Black carbon footprint of human presence in Antarctica

The study investigates how human activities—primarily research operations and tourism—contribute to black carbon (BC) deposition in Antarctica, thereby darkening snow, reducing albedo, and accelerating melt. While BC has been detected in remote snow globally, background Antarctic levels are typically below 1 ng/g—far lower than in the Arctic or mid-latitudes. Limited meridional transport suggests local sources dominate BC deposition in Antarctica. With a surge in human presence (76 stations, ~5,500 summer personnel; ~74,000 tourists in 2019–2020), local BC emissions from vessels, aircraft, generators, and vehicles may substantially impact snow albedo and melt. The research aims to quantify BC concentrations across widely visited regions, apportion absorbing constituents, and estimate associated radiative forcing and premature snowmelt attributable to local human activities.

Prior work shows BC in snow reduces albedo and increases melt. Global observations document BC in snow across the Arctic, North America, China, the Tibetan Plateau, the Himalayas, and the Andes. Antarctic measurements and ice cores confirm BC presence, with background snow concentrations consistently below 1 ng/g—about an order of magnitude lower than in the Arctic. Long-range transport events (wildfire smoke from South America and Australia; dust from Australia and Patagonia) reach Antarctica but back-trajectory analyses suggest minor contribution to Antarctic absorbing aerosol deposition. Local sources near stations (e.g., Palmer, McMurdo Dry Valleys, South Pole) elevate BC above background. The Antarctic Peninsula and archipelagos, with numerous stations and tourist landings, are particularly exposed. Increasing tourism and expansion of research infrastructure likely enhance the BC footprint. Dust from extensive nearby ice-free areas also contributes to light absorption, complicating attribution using spectral properties.

Sampling and sites: 155 snow samples were collected over four consecutive summers (2016–2017 to 2019–2020) at 28 sites along ~2,000 km from King George Island (62°S) to Union Glacier Camp in the Ellsworth Mountains (79°S). Coastal sites on the South Shetland Islands, Palmer Archipelago, and Antarctic Peninsula were prioritized; two interior sites were sampled near Union Glacier. Sampling was generally late summer (end of melt season) except Union Glacier (early December). Sites were selected hundreds of meters to several kilometers away from apparent sources (stations, landings, roads, airfields) to represent broader areas. Duplicate surface samples (~1 m apart) were taken; vertical snowpits (10–20 cm intervals) at four Peninsula sites and a 2.6 m snowpit at Union Glacier captured vertical and interannual variability. Sample masses were 1–2 kg. Field blanks indicated contamination at tenths of ng/g.

Filtration and spectroscopy (Meltwater Filtration Technique): Snowmelt was vacuum-filtered through 0.4 µm Nucleopore filters. Spectrophotometry (340–750 nm, 5 nm resolution) measured transmittance to compute absorbance. Calibration curves from filters loaded with known Monarch-71 soot provided wavelength-dependent BC-equivalent loading and concentration. Because non-BC absorbers (dust, algae) attenuate more at short wavelengths, the average concentration over 700–750 nm provided C_MAX (maximum possible BC). The absorption Ångström exponent (α) was derived by fitting AOD(λ)=βλ^−α (420–620 nm) using MAC values for soot. Best-estimate BC concentration (C_EST) was obtained by attributing measured absorption to BC and non-BC components: α = zα_BC + (1−z)α_NBC with α_BC=1.1 and α_NBC=4, yielding the BC fraction z and AOD_BC, which was then used to scale C_MAX to C_EST via the 700–750 nm absorption ratio.

Albedo reduction and radiative forcing: Regional albedo reductions (ΔA) due to BC were estimated using Dang et al. parameterizations as a function of cloud fraction (CF), snow grain radius (r = 200–600 µm), and BC concentration. Over the Peninsula and archipelagos, CF approaches 1; thus ΔA≈ΔA_cloudy. ERA5 provided DJF downwelling shortwave irradiance (I≈250 W/m²) and CF; MERRA-2 provided albedo context.

Snowmelt estimation: The extra absorbed energy over DJF was computed as ΔA·I·d (d=90 days), converted to melted snow mass per unit area W by dividing by the latent heat of fusion (E=334 kJ/kg). Total seasonal premature melt W_L was estimated by multiplying by affected area D. Monte Carlo simulations sampled ΔA (0.001–0.004) and D ranges to produce distributions of W_L for King George Island (D=6–48 km²) and tourism-impacted areas on the Peninsula/archipelagos (D=100–500 km²).

Back-trajectory and statistics: HYSPLIT (GDAS) 72 h backward trajectories for DJF (2010–2020) and pre-sampling periods at selected sites assessed source regions; clustering by total spatial variance supported local dominance. Statistical analyses (ANOVA and Tukey HSD) tested site groupings and differences in BC and α; one-way t-tests assessed deviations from 0 or 1 ng/g background. Python/Matplotlib was used for plotting; data and code repositories are provided.

• BC concentrations around research stations and tourist-landing areas on the Antarctic Peninsula and associated archipelagos are substantially above Antarctic background (~1 ng/g), with a median ~3 ng/g across all samples.
• Spatial gradients: Northern South Shetland Islands (King George, Greenwich): ~3–5 ng/g; Southern South Shetlands and Trinity Peninsula: ~4–7 ng/g (peak ~8 ng/g near Esperanza Base at 63.4°S); Palmer Archipelago: Trinity Island ~3–5 ng/g; Doumer Island ≤2 ng/g; further south: Petermann Island and Detaille Island ~1–2 ng/g.
• Interior sites (Union Glacier, 79°S): surface snow 1–3 ng/g; Ångström exponent (α) 1–3, lower than Peninsula sites, indicating less dust contribution relative to coastal archipelagos.
• Ångström exponent generally 2–3 over Peninsula/archipelagos, indicating significant non-BC absorption (dust). Ice-free areas (e.g., Byers and Ulu Peninsulas) likely supply dust. Presence of snow algae at several sites influenced some filters (α up to 4–6), but algae contribute negligible absorption beyond 700 nm.
• Union Glacier 2.6 m snowpit suggests interannual variability: BC peaked around 2013–2014 (~3 ng/g) coincident with initiation of Chilean Union Glacier Camp operations; α varied inversely with BC, suggesting relatively constant dust with varying BC inputs.
• Estimated albedo reductions attributable to local BC at impacted sites: ΔA ≈ 0.001–0.004, yielding positive shortwave forcing of ~0.25–1.0 W/m² (DJF, I≈250 W/m²).
• Estimated seasonal premature melt at impacted sites: 5–23 kg/m² (5–23 mm w.e.) over DJF.
• Aggregated impacts: • King George Island (research-driven): 0.4 ± 0.2 Mt seasonal premature snowmelt; ~0.6 ± 0.3 kt snow per researcher bed (11 stations; ~700 beds). • Tourism across Peninsula/archipelagos: 4.4 ± 2.3 Mt seasonal premature snowmelt; ~83 ± 43 tons per visitor (average ~53,000 visitors per year during 2016–2020).
• Back-trajectory analyses support the conclusion that local emissions dominate BC deposition at sampled sites; no major Patagonia wildfire events occurred during sampling years to explain elevated BC via long-range transport.

Findings demonstrate that human presence in Antarctica measurably elevates BC in snow near research infrastructure and tourist landing zones, compared to the continent’s very low background. Elevated BC, often accompanied by dust contributions, reduces albedo and induces radiative forcing sufficient to accelerate seasonal melt by up to ~23 mm w.e. at impacted sites. Spatial patterns align with intensity of human activities: higher BC near populated stations and frequent tourist stops in the northern Peninsula and archipelagos, diminishing southward and in the continental interior. The Union Glacier record illustrates that even deep-field logistical hubs raise BC above background with temporal variability tied to operational activity. These results indicate that local emissions, rather than long-range transport, are the principal drivers of absorbing aerosol deposition in these regions during the study period. The quantified per-capita/per-bed footprints reveal that research operations, with their fuel-intensive logistics and power generation, produce larger per-person impacts than individual tourists, while the aggregate impact of tourism is substantial due to high visitor numbers. The study provides actionable evidence that mitigating local BC emissions can reduce premature melt in a sensitive, rapidly changing region.

The study provides a continent-scale assessment along a 2,000 km transect showing that local human activities (research and tourism) measurably increase BC in Antarctic snow, leading to significant albedo reductions, positive radiative forcing, and accelerated seasonal melt at impacted sites. Quantified impacts include island- and region-scale melt burdens and per-capita footprints for researchers and tourists. The work underscores the need for mitigation: limiting station footprints and emissions, transitioning to cleaner fuels and hybrid/electric technologies for ships and aircraft, improving energy efficiency, and adopting renewable power at research facilities. Future research should expand spatial and temporal coverage, refine source attribution and deposition modeling, quantify dust and biological contributions, and evaluate the efficacy of mitigation measures through sustained monitoring.

• Spatial and temporal coverage is limited: sampling occurred at 28 sites over four summers, mostly in late summer and primarily coastal, potentially biasing estimates toward higher BC due to surface concentration during melt.
• Representativeness of BC-impacted areas is uncertain; total melt estimates rely on assumed impacted-area ranges and Monte Carlo simulations rather than exhaustive mapping.
• Albedo reductions were estimated using parameterizations (Dang et al.) and regional averages of irradiance and cloud fraction, not site-specific radiative transfer for each location.
• Spectral attribution of absorption to BC vs. non-BC (dust, algae) via Ångström exponent involves assumptions (α_BC=1.1; α_NBC=4) and may introduce uncertainties where mixed or unusual particle types occur.
• Some samples showed influence of snow algae; although long-wavelength absorption was used to constrain BC, biological variability may affect α and inferred partitioning.
• Potential small contamination (field blanks: tenths of ng/g) and sub-meter horizontal variability can introduce uncertainty; duplicates sometimes differed by slightly >1 ng/g.
• Back-trajectory analysis was performed for selected sites and periods; not comprehensive across all sampling events.