetric emissions from each eruption, classified on the logarithmic scale of the Volcanic Explosivity Index (VEI). As VEI-6 (VEI-5), the Pinatubo (Cerro Hudson) eruption released at least 10 km 3 (1 km 3 ) of particles and gases to the atmosphere, whilst the El Chichón VEI-5 eruption, ejected only 1 km 3 . The difference in the scale of these two eruptions accounts for the different signal strengths observed in the 1994-1992 and 1984-1982 CE ice layers. Our results are consistent with observations from several Antarctic ice cores presenting a stronger signal for Pinatubo/Hudson and a weaker signal for El Chichón (Inoue et al., 2017;Plummer et al., 2012;Thoen et al., 2018).
The records identified for the Pinatubo/Hudson and El Chichón eruptions provide two examples of how different parameters measured in ice cores from the southern Antarctic Peninsula, Ellsworth Land and Marie Byrd Land can effectively record recent major low-latitude volcanic eruptions. The findings also highlight the limitations of using SO 4 2− , nss SO 4 2− in the ice core sites downwind of the Balleny islands. The high background concentrations, from adjacent oceanic sources, and influence of extreme precipitation events (Turner et al., 2019) make it difficult to identify even large volcanic eruptions clearly or consistently.

The 2001 CE Ice Core Horizon
In 2001, remote sensing observations from Sturge Island presented inconclusive evidence for recent volcanic activity in the Balleny Islands. Numerous lines of evidence from the ice core record provide independent evidence that a high-latitude volcanic eruption occurred at this time.
The 2001 ice core layer (2001L) presents some of the most striking features identified in the 1977-2007 CE period. This layer exhibited a prominent and synchronous SO 4 2--EC > 2σ-peak on each ice core analyzed. Its presence in ice across Marie Byrd Land, Ellsworth Land, and the southern Antarctic Peninsula, highlights it as a persistent regional feature. Even though the >2σ signal is only represented by a single data point in two ice core records (SO 4 2− in JUR and WAIS), its consistent presence across the region rules out the possibility of analytical outliers or sample contamination. The magnitude and regional distribution of the SO 4 2− -EC > 2σ-peaks suggest it was caused by an exceptional input of sulfates to an extended area of the ice sheet during mid-2001. Further analyses of the MSA record showed the absence of peaks during the austral autumn/winter of 2001. Thus, demonstrating the SO 4 2− -EC > 2σ-peak was not produced by increased austral autumn/winter marine biogenic productivity (Mulvaney et al., 1992). The lack of evidence for a biogenic source suggests the SO 4 2− >2σ-peak identified in the 2001 ice core layer could have been caused either by inputs from a volcanic source or by large inputs of sea salts. However, the detection of numerous cryptotephra glass shards in the mid-2001 CE ice core layer from the JUR and GOM ice cores provides direct evidence for a volcanic eruption. Combined with the observed MPC>2σ-peak in WAIS and JUR for the 2001L, this supports the assertion that the excess SO 4 2− is derived from a volcanic source. The elevated inputs of SO 4 2− & MPC recorded in 2001L are comparable to those observed in the 1994-1992CE and 1984-1982 ice core layers, attributed to Pinatubo/Hudson and El Chichón eruptions, respectively (Section 4.1).
Results from PSD analyses spatially constrain a possible source for the 2001 volcanic products. In particular, the PSD profile for the 2001 MPC > 2σ-peak presented a considerable increase in the volume and size of particles (mode diameter = 10 μm), compared with the background PSD (mode diameter = 5.7 μm). The coarser-than background PSD in 2001L suggests a more proximal volcanic source (Koffman et al., 2013) because coarser particles are unlikely to be transported large distances. Additionally, the spatial gradient identified in the VSF suggests the transport and deposition of volcanic sulfate was from the Amundsen Sea sector toward the Bellingshausen Sea sector. Thus, establishing an eastward dispersion of the volcanic cloud. Results from the PSD and the net VSF are complemented by the synchronous deposition of the SO 4 2− , EC, and MPC peaks that indicate rapid tropospheric transport and therefore also support a closer volcanic source (Koffman et al., 2017). Similarly, the angular texture of the cryptotephras suggests these glass shards were produced, transported, and deposited within a short interval, without being altered by reworking or weathering processes.
To date, the Global Volcanic Program list of volcanic emissions only records two major volcanic events (VEI ≥ 4) for the 1998-2001 CE period. These events correspond to Shiveluch volcano in Russia (ongoing eruption since 1999 CE, VEI = 4) and to Ulawun volcano in Papua New Guinea (September 2000 CE, VEI = 4). Additionally, during the same period, there is only one confirmed major source of SO 2 volcanic emissions, the Nyamuragira volcano in equatorial Africa (Shinohara, 2008). In the 1998-2001 CE period, the Nyamuragira volcano erupted three times: October 1998 CE, January 2000 CE, and February 2001 CE. All these eruptions presenting a VEI = 2. Despite the evidence of two major volcanic events and considerable SO 2 volcanic emissions from Nyamuragira during the 1998-2001 CE period, all these events occurred either in the equatorial region or in the northern hemisphere mid-latitudes. The distant location of these eruptions, their magnitude (VEI ≤ 4), and their timing cannot explain the synchronous SO 4 2− , EC and MPC peaks or the presence of coarse cryptotephra shards in the 2001L. Thus, establishing a small to moderate Antarctic eruption as the potential source of volcanic products present in the 2001 ice core layer. Mount Erebus has been continuously active (lava lake activity) since 1972 CE (Kyle, 1994) and is the only volcano listed to have documented volcanic activity in Antarctica between 1998 and 2001 CE. The eruptions recorded during this period were of Strombolian type and did not exceed VEI = 2. There was a substantial increase in the number of eruptions per month during 1998 CE and 2000 CE. However, there was a sharp decrease in the number of eruptions after May 2000 CE, leading to the absence of eruptions between March 2001and February 2002(Global Volcanism Program, 2006. Despite the potential of Mount Erebus to be considered as the source to the volcanic signature in the 2001L, the relatively continuous emission from small eruptions would be represented in the ice core record as the background signal, rather than as prominent peaks. Moreover, the proximity of Mount Erebus to the ice core sites would require a small to moderate eruption to have occurred in mid-2001. However, the lack of eruptions during this period rules out the possibility of Erebus as the source of volcanic products seen in the 2001 ice core layer. Although Mount Erebus is discarded as the volcanic source, a potential small to moderate Antarctic eruption is consistent with remote sensing observations from Sturge Island on the 12 June 2001. The analyses of air mass transport pathways provide key evidence linking remote sensing observations to the ice core record. Trajectory analyses confirm air parcels passing over Sturge Island at low elevations (1,500 and 2,000 m a.s.l) during the unusual cloud formation, were transported to the ice core sites in Ellsworth Land and southern Antarctic Peninsula within a week. The detection of a volcanic signal in the WAIS ice core, despite trajectories not passing directly over the ice core site, can be explained by the coarse resolution of the trajectory model (2.5° lat.-long.), potential mixing with neighboring air parcels (both vertically and horizontally), and/or trajectories reaching the site after the designated 10 days period. The air masses passing over Sturge Island on the 12 June 2001, likely incorporated particles and chemical compounds from the eruption cloud then carried and deposited them over Marie Byrd Land, Ellsworth Land, and the southern Antarctic Peninsula.
Sturge Island appears as a strong candidate. However, the lack of ground-based observations and inconclusive remote sensing observations prevent the confirmation of it as the unequivocal source. No published information exists for Sturge Island, which appears not to have been visited by geologists. However, the Balleny islands are entirely volcanic (Hatherton et al., 1965;Wright and Kyle, 1990). From the examination of photographs and the few published descriptions and analyses (Johnson et al., 1982), each of the islands appears to be formed from the products of multiple overlapping mafic volcanic shields. Based on the strata exposed in the cliffs forming much of the coastlines, the eruptions were predominantly effusive, although a description of tuff and 'agglomerate scoria' indicates that at least some explosive events also occurred and may have affected Sturge Island. The subdued, shield-like morphologies indicate that explosivity was probably minor in the islands, however, and probably mainly low-energy, Strombolian in character. Sturge Island rises gradually from a low cliff on its west coast to >1,000 m, then falls away in prominent cliffs that form the entire east coast. Satellite images of Sturge Island show no obvious primary volcanic landforms, which is puzzling if an explosive eruption occurred as recently as 2001 CE. However, any feature would rapidly get filled by snow. Indeed, the faint outline of a possible crater rim is present at the south end of the island, close to Cape Smyth. With a diameter of >2 km, it is relatively large and low, inconsistent with a Strombolian origin. It may therefore be hydrovolcanic, which would give greater dispersal of any tephra (J. Smellie, pers. comm). Based on the topographical features of the island, the unusual cloud formation identified on the 12 June 2001 could be interpreted either as a banner cloud or as a volcanic plume. The absence of analogous cloud features associated with nearby islands could be either explained by the effect of weather interacting with variations in the island topographies, or by a volcanic eruption occurring in Sturge Island. Either way, remote sensing evidence alone is not enough to confirm or reject a volcanic eruption.
Results presented here support the volcanic origin of the unusual cloud formation and are consistent with previous studies identifying the Balleny Islands as the source of earlier volcanic products preserved in ice cores from Marie Byrd Land (Dunbar et al., 2003;Koffman et al., 2013;Kurbatov et al., 2006). Buckle Island is suggested as the source of three cryptotephra layers deposited in 1839, 1809, and 1804 CE (Kurbatov et al., 2006). The most recent of which was confirmed by historical records from sailors who observed a volcanic plume rising from Buckle Island in 1839 CE (LeMasurier et al., 1990). This 1839 eruption was recognized in the WAIS ice core and characterized by the presence of cryptotephra in two horizons with elevated particle concentration and a considerable increase of the mode particle diameter (>10 μm) over the PSD background dust (5.1 μm; Koffman et al., 2013). Likewise, several cryptotephra layers identified in the ice core record show that Antarctic eruptions typically increase the mode particle diameter (>6 μm; Koffman et al., 2013;Narcisi et al., 2010) corresponding with elevated SO 4 2− -EC and MPC deposition. In addition, air mass trajectories are consistent with previous studies showing the absence of a 2001 Sturge Island eruption record in ice cores from the inland sites of Mount Johns and the Ellsworth Mountains in the West Antarctic ice sheet (Hoffmann et al., 2020;Thoen et al., 2018).
In summary, the ice core records, air mass trajectory analyses, and remote sensing observations presented here provide strong evidence that a short-lived, small to moderate volcanic eruption took place on Sturge Island in mid-2001. Evidence suggests that volcanic products from this eruption were rapidly transported through the troposphere and deposited inland over Marie Byrd Land, Ellsworth Land, and the southern Antarctic Peninsula. The deposition over the ice sheet produced a volcanically enriched layer that has been preserved in the ice core record.
The detection of this 2001 eruption may demonstrate Sturge Island is an active volcano capable of producing small-moderate explosive events. It is possible that previous eruptions recognized in Antarctic ice core records and attributed to Buckle Island, could instead, have originated from Sturge Island. If true, Sturge Island could be at least as active as Buckle Island. Thus, suggesting a reinterpretation of the Balleny Island hot-spot dynamics (Green, 1992). Additionally, the prominent and consistent volcanic signal identified in the 2001 ice core layers from Marie Byrd Land, Ellsworth Land, and the southern Antarctic Peninsula highlight this ice core horizon as a new, 21st century, chronostratigraphic marker between the eruptions of Pinatubo/Hudson (1991 CE, VEI = 6 and VEI = 5) and Puyehue-Cordon Caulle (2011 CE, VEI = 5). As such, the Sturge Island 2001 eruption provides a valuable volcanic horizon to date ice cores from the Amundsen and Bellingshausen Seas sectors.
The evidence presented in this work of a new 21st century volcanic horizon in West Antarctic ice cores supports the occurrence of a volcanic eruption in Sturge Island in 2001. Whilst geochemical analyses of the cryptotephra shards would be required to unequivocally determine the Balleny Islands as the volcanic source, the evidence presented here is sufficiently robust to assign this cryptotephra layer in ice cores as a chronostratigraphic marker.

Conclusions
Antarctica is one of the most enigmatic volcanic regions on Earth, with potentially hundreds of volcanoes hidden beneath the ice sheet. New historical records of active volcanism in Antarctica provide valuable information to study how volcanic activity can shape the polar climate and its potential impacts on the cryosphere. A set of ice core records from Marie Byrd Land, Ellsworth Land, and the southern Antarctic Peninsula have been analyzed to validate previous inconclusive evidence of a 2001 volcanic eruption on Sturge Island, part of the Balleny Island chain. The 2001 ice core layer contains a regional input of sulfates and microparticles, which when evaluated together, are consistent with a volcanic source. The combined presence of volcanic tephra shards, particle coarsening and in-phase deposition of potentially volcanic products indicate a small to moderate Antarctic eruption as the source and a rapid tropospheric dispersion as the transport mechanism. Air mass trajectory analyses proved air parcels passing over Sturge Island during the 2001 eruption were effectively transported, within a week, to the ice core sites.
The evidence presented here builds on previous inconclusive remote sensing observations to advocate Sturge Island as an active volcano with recent eruptive states. The regional extent of this volcanic event across several Antarctic ice core records may highlight its potential as a chronostratigraphic marker to improve the accuracy of 21st century ice chronologies. The detection of small to moderate eruptions in Antarctic ice core records is a challenging endeavor, only possible by following a multi-proxy approach. No single record can provide unequivocal evidence of a volcanic event, however, when used together they present a convincing argument. Further research should be focused on performing geochemical analyses of the cryptotephra shards to unambiguously fingerprint the volcanic source.