Past eruptions of a newly discovered active, shallow, silicic submarine volcano near Tokyo Bay, Japan

Shallow, silicic submarine volcanoes pose significant, yet poorly constrained, hazards, as shown by the recent eruptions of Fukutoku-Oka-no-Ba (south of Ioto island in the Ogasawara Islands, Japan) in 2021 CE, which produced pumice rafts impacting ships and coastal communities throughout Japan (Geological Survey of Japan, 2021), and Hunga Tonga–Hunga Ha'apai in 2022, which inundated the Kingdom of Tonga with tsunami waves and ash fall, severed communications, and triggered global tsunami activity (Global Volcanism Program, 2022). The difficulties of observing eruptions, sampling submarine deposits, and constraining eruption ages mean much remains unknown about the ways in which submarine eruptions vary with water depth and eruption magnitude, while a lack of detailed eruptive histories prevents assessment of volcanic hazard.

A recent study (Mitchell et al., 2018) of pumice volatile contents from the deep, silicic Havre eruption in 2012 (in the Kermadec Islands, southwest Pacific Ocean), obtained using new Fourier transform infrared spectroscopy (FTIR) analytical techniques, demonstrated how these data can yield new insights into submarine eruption processes. Our study used volatile contents of both subaerial and submarine deposits from Oomurodashi, a newly discovered active volcano in the Izu-Bonin arc, to investigate late Quaternary eruptions of a shallow, silicic submarine volcano.

Oomurodashi is located in the northern Izu-Bonin arc, 20 km southeast of Izu-Oshima Island (population ~9000) and 60 km southwest of Tokyo Bay (Fig. 1). Its 20-km-wide, flat-topped summit lies ~120 m below sea level (mbsl) and contains a small (1.2 × 0.7 km, ~100 m deep) crater named Oomuro Hole. Oomurodashi was previously thought to be inactive; however, cruise NT07–15 of R/V Natsushima in 2007 (JAMSTEC, 2007), using a remotely operated vehicle (ROV), measured high heat flow in the crater floor and observed fresh rhyolite pumice deposits (Tani et al., 2013). Subsequent ROV surveys (NT12–19 of R/V Natsushima in 2012 [JAMSTEC, 2012]; KS-16–6 of R/V Shinseimaru in 2016 [data available from JAM-STEC]) obtained samples and discovered an active hydrothermal field within the crater. We present volatile and density data for a subset of rhyolite eruption products sampled directly from submarine outcrops and from subaerial tephra deposits on the nearby islands of Izu-Oshima and Toshima (Fig. 1).

A distinctive tephra of dark, fine-grained material dotted with white pumice is found among the predominantly basaltic tephra deposits on the islands of Izu-Oshima and Toshima, named layer O58 (¹⁴C age of 13.8–13.2 ka; Saito and Miyairi, 2008; recalculated with IntCal20, Reimer et al., 2020) and layer O3T (found between layers dated to 15 and 11 ka; Oikawa and Tani, 2020), respectively (Fig. 1). The pumice clasts are generally <1 cm (longest axis) but range up to 4 cm in size (Fig. 2). We compared their major- and trace-element geochemistry with those of samples from Oomuro Hole as well as published geochemical data for two neighboring rhyolitic volcanic islands (see the Supplemental Material¹). These data confirmed that Oomurodashi was the source of both tephra layers.

The dark, fine-grained matrix forms the majority of both layers O58 and O3T, with particles <250 μm comprising >70% by mass. Most of these are aggregates, resistant to disaggregation methods, of even finer grains that are dense or only weakly vesicular; some have stepped features indicative of intense brittle fragmentation (Fig. 2C). Such particles may be associated with explosive phreatomagmatic activity (Zimanowski et al., 2015). Other features are particles with manganese coatings likely formed in the submarine environment, and dense angular lithics ≤1 cm in the base of O58 (see the Supplemental Material). We propose that these tephras are the distal deposits resulting from the explosive formation of the Oomuro Hole crater by a shallow submarine, phreatomagmatic eruption.

Volatile and porosity data were obtained for both subaerial tephra pumice clasts and submarine samples: a lava (719-R14) and loose, weakly vesicular clast (1409-R10) from the rim of Oomuro Hole, a lava (1970-R01) from a small (100-m-wide, 24-m-high) knoll on the flat summit, and a loose pumice clast (1407-R01) from the northeast flank (Fig. 1). Total water (H₂O_t), molecular water (H₂O_m), and hydroxyl water (OH) contents of matrix glasses were obtained by FTIR using the species-dependent ε3500 (not OH-by-difference) method following the procedures of McIntosh et al. (2017) and Mitchell et al. (2018). CO₂ was below detection limits for all samples. All samples had <5% isolated porosity. See the Supplemental Material for full details of the methods, sampling depth, porosity, and FTIR data.

The equilibrium concentrations of H₂O_m and OH (equilibrium speciation) at a given H₂O_t concentration and temperature are controlled by a species interconversion reaction and vary in a known way (e.g., Stolper, 1982; Nowak and Behrens, 2001). The interconversion reaction rate decreases dramatically during cooling until reaching the glass transition temperature, T_g, at which H₂O_m and OH concentrations become fixed (e.g., Dingwell and Webb, 1990; Zhang et al., 1995). Interpreting H₂O species data thus requires an assumption of each sample's T_g. T_g depends on H₂O_t and cooling rate and is lower for higher H₂O_t and/or slower cooling. Sample OH and H₂O_t data are therefore plotted (Fig. 3A) against values expected for T_g ranging from 800 °C (instant quench) to 600 °C (equivalent to ~0.8 wt% H₂O_t and 10 °C/min cooling; Giordano et al., 2008). Our sample data plot below these curves, so they have excess H₂O_m; i.e., they have been hydrated by disequilibrium addition of H₂O_m.

Diffusion of water in melts and glasses occurs by movement of H₂O_m, with some subsequent interconversion of H₂O_m to OH to maintain local equilibrium (Zhang et al., 1991). Excess H₂O_m thus indicates H₂O_m addition at temperatures where the interconversion reaction rate was too slow to maintain equilibrium speciation. This could be due to slow low-temperature (secondary) hydration (e.g., Giachetti and Gonnermann, 2013) or more rapid hydration at intermediate temperatures (e.g., Zhang et al., 1995; McIntosh et al., 2014).

FTIR data for pumice from the 2012 submarine Havre eruption revealed they were hydrated, and it could be inferred that hydration occurred during cooling in the water column because their young age excluded slow low-temperature hydration as a possible cause (Mitchell et al., 2018). The older Oomurodashi samples, however, could have experienced either hydration mechanism. H₂O species data along hydration profiles can help to distinguish these scenarios; unfortunately, only pumice 1407-R01 was successfully prepared for profile analysis (Figs. 3B and 3C). Unlike H₂O_t and H₂O_m, OH (wt%) does not increase toward its bubble walls. Hydration thus occurred with little to no conversion of excess H₂O_m to OH. For the measured H₂O_t range, even a slow cooling rate of 10 °C/min would give T_g > 650 °C for this pumice (Giordano et al., 2008), which would be reached in ~15 min. This is enough time for the observed H₂O_m diffusion profile to form, despite the order of magnitude drop in H₂O_m diffusivity for cooling from 800 °C to 600 °C (Ni and Zhang, 2008); however, at these temperatures and time scales, at least some interconversion of excess H₂O_m to OH would be expected before OH became fixed at T_g (Zhang et al., 1995). Hydration thus probably occurred at low temperature. Low-temperature H₂O diffusivity data span several orders of magnitude (e.g., Giachetti and Gonnermann, 2013), but such a profile could form in ~1–100 k.y., assuming H₂O_t diffusivity of ~10^-21 to ~10^-23 m²/s. Without profiles for the other samples, it is impossible to say conclusively whether all hydration occurred at low temperature. The observed hydration of the ca. 13.5 ka subaerial pumice clasts is, however, also consistent with low-temperature hydration after their eruption, particularly as bulk hydration is enhanced for vesicular samples that experience large diurnal temperature fluctuations (Friedman and Long, 1976).

Unlike subaerial tephra layers, it is difficult to date late Quaternary submarine eruption deposits. We estimated eruption ages of the inplace lavas by comparing their OH contents with their sampling water depth and past sea level, in a manner analogous to that used to constrain paleo-ice thickness for subglacial eruptions (e.g., Tuffen et al., 2010). H₂O solubility is pressure dependent, so rising magma degasses H₂O until final emplacement as a seafloor lava flow. The H₂O_t content of (unhydrated) matrix glasses records this emplacement pressure, which can be converted into an equivalent water depth. Hydration has altered the H₂O_t record of these lavas, so we instead used their OH contents, fixed at T_g, to estimate their quench pressure, hence emplacement depth. Figure 3D shows sample OH data plotted along curves of OH versus pressure (bottom axis) and equivalent water depth (top axis) calculated for T_g of 600–800 °C, with their ROV sampling depths shown above the top axis. For past sea level, we used a global sea-level curve adjusted for regional tectonic uplift (Fig. 3E; see the Supplemental Material).

Cooling rates of submarine rhyolite flows are not constrained, but given the low original H₂O_t content of the shallow lavas, we assumed a likely T_g of 700–800 °C. If slow cooling led to T_g < 700 °C, then lavas will be younger than calculated, but this potential error is typically <1 k.y. due to the convergence of T_g curves at low OH. Analytical uncertainties were considered by propagating ±1σ of the FTIR mean (Fig. 3D).

Accordingly, the summit knoll lava 1970-R01 has an OH content consistent with quench when sea level was up to 28 m shallower than today, giving an eruption age of ca. 9.8–7.3 ka. Lava 719-R14 also contains less OH than expected for its sampling water depth, but its larger FTIR uncertainty translates to large uncertainty in age, giving an eruption age of ≤14 ka. This lava is exposed at the top of the Oomuro Hole crater with vertical cooling joints, implying it was emplaced on the summit and then cut by the formation of Oomuro Hole, which we infer from tephra O58 to have formed 13.8–13.2 ka. This sequence is also consistent with a preliminary quartz electron spin resonance date of 12 ± 4 ka for this lava (Asagoe et al., 2013). Although FTIR data are consistent within error with an age ≥13.2 ka, we considered additional factors that may have affected the estimated age (see also the Supplemental Material). Ages will be underestimated if lavas have been uplifted since their eruption; however, Oomurodashi's flat summit, formed via wave planation at the last sea-level lowstand, agrees well with our uplift-adjusted sea-level curve and excludes any significant local deformation since that time. Ages will also be underestimated if lava volatiles had not fully equilibrated with the emplacement pressure before quenching. Such incomplete degassing is more likely for lava 719-R14 because the cross-sectioned flow interior was sampled rather than (as for the summit knoll) the exterior, and this may be the cause of the greater variability in its OH data. Generally, flow exteriors are more likely to have fully degassed to their quench pressure, as seen for subaerial lavas at Mount St. Helens (USA) and Santiaguito (Guatemala), although the most vent-proximal samples may retain up to 0.1 wt% excess H₂O_t (Anderson et al., 1995). However, even if lava 719-R14 retained as much as 0.15 wt% excess OH from incomplete degassing, its age cannot exceed 15.5 ka.

Tephra pumices had higher OH contents than the submarine samples (Figs. 3A and 3D), implying that they quenched at higher pressures despite being deposited subaerially. Assuming T_g of 700–800 °C, their OH contents are equivalent to quench pressures of 3.6–4.0 MPa (O58) and 2.1–2.3 MPa (O3T). At 13.8–13.2 ka, the Oomurodashi summit was ~55 mbsl, equivalent to ~0.7 MPa. These OH contents therefore record fragmentation and quenching within the shallow edifice. Assuming a magma/lithostatic pressure gradient (with density of ρ = 2300 kg/m3), we calculated quench depths within the conduit of ~130–150 m (O58) and ~60–70 m (O3T); these depths would have been shallower if the syneruptive crater had already begun to form (e.g., ~80–100 m and ~20–30 m depth for a 100-m-deep crater like today). Such depths indicate shallow fragmentation, consistent with shallow magma-water interaction (e.g., Dellino et al., 2012), or maar-type explosions within a partly infilled vent that would have been sufficiently shallow to produce surface ejecta (Graettinger et al., 2014). Thus, the volatile data support the interpretation that these tephras were produced by shallow phreatomagmatic explosions that formed the Oomuro Hole crater, and they demonstrate the ability of new FTIR methods to obtain fragmentation depths of juvenile material.

Pumice clasts around the rim of Oomuro Hole indicate that the eruption (or subsequent eruption) shifted to a pumice-producing style after crater formation. Clast 1409-R10, collected at 118 mbsl, was dense (33% porosity) and would not have been buoyant in water upon eruption. Its OH content indicates a quench depth of 70–80 mbsl. If it erupted at 13.8–13.2 ka (when the sampling location was 50–56 mbsl), it would thus need ~20 m of nonbuoyant postquench ascent to reach its final deposition location. As momentum-driven ascent in water is limited, it is likely from a more recent eruption, when its deposition location was also 70–80 mbsl (ca. 11.3–10.7 ka). By contrast, pumice 1407-R01 has an OH content equivalent to a quench depth of ~30 mbsl. This pumice (72% porosity) would initially have been buoyant in water, and its final location 5 km from the volcanic center suggests it drifted at shallow depths prior to waterlogging and deposition. Waterlogging of hot pumices that ingest water during cooling can occur rapidly, potentially limiting buoyant ascent and dispersal (Whitham and Sparks, 1986). ROV surveys extending 500 m east of Oomuro Hole (i.e., downstream of the strong Kuroshio current) during Cruise NT12-19 found abundant fresh rhyolitic pumice covering the summit. However, in assessing potential future hazards, it must be noted that if shallow eruptions also produced pumice that cooled within a subaerial plume and thus ingested air rather than water (Whitham and Sparks, 1986), or had a greater percentage of isolated porosity (Manga et al., 2018), such pumice could have formed buoyant rafts that are not preserved in Oomurodashi's local deposits (e.g., Carey et al., 2018).

The recently discovered active Oomurodashi volcano has had at least three eruptions since ca. 14 ka, which include submarine lava flows, phreatomagmatic explosions creating subaerial tephras, and submarine eruptions of pumice with varying buoyancies. Oomurodashi's location next to inhabited islands and major shipping lanes entering Tokyo Bay therefore makes it a significant hazard. FTIR volatile data of lavas and pyroclasts integrated with porosity and textural data will provide a new framework for interpreting submarine eruption processes. Acquisition of these data from both local and distal deposits, such as those in marine sediment cores, will enable us to investigate the full spectrum of submarine eruption styles and better understand their potential hazards.

We thank the scientists and crew of the R/V Natsushima and the R/V Shinseimaru. This work was supported by Japan Society for the Promotion of Science KAKENHI grants JP00470120 and 16K05584. We thank S. Bryan and five anonymous reviewers for their constructive reviews.