Hazardous explosive eruptions of a recharging multi-cyclic island arc caldera

Large silicic volcanic systems can generate catastrophic caldera-forming eruptions with global impacts. Many such systems undergo recurrent caldera cycles that typically include an initial rejuvenation and recharge stage with frequent smaller eruptions, followed by a build-up stage culminating in a climactic caldera-forming event. Although early- to mid-cycle eruptions are smaller, they can still pose major hazards to nearby populations and infrastructure. The timing of reactivation and onset of explosive activity after caldera collapse is poorly constrained because early-cycle deposits are often buried within intra-caldera fill. Santorini, a classic multi-cyclic caldera in the South Aegean Volcanic Arc, last experienced a caldera-forming eruption around 1600 BCE (Minoan eruption; 34.5 ± 6.8 km3 DRE). Post-collapse volcanism built the intra-caldera Kameni Volcano, with historical records since 197 BCE describing mainly effusive activity. This study aims to reconstruct the submarine architecture and eruptive history of intra-caldera deposits associated with Kameni Volcano to assess the potential for hazardous explosive eruptions early in the caldera cycle.

Previous work established Santorini’s multi-cyclic caldera behavior with interspersed Plinian eruptions and multiple caldera collapses over the past ~650 ka. Post-Minoan activity at Kameni was regarded as predominantly effusive to mildly explosive, based on onshore deposits and historical accounts. Seismic studies previously subdivided the caldera infill into three main units and variably interpreted them as Minoan, mixed, or post-Minoan deposits. The Minoan eruption DRE volume was recently revised to 34.5 ± 6.8 km3. Comparable caldera systems (Campi Flegrei, Taupo, Rabaul, Nisyros) show hazardous activity during early cycle stages, underscoring the need to understand reactivation timescales. Historical records of Santorini document eruptions at 197 BCE, 46–47 CE, 726 CE, and numerous Nea Kameni eruptions from 1570–1950 CE. Despite historical descriptions of abundant 726 CE floating pumice across the Aegean, substantial stratigraphic deposits had not been identified onshore, aside from a thin pumice layer on Palea Kameni, leaving a discrepancy between accounts and geological evidence.

- Drilling and core analysis: IODP Expedition 398 (Dec 2022–Feb 2023) drilled four intra-caldera sites (U1594–U1597) in Santorini’s southern and northern basins. Multiple holes at U1595 (three) and U1596 (two) improved recovery; single holes at U1594 and U1597 faced challenging conditions. Lithologies were logged using standard pyroclastic terminology, noting drilling artefacts. - Lithostratigraphy: Five lithologic units (L1–L5) were identified. L1 (~0–20 mbsf) consists of well-sorted ash with a metre-thick lapilli layer; L2 (~20–55 mbsf) grades from pumice lapilli (to 5 cm) to lapilli-ash; L3 (~55–60 mbsf at U1595) comprises interbedded ash and tuffaceous mud; L4 (~60–100 mbsf) contains lithic gravels/sands with blocks; L5 (>100–126+ mbsf) includes pumice lapilli, red ash, and lithics. - Seismic data: High-resolution 2D reflection seismic profiles from multiple cruises (2006–2019) with vertical resolution ~2.5–15 m were processed (filtering, deconvolution, multiple suppression, migration). Seismic units S1–S3 were mapped, with S1 shallow and incoherent, S2 with subparallel higher-amplitude reflections and an undulating unconformity, and S3 onlapping an incoherent acoustic basement. Acoustic blanking under Kameni limited subsurface imaging below the edifice. - Core–seismic integration: Shipboard P-wave velocities and densities (WRMSL, discrete measurements) enabled correlation of lithologic units to seismic units: S1 ↔ L1, S2 ↔ L2; L3 lies at the S2–S3 boundary; L3–L5 correlate within S3. A distinct high-amplitude reflector at the S2–S3 boundary locally corresponds to an incoherent subunit interpreted as a buried lava flow. - Mapping and volumetrics: Seismic horizons were traced to generate base and thickness maps for S1–S3. Two-way travel times were converted to thickness using measured velocities. Bulk and dense-rock equivalent (DRE) volumes were computed. Minimum volumes exclude the Kameni edifice; maximum volumes for S2 and S3 were estimated via interpolation beneath Kameni. - Chronostratigraphic ties: Offshore extensions of dated onshore lava flows (46–47 CE, 726 CE, 1707–1711 CE, >1711 CE) were imaged as laterally confined subunits with chaotic internal character and high-amplitude tops, typical of submarine lavas. These constrain S1 as post-726 CE, S2 between 47–726 CE, and S3 pre-47 CE. Buried lavas at the S2–S3 boundary likely correlate with 197 BCE. - Geochemistry: Major and trace element analyses of glass shards from unit L2 and distal marine ash were compared with established Aegean arc tephra datasets, linking unit L2 to the Kameni 726 CE compositional field. - DRE conversion: Moisture and density measurements (wet/dry mass, pycnometer volumes) on 86 samples provided porosity and densities to derive DRE conversion factors. A solid density of 2,660 kg m−3 was assumed from maximum grain densities to mitigate vesicle effects.

- Identification of a major submarine explosive eruption associated with Kameni Volcano around 726 CE. Seismic unit S2 consists entirely of pumice and ash with no intercalated muds, indicating a single uninterrupted eruptive event. - Volumes: S1 bulk 0.30 ± 0.04 km3 (DRE 0.09 ± 0.01 km3); S2 bulk 2.0 ± 0.7 km3 (DRE 0.53 ± 0.18 km3) within the caldera; adding a distal ash layer (~0.4 km3 DRE; age 200–950 CE) yields a total bulk up to 3.1 km3 and ~1 km3 DRE for the 726 CE eruption; S3 bulk 2.65 ± 0.75 km3 (DRE 1.30 ± 0.37 km3). - Eruption magnitude: The 726 CE eruption attains VEI 5, larger than previously assumed worst-case scenarios (VEI 3–4) for Santorini; comparable in order to the 2022 Hunga Tonga-Hunga Ha'apai event. - Stratigraphic relationships and ages: S1 post-dates 726 CE and likely records volcaniclastic deposits from mainly effusive Nea Kameni eruptions (1570–1950 CE), including some explosive intervals. S2 is bracketed between the 46–47 CE and 726 CE lavas; S3 predates 46–47 CE. Buried lava flows at the S2–S3 boundary in both basins probably correspond to the 197 BCE eruption. - Vent location and depositional style: Mapping suggests a vent between Nea and Palea Kameni. The caldera-filling nature of S2 indicates emplacement by gravity flows and fallout of water-saturated pumice, largely confined to the submarine realm, explaining the paucity of onshore deposits. - Acoustic basement and timing: Onlap of S3 onto the acoustic basement and correspondence of basement ridges to landslide scarps indicate S3 post-dates the Minoan collapse; no Minoan deposits were recovered intra-caldera. - Hazard implications: Demonstrates Santorini’s capacity for large explosive eruptions early in the caldera cycle; significant unconsolidated pumice/ash under Kameni lavas and steep submarine slopes imply flank instability risk near the 2011–2012 unrest deformation focus.

The integration of intra-caldera cores with high-resolution seismic stratigraphy resolves the timing and magnitude of post-Minoan eruptive phases at Santorini. The recognition that unit S2 represents the voluminous, predominantly submarine 726 CE eruption reconciles extensive historical reports of floating pumice with a previously missing geological record, addressing the question of how soon after caldera collapse explosive activity can resume. The demonstrated VEI 5 magnitude indicates that early-stage caldera-cycle volcanism at Santorini is not limited to small effusive eruptions. This elevates the assessed hazard for Santorini and the broader eastern Mediterranean, where tsunamis from submarine explosions or slope failures, widespread pumice rafts, and ash plumes could disrupt coastal communities, aviation, shipping, and subsea infrastructure. The locus of deposition and the structural setting (topographic step along the Kameni Line and proximity to a subcircular depression) suggest susceptibility to sector collapse, analogous to other marine volcano failures (for example, Anak Krakatau). More broadly, the findings highlight that onshore eruption archives can be incomplete for submarine-dominated events, implying that other recharging silicic calderas may also have experienced underestimated explosive activity, with consequences for regional hazard models.

This study reconstructs the intra-caldera stratigraphy of Santorini using core–seismic integration, revealing that a major VEI 5 submarine explosive eruption occurred in 726 CE, depositing up to 3.1 km3 of pumice and ash largely within the caldera. It establishes a refined chronostratigraphy (S3 pre-46–47 CE, S2 between 47–726 CE, S1 post-726 CE) and quantifies deposit volumes, demonstrating that Santorini can generate hazardous explosive eruptions early in the caldera cycle. The results mandate updated hazard assessments for Santorini and similar systems, considering submarine eruption dynamics and slope instability. Future work should: improve imaging beneath Kameni to reduce volumetric uncertainty; obtain tighter geochronological constraints on buried lava flows (for example, 197 BCE); expand distal tephra correlations; and develop integrated shoreline-crossing monitoring and early-warning strategies for marine volcano hazards.

- Drilling limitations: Acoustic basement was not penetrated; core recovery decreased with depth, with low recovery in L4 and L5 limiting detailed correlation within S3. - Imaging gaps: Acoustic blanking beneath Kameni prevented tracing units directly under the edifice, requiring interpolation for maximum volume estimates of S2 and S3. - Chronological uncertainties: Ages of some Palea Kameni lavas are inferred from historical accounts; correlation of buried lava flows to the 197 BCE eruption is tentative. The distal ash layer correlated to 726 CE has an age range of 200–950 CE. - Volume uncertainties: DRE conversions rely on limited moisture–density samples and assumed solid density; unknown quantities of pumice raft deposition across the Aegean seafloor were not fully quantified. - Spatial resolution: Seismic vertical resolution (2.5–15 m) may not resolve thin interbeds or small-scale heterogeneities within thick units.