Magmatic degassing often provides the earliest clues to changes in magmatic systems, yet detection and characterization of the escaping gases is challenging. Recent work suggests that satellite-based normalized difference vegetation index (NDVI) calculations may provide an indirect method for detecting changes in degassing. Here, we use tree ring δ13C, δ18O, and radiocarbon (14C) data to assess NDVI as a proxy for diffuse magmatic degassing and to better understand the development of the Tern Lake thermal area (TLTA) in the Yellowstone Plateau volcanic field, USA. We measured δ13C, δ18O, and 14C in wood from four TLTA trees and two control trees. The 14C data indicate that TLTA trees were exposed to significant magmatic CO2 degassing during the 1990s, corroborating satellite-based NDVI changes observed at a time when no other indications of degassing existed. Magmatic degassing also increased water availability for trees closest to the degassing center, as indicated by decreasing δ13C and δ18O values relative to background throughout the 1980s and 1990s. Increased H2S degassing in the early 2000s impaired stomatal functioning, increasing δ13C and δ18O values. These isotope values were strongly correlated with NDVI, demonstrating that ground- and satellite-based approaches yield similar conclusions. Together, the satellite and isotope datasets suggest that TLTA degassing was characterized by cold CO2 emissions with minimal H2S until the early 2000s, when it transitioned to a high-temperature system with significant CO2 and H2S emissions. These results add to a growing body of evidence that vegetation responses can be an effective proxy for magmatic degassing.
The hazard posed by volcanic ballistic projectiles (VBPs) is commonly assessed using trajectory models that estimate impact locations and energies. Recent models simulate the one-way coupling between the erupting gas–ash mixture and particles by inserting the numerically obtained flow field into the equation of particle motion. Model testing typically relies on reproducing observed travel distances in the field, but this approach is hindered by poorly constrained pre-eruptive conditions such as vent overpressure. To overcome this limitation, this study applies a trajectory model based on the steady-state supersonic jet flow field to reproduce experimental trajectory data obtained from high-speed shadowgraph imaging of spherical glass beads initially at rest and accelerated by supersonic jets from an open-ended shock tube. The model successfully reproduces the observed velocity profiles, demonstrating the viability of the one-way coupling approach for dilute particle populations accelerated by supersonic jets. The reproduced trajectories yield particle Reynolds numbers between 103 and 2 × 105. In this range, drag coefficients for both spherical and irregular-shaped particles are largely independent of particle Reynolds number and depend primarily on particle Mach number. Furthermore, an equation for the duration of steady-state supersonic flow conditions is derived, validated, and then used. The equation shows that, in many explosive eruptions, VBPs are predominantly accelerated during the period in which the supersonic jet can be approximated as steady-state. Particle acceleration is strongest near the vent and decreases rapidly with distance, implying that early jet collapse has little influence on maximum VBP velocities.
Paroxysmal eruptions at Stromboli are commonly attributed to the ascent of deep CO2-rich magma, whose early exsolution at ~200 MPa generates overpressure capable of triggering explosive activity. However, new geochemical observations from the July 2024 paroxysm challenge this paradigm. In this study, we investigate plume degassing dynamics through analysis of CO2, SO2, and H2O molar ratios and SO2 and CO2 fluxes measured by the INGV monitoring network, including an unprecedented dataset acquired during the 2024 paroxysm itself. We first demonstrate that differential atmospheric processing of CO2 and SO2 can significantly bias gas-ratio interpretations, requiring rigorous post-processing to retrieve primary magmatic signals. Reassessment of the full geochemical record indicates that, although the 2007 paroxysm is consistent with a deep CO2-driven, bottom-up mechanism, the 2019 and especially the 2024 events display contrasting precursory behaviour. The 2024 eruption was not preceded by elevated CO2 flux; instead, it was characterized by declining CO2 emissions, low CO2/SO2 ratios, and increasing SO2 flux within days of the event. These trends, coupled with reduced Strombolian explosion frequency, point to volatile accumulation and overpressurization within the shallow plumbing system (~2 km depth). Degassing was dominated by SO2 and H2O, indicating that shallow reservoir dynamics, rather than deep CO2 input, can independently trigger paroxysmal explosions. We propose that Stromboli operates under at least two distinct degassing regimes: a deep CO2-driven piston-like ascent, and a shallow, top-down overpressure mechanism capable of mobilizing deeper magma as a secondary effect. These findings demonstrate that deep CO2 flux is not an exclusive precursor of paroxysmal activity and underscore the need to integrate shallow volatile dynamics into eruption forecasting frameworks. Recognizing multiple trigger mechanisms is therefore critical for improving warning strategies and strengthening civil protection measures in communities exposed to high-energy explosive eruptions.
The eruption of submarine volcanoes is a poorly-characterised source of marine hazard, largely due to the difficulty of observing such eruptions. Pumice rafts are a potentially-hazardous product of submarine volcanic eruptions: they pose danger to shipping, wildlife and impact livelihoods of those living in small island nations. However, our understanding of pumice rafts is limited due to rafting events often going unnoticed and occurring in remote areas. In this study, we analyse the pumice raft produced in the 2019 eruption of Volcano F in the Tonga Arc, southwest Pacific. We demonstrate that raft evolution, including waterlogging, can be detected using remote sensing methods, by identifying diagnostic spectral characteristics associated with saturated pumice. Using field spectroscopy, we observe unique spectral signatures in pumice samples, from this eruption and other eruptive settings, and compare these to the satellite-derived spectral measurements of the raft. We additionally show that pumice spectra are distinct from spectra of other floating debris.These spectral signatures can be used to identify pumice from samples and remotely using satellites. Using pumice samples from the 2019 raft, collected at various distances from the vent of Volcano F, we show that samples that spent longer at sea are more eroded although their circularity remains similar regardless of collection location. We show that with increased time at sea, the raft thickness decreases, whilst its saturation and uniformity increase. Our study demonstrates that, beyond capturing the presence of pumice, spectral satellite analysis can be linked to pumice raft physical properties.We detect two previously unreported eruptive periods at Volcano F, Tofua-Kermadec Arc, using a combination of satellite remote sensing and machine learning analysis of discoloured seawater. Additionally, we use the same methods to detect and observe submarine volcanic activity at Home Reef, Tofua-Kermadec Arc.Plain Language Summary In this paper, we use a mixture of satellite imagery and sample analysis to learn more about submarine volcanic processes at two volcanoes: Volcano F and Home Reef. We can observe periods of submarine volcanic activity using satellites by looking for pumice rafts, changes in ocean colour and volcanic plumes in the atmosphere. Pumice is a type of rock that is produced during explosive eruptions and forms into a raft which can float on the ocean. In our study, we analyse spectral properties of pumice samples from the 2019 Volcano F eruption and compare them to satellite images of the pumice raft to determine how saturated the raft is. We study how the size and shapes of the rock samples change with distance from volcano. We use machine learning to scan through satellite images for 25 years at Volcano F and Home Reef for potential eruptions. We detect three previously known eruptive periods–two at Volcano F and one at Home Reef and then two new periods of discoloured seawater at Volcano F.
Cerro Machín Volcano (CMV) ranks third among the 21 preliminary geothermal areas identified in Colombia, with an estimated electrical power of 129.94 MWe (Alfaro et al., 2021), yet its shallow structure had not been resolved at high resolution. We conducted terrestrial gravity and magnetic surveys within and around the volcanic edifice: 137 gravity stations and 38,789 magnetic records in walk mode. Gravimetric reductions used a Nettleton-derived reference density of 2.65 g/cm3, magnetic processing included diurnal correction, IGRF subtraction, and reduction to the pole. Residual anomalies were isolated through upward continuation to 500 m, and two joint 2D geological–geophysical models were built along NW–SE (~5.1 km) and SW–NE (~5.2 km) profiles with GMSys-2D. The dome is 720–840 m thick, its base resting directly on the Cajamarca Metamorphic Complex, shallower than previous seismic estimates of up to 2 km. Lateral variations in density (2.4–2.55 g/cm3) and susceptibility (0.002–0.12 SI) record weathering and hydrothermal alteration restricted to discrete fault zones, so that no continuous clay cap is developed. Beneath the dome, a fractured metamorphic conduit (ρ = 2.6 g/cm3, k = 0.008 SI) is interpreted as the upflow reservoir (~0.6 km2, at least 800 m thick, truncated at the base of the models), while an outflow zone of about 2 km2 extends at least 2 km toward the Santa Marta ravine. The dome base, resolved between 1920 and 1850 m a.s.l., coincides with the reservoir top independently imaged by the seismic and magnetotelluric models of Herrera and Calò (2025), to within less than the vertical resolution of either dataset. Two methods with unrelated physical sensitivities therefore converge on the same boundary, which we interpret as the top of the upflow zone, sealed above by the unaltered dome core.
