Large volcanic eruptions can significantly impact global climate due to the radiative effects of sulfate aerosols released into the atmosphere. The Millennium Eruption (ME) of Mt. Baekdu (Paektu, Changbaishan), occurring in late 946 CE, is among the largest eruptions of the Common Era, with a Volcanic Explosivity Index (VEI) of 6 or 7. Tephra from this eruption is widely distributed across East Asia and has been identified in Greenland ice cores. Two distinct compositional phases are recognized: an initial rhyolite (Phase 1) followed by a trachyte (Phase 2). The temporal relationship between these phases remains unclear, with some studies suggesting a hiatus of several months based on proximal and medial ash analyses. Historical records, such as chronicles from Japanese temples, provide some evidence of ash fall and thunderous sounds around the eruption period, but these records are not precise enough to clarify the eruptive sequence. Uncertainties regarding the duration of any hiatus and the total sulfur emissions are critical factors influencing estimates of the eruption's climate forcing. Estimates for syn-eruptive sulfur emissions range from 2 to 11 Tg, but this petrologic method can underestimate total emissions, particularly when pre-eruptive sulfur is considered. Total sulfur release estimates for the ME vary widely, from 2–7 Tg to as much as 45 Tg, impacting the uncertainty in evaluating its climate impact. Estimates of plume height also vary significantly, influencing the altitude of sulfate aerosol dispersal. The lack of strong evidence for transient Northern Hemisphere summer cooling following the ME in tree-ring temperature reconstructions and historical records suggests a potentially limited climatic impact. Ice cores provide a highly time-resolved record of aerosol deposition, offering an opportunity to refine our understanding of the ME timeline, atmospheric transport of sulfur emissions, and the eruption's climate forcing.

Previous research on the Mt. Baekdu Millennium Eruption has focused on tephra distribution, geochemistry, and dating methods. Studies using proximal and medial ash deposits have yielded conflicting conclusions regarding the eruption's phasing. While some suggest a single eruptive episode, others propose a hiatus between the rhyolite and trachyte phases, based on observations such as lahar deposits and grain-size analysis. Estimates of sulfur emissions vary widely, ranging from 2 to 45 Tg S, highlighting the difficulty in accurately determining total sulfur release from volcanic eruptions, particularly when pre-eruptive sulfur is considered. Similarly, estimates of plume height and sulfur injection altitude differ significantly in existing literature, with implications for understanding the eruption's climate impact. Previous studies using tree-ring data and ice core analyses of the eruption have suggested that the eruption’s climate impact might have been less than originally estimated.

This study involved both discrete and continuous analyses of the NGRIP1 ice core from Greenland. Discrete samples were analyzed for tephra geochemistry using scanning electron microscopy (SEM) and electron probe microanalysis (EPMA). Continuous flow ice-core melter system at the Desert Research Institute (DRI) was used to obtain high-resolution measurements of liquid conductivity, elemental chemistry (including sulfur), and insoluble particle concentrations. The effective depth resolution for these measurements was 1.8 mm, 20 mm, and 3 mm, respectively. Sea-salt sulfur (ssS) was calculated from sodium and calcium concentrations and subtracted from the total sulfur to determine non-sea-salt sulfur (nssS) concentrations. An annual net accumulation model was developed based on seasonal snowfall data from north-central Greenland (1991-1995). This model, along with air trajectory calculations using the Hybrid Single-Particle Lagrangian Integrated Trajectory (HYSPLIT) model, was used to estimate the time interval between the rhyolite and trachyte phases. Multiple sulfur isotopes (δ³⁴S and Δ³³S) were measured using a Neptune Plus Multi Collector Inductively Coupled Plasma Mass Spectrometer (MC-ICP-MS) to assess the altitude of sulfate aerosol dispersal. Background-corrected sulfur isotope values were calculated using the background sulfate concentration before the eruption. The study also estimated sulfate deposition in Greenland based on the NGRIP average accumulation rate and the high-resolution nss-sulfate concentration in the ice core.

The analysis of the NGRIP1 ice core revealed a clear transition from rhyolite to trachyte tephra, corresponding to distinct spikes in insoluble particle concentrations. The depth interval between these spikes, coupled with the annual net accumulation model, suggests a hiatus between the two eruptive phases lasting approximately one to two months. The study found negligible sulfur mass-independent fractionation (S-MIF) in the ice core sulfate associated with the ME. This, along with the near-synchronicity between particle and sulfate deposition, and the peak sulfur fallout in winter, indicates that the sulfate aerosols were primarily confined to the upper troposphere/lower stratosphere (UT/LS). The estimated sulfate deposition at the NGRIP site (11 kg km⁻²) is significantly lower than that observed from the 1815 CE Tambora eruption (40 kg km⁻²), reflecting a lower sulfur yield and injection height for the ME. The model suggests that the sulfate deposition spanned approximately 232 ± 42 days (2σ), primarily during boreal winter and spring. The absence of significant stratospheric sulfur loading, coupled with the short atmospheric lifetime of the UT/LS aerosols and the winter timing of the deposition, resulted in minimal climate forcing from the ME. The conductivity peak and width of Spike 2 (trachyte phase) were found to be larger than Spike 1 (rhyolite phase), suggesting a higher SO₂ release during Phase 2. This finding can help in understanding sulfur budgets of magma compositions in relation to eruptive volumes and degassing histories.

The findings of this study address the long-standing uncertainty regarding the phasing of the Mt. Baekdu Millennium Eruption. The evidence strongly supports the hypothesis of a hiatus of approximately one to two months between the rhyolite and trachyte phases. This has implications for understanding volcanic hazards at large caldera complexes, specifically the timing required to conclude an eruption. The muted climate forcing observed is consistent with the limited stratospheric injection of sulfur. This is attributed to the combination of low-altitude sulfate aerosol dispersal, a short atmospheric lifetime of the aerosols (due to their being in the UT/LS), rapid scavenging of aerosols, and the timing of the eruption in boreal winter. This case study underscores the need for detailed, high-resolution analyses to assess the climate impact of individual volcanic events, as generalized models may not accurately capture the nuances of each eruption. The study's findings contribute to a more refined understanding of the eruption's timeline, atmospheric processes, and its ultimate climatic impact, particularly by showing the importance of the altitude of the aerosol plume.

This high-resolution study of the Mt. Baekdu Millennium Eruption provides strong evidence for a one-to-two-month hiatus between the rhyolite and trachyte phases. The limited climate forcing is attributed to UT/LS sulfur injection height, a short atmospheric lifetime of the sulfate aerosol veil, and peak deposition occurring during the boreal winter. These findings refine our understanding of the eruption's timeline and impact, improving our ability to assess volcanic risks at similar caldera complexes. Future research could focus on further refining the accumulation model for Greenland, potentially using longer-term data sets or higher resolution models, to reduce uncertainties in the time interval estimate. More detailed atmospheric models incorporating the high mass eruption rate could further elucidate the discrepancy between plume height and sulfur injection height.

The primary limitation of this study is the reliance on a relatively short dataset (1991-1995) for the annual net accumulation model. This could introduce some uncertainty into the time interval estimate between the two eruptive phases. Additionally, while the HYSPLIT model provides a reasonable estimate of transport time, it does not fully capture the complexities of ash dispersion and deposition. Finally, the background correction of sulfate concentrations for isotopic analysis may be influenced by potential overlap of other sulfur sources with the ME signal.