The search for reliable eruptive precursors is a central challenge in volcano monitoring, essential for optimizing volcanic early-warning systems. Recent studies have shown that the Shannon entropy of seismic signals is a promising precursor, capable of forecasting imminent eruptions with high reliability. In addition, cross entropy computed between pairs of seismic stations can help pinpoint the location of an impending eruptive vent.

In this study, we analyse the behaviour of these two entropy measures for the Island of Hawai‘i from 2017 to 2025. We examine eruptions from both Mauna Loa and Kīlauea, yielding forecast lead times of 30 minutes to 24 hours. Differences in these lead times may reflect the complexity of the volcano-structural setting of the island and its underlying volcanic plumbing systems. Highly fractured areas may favour rapid magma ascent, leading to a short eruption warning. Heat maps of cross-entropy across all station pairs in the network enabled precise forecasting of the locations of forthcoming eruptive sources, except when the new vent formed outside the seismic network.

How to cite: Santos Campos, I., D'Auria, L., Álvarez-Hernández, A., Rey-Devesa, P., Ibáñez, J. M., Prudencio, J., Titos, M., and Benítez, C.: Shannon Entropy as an eruptive precursor: a practical study on Hawai‘i Island from 2017 to 2025, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-388, https://doi.org/10.5194/egusphere-egu26-388, 2026.

Kīlauea, Hawaii, one of the world's most active volcanoes, has experienced numerous (>40) eruptive episodes since December 2024 with remarkable lava fountain heights (up to 450m) in Halemaʻumaʻu crater. Following a dike intrusion within the Halemaʻumaʻu crater in December 2024, the eruption entered a stable pressurization and release pattern from January 2025 onwards with lava flows during the sequence confined to Halemaʻumaʻu crater.

We study the 2024-26 sequence focusing on relative seismic velocity changes (dv/v). We use ~20 seismic stations located within 10 km of the Halemaʻumaʻu crater and process the data using the traditional cross-station and less conventional single-station approach and estimate the dv/v using the wavelet approach. The dv/v patterns highlight at least three distinct phases of activity during the 2024-26 eruption sequence, as well as some interesting velocity decreases prior to the onset of the sequence in December 2024 although these are spatially confined. 

We inspect the differences between our new results with previous seismic velocity patterns (2015-2024) and explore the nature of the changes using complementary observations (seismic and geodetic data), as well as numerical modeling. Our study suggests a change in strain patterns at the shallow Halemaʻumaʻu reservoir which implies a dynamic evolution of the magmatic system feeding the eruption. Additionally, we show how deformation (deflation) of the deeper South Caldera reservoir contributes to the observed dv/v patterns. Our study sheds light on the dynamics between different magma reservoirs and links to surface processes.

How to cite: Caudron, C., Reiss, M. C., Bennington, N., Hotovec-Ellis, A., Rohnacher, A., Lanza, F., Sens-Schönfelder, C., Jolly, A. D., Roman, D., Wauthier, C., Lo, A. W. K., Anderson, K., and Flinders, A.: Seismic velocity changes during the 2024-26 fountaining sequence at Kīlauea, Hawaiʻi, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-17603, https://doi.org/10.5194/egusphere-egu26-17603, 2026.

The ongoing (2024-) eruption sequence at Kīlauea Volcano, Hawai’i, has comprised 40 (as of mid-January, 2026) episodes of high fire fountaining, as well as episodes of dome fountaining, effusive eruption, and gas pistoning (series of cyclical overflows and drain-back of lava). While many episodes of gas pistoning are visually apparent in webcam footage, seismic characterization of gas pistoning and VLP swarms allows for an objective analysis of pistoning frequency, amplitude, and duration, providing greater insight into the gas ascent and escape process and its relationship to fire fountaining episodes. We thus analyze continuous RSAM to quantify the timing, duration, frequency, and amplitude of gas pistoning throughout the 2024-2026 eruption sequence and its relationship to fire fountain heights and durations.

Gas pistoning was first observed in March of 2025 as an immediate precursor to fire fountaining, and post-March fire fountains have generally been preceded by gas pistoning. The onset of precursory gas pistoning corresponds to a shift to shorter-lived fire fountains, suggesting that precursory gas pistoning contributes to the duration of fire fountain episodes. However, beginning in late May, gas pistoning became more decoupled from high fountaining, with a marked delay between the end of gas pistoning and the high fountain onset. The fountains that follow a delay after pistoning are among the highest in the eruption sequence, suggesting that a short-term sealing of gas pathways contributes to greater fountain heights. The most recent (November 2025 onwards) episodes of fire fountaining have also been followed by swarms of repeating VLP events or by additional gas-pistoning tremor, suggesting ongoing gas escape following the end of high fire fountaining. Overall, seismic observations indicating increased precursory and post-fountaining degassing suggest increasing degrees of vertical connectivity in Kīlauea’s magma transport system.

How to cite: Roman, D., Reiss, M., and Caudron, C.: Seismic characterization of pre- and post-fountaining phenomena during the 2024-2026 Kīlauea eruption sequence, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-14959, https://doi.org/10.5194/egusphere-egu26-14959, 2026.

In 2014-15, the subglacial Bárðarbunga caldera collapsed, subsiding 65 metres as magma flowed out from beneath it to feed a fissure eruption at Holuhraun. Subsequently, the caldera has been re-inflating, indicating recharge of the crustal magma reservoir. Sustained seismicity along the caldera ring faults – but with reversed focal mechanism polarity compared to the eruption period – further supports its ongoing resurgence. In summer 2021, 2024 and 2025 we installed temporary broadband seismic arrays on the ice cap above Bárðarbunga, to provide improved constraints on earthquake hypocentres and focal mechanisms.

We use QuakeMigrate to produce catalogues of microseismicity, with 8,500 and 19,500 events located in the campaigns in 2021 and 2024, respectively. The magnitude of completeness, M_C is ~ -1. Relative relocation reveals a sharply defined ring fault, consistent in geometry with geodetic constraints obtained during the 2014-15 collapse, thus providing strong evidence that the same structure is being reactivated as the caldera re-inflates. Tightly constrained focal mechanisms show excellent agreement with the local ring-fault geometry defined by the relocated microseismicity, and steep dip-slip faulting corresponding to uplift of the caldera floor. Low frequency earthquakes observed between 15 - 25 km depth in the normally ductile part of the crust below Bárðarbunga, and at around 6 km depth below the caldera, signify activity in the deeper plumbing system of the volcano, which may indicate magma ascent pathways. These events contribute to excellent ray coverage for tomography, which we will use to image the shallow melt reservoir and its geometry relative to the ring-fault.

How to cite: Winder, T., Heimisson, E. R., Rawlinson, N., Brandsdóttir, B., Jónsdóttir, K., and White, R. S.: High-Resolution Microseismicity Provides Insights into Ring-Fault Geometry at the Re-inflating Bárðarbunga Caldera, Iceland, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-1110, https://doi.org/10.5194/egusphere-egu26-1110, 2026.

Monitoring seismic velocity variations in the shallow crust using noise-based techniques has emerged as an effective approach for tracking temporal changes in the local stress field. Despite their potential, real-time implementations of these methods have so far been mostly restricted to volcanic areas and are operational at only a limited number of volcano observatories due to the relative simplicity and speed of application in those contexts. In this work, we present the first real-time monitoring of seismic velocity variations applied to a tectonic seismic swarm in the Aegean Sea, which initiated on January 31, 2025. We introduce objective and rapid procedures to identify the key parameters necessary for the analysis and to compute the probability that new observations belong to the same statistical distribution as the background, allowing us to highlight potential anomalies as they occur. Finally, we interpret the processes driving the Aegean seismic swarm and suggest the presence of distinct recovery patterns in relative velocity variations following the abrupt drops typically observed during co-seismic effects.

How to cite: Mandler, E., Zaccarelli, L., Faenza, L., and Melis, N.: Real-time monitoring of seismic velocity variations during the 2025 Aegean Sea seismic swarm, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-19911, https://doi.org/10.5194/egusphere-egu26-19911, 2026.

Taupō volcano is a frequently active rhyolitic caldera volcano in the central North Island of Aotearoa New Zealand that was the site of Earth’s most recent supereruption (Ōruanui, ∼25.5 ka), as well as one of the most violent eruptions globally of the last 5000 years (Taupō, 232±10 CE). Taupō has erupted 28 times since the Ōruanui event and displays unrest activity (seismicity and surface deformation) on roughly decadal timescales. In 2022–23, Taupō volcano underwent a period of unrest with elevated levels of earthquakes and ground deformation, including a M 5.7 earthquake that caused a tsunami within Lake Taupō. This elevated activity resulted in the Volcanic Alert Level for Taupō being raised to Level 1 for the first time. Here, we present results from a detailed characterisation of the activity beneath Taupō throughout the year-long unrest episode including a catalogue of earthquake locations; relative relocations; magnitudes; and focal mechanisms. We focus particularly on the detail in the catalogue that reveal the processes, state and structure of the modern magma reservoir beneath Taupō and builds our ability to interpret future unrest and possible eruption at the volcano. 

How to cite: Mestel, E., Illsley-Kemp, F., Savage, M., Wilson, C., and Hreinsdóttir, S.: Characterisation of the 2022–23 unrest episode at Taupō volcano, Aotearoa New Zealand, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-14560, https://doi.org/10.5194/egusphere-egu26-14560, 2026.

Seismograms recorded near active volcanoes contain numerous volcanic earthquakes and tremors that capture signatures of diverse volcanic processes and offer insights into the state of the volcano’s plumbing system and its underlying physical mechanisms. However, the strong variability of the seismo-volcanic signals makes their interpretation in terms of associated physical processes difficult. To address this, we employ unsupervised machine learning techniques, specifically, the scattering transform and Uniform Manifold Approximation and Projection (UMAP), to extract statistically significant features from continuous seismograms and to identify meaningful patterns related to volcanic activity. This approach eliminates the need for discrete event catalogs, enabling a comprehensive analysis of seismic manifestations of the volcanic activity.

Our study focuses on Piton de la Fournaise (PdF), a highly active basaltic volcano on La Réunion island, France, which erupted 25 times between 2014 and 2024. As one of the world’s best-monitored volcanoes, PdF represents an ideal natural laboratory for testing and refining our methodology. We analyze three-component seismograms from multiple stations and validate our findings using complementary datasets, including eruption, earthquake, and tremor catalogs.

The two-dimensional UMAP representation of the analyzed seismic data reveals distinct patterns that correlate with volcanic activity. The resulting seismogram atlas shows isolated clusters of points forming continuous features, which correspond to co-eruptive tremors. During non-eruptive periods, the analyzed time windows accumulate in a dense point cloud. Within this cloud, a predominantly random distribution of points is evident. However, some points form nearly linear, continuous pathways within the cloud, correlating with periods of magmatic intrusions. Adjacent to the dense point cloud, pre-eruptive seismic swarms are grouped in a specific region of the UMAP space, suggesting a common underlying mechanism.

How to cite: Gärtner, M. A., Campillo, M., and Shapiro, N.: Unsupervised Machine Learning for Analyzing Continuous Seismic Recordings: Insights from Piton de la Fournaise Volcano, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-7083, https://doi.org/10.5194/egusphere-egu26-7083, 2026.

Imaging volcanic interiors is of paramount importance for understanding volcano-seismic signals and their underlying sources. However, determining fine scale structure in highly heterogeneous media is a significant challenge using traditional imaging approaches. Furthermore, modelling and inversion tools often employ cumbersome and lengthy procedures, which can be slow to implement, especially during volcanic crises when results are needed swiftly as large data volumes  arrive at Volcano Observatories. Machine-learning (ML) methods, which have experienced rapid growth over the last decade, have strong potential to address this challenge due to their suitability for complementing physics-based numerical simulations and inversion. In particular, we examine the feasibility of imaging small-scale heterogeneities beneath volcanoes, such as propagating individual dykes, directly from seismic data using rapid ML-based imaging.

Here we build on previous work where a large suite (> 5000) of seismic earthquake gathers (i.e. seismic records from individual earthquakes) derived from numerical simulations in highly heterogeneous 2D velocity models, were used to train a Fourier Neural Operator (FNO). Subsequently that FNO was used to invert for complex structure in previously unseen geologically realistic 2D models. As the training procedure is extremely computationally expensive, and is likely prohibitive in 3D, here we ask: “can meaningful information be retrieved from seismic data derived from 3D simulations, based on an FNO that was trained only on 2D seismic data”?  We see the answer to this question as important, as it helps determine the nature of the FNO training required in order to apply this new methodology beyond the numerical domain into the 3D physical world.

We build 3D models that are consistent with the 2D models used for machine learning training. Seismic data are generated from these models, and we evaluate how well a 2D pre-trained algorithm can recover geological structures and velocity characteristics from the 3D data.

How to cite: Evans, C., Lokmer, I., Bean, C., and Totten, E.: Direct seismic data inversion for volcanoes using machine learning: a comparison of 2D and 3D cases, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-10295, https://doi.org/10.5194/egusphere-egu26-10295, 2026.

This research employs a Random Forest machine learning method to forecast eruption probability for three Indonesian volcanoes: Semeru, Lewotobi Laki-laki, and Ruang. The primary objectives are to (1) evaluate model performance under varying data quality conditions and (2) test the transferability of forecasting models between different volcanic systems. We compare three scenarios: Semeru's major eruptions (December 2020 and 2021) with significant data gaps, Lewotobi Laki-laki's seven major eruptions (March - August 2025) with high data completeness, and Ruang's eruption (April 2024).

Seismic data from the vertical component (Z) were processed using Real-time Seismic Amplitude Measurement (RSAM), Displacement Seismic Amplitude Ratio (DSAR), and MSNoise to monitor seismic amplitude variations and relative velocity changes. Statistical methods extracted an initial set of 768 features from these processed signals. After removing highly correlated features, the top 20 most relevant features were selected for model training.

For Semeru, a model trained on the 2020 eruption successfully forecasted the 2021 eruption, with forecast probability exceeding the 0.7 threshold 12 hours prior to the eruption. For Lewotobi Laki-laki, models trained on earlier eruptions (March-April 2025) successfully forecasted subsequent event in May 2025, achieving lead times ranging from 6 hours to 1 day. Cross-volcano testing revealed that the Semeru-trained model failed to forecast the Ruang eruption, likely due to data incompleteness. In contrast, the Lewotobi Laki-laki model successfully forecasted the Ruang eruption 6 hours in advance, demonstrating successful model transferability. These results highlight the critical importance of data completeness for developing robust, transferable eruption forecasting systems.

How to cite: Martanto, M., Caudron, C., Lecocq, T., Syahbana, D. K., and Nugraha, A. D.: Eruption Forecasting Using Random Forest on Single-Component Seismic Data: Insights from Three Indonesian Volcanoes (Semeru, Lewotobi Laki-laki, and Ruang), EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-16967, https://doi.org/10.5194/egusphere-egu26-16967, 2026.

Monitoring volcanic unrest at complex volcanoes such as Mt. Etna remains challenging due to the coexistence of diverse seismic sources, including volcano-tectonic (VT) earthquakes, sustained tremor and strong scattering in heterogeneous structures. Traditionally, such processes are investigated using translational seismometers alone, potentially limiting the characterization of the underlying wavefield and its physical interpretation.

In this study, we explore the added value of combining translational, rotational, and distributed dynamic strain sensing (DDSS) observations to investigate seismic activity during the December 2025/ January 2026 eruptive activity of Mt. Etna. We analyze six-component ground-motion recordings from a rotational sensor co-located with a conventional seismometer, complemented by DDSS measurements along a nearby fiber-optic cable. Using root-mean-square (RMS) amplitude analyses we examine the temporal evolution of seismic energy associated with tremor and VT activity across the different sensing modalities.

Preliminary results indicate that rotational and translational measurements capture complementary aspects of the volcanic wavefield, with rotational data emphasizing continuous, wavefield-dominated energy components, while translational recordings highlight both impulsive and sustained signals. DDSS observations provide dense spatial sampling, offering additional constraints on signal coherence, propagation characteristics, and source localization. When analyzed jointly, these datasets reveal a more coherent and interpretable picture of volcanic unrest than any single sensor type alone.

Our observations suggest that multi-sensor seismic monitoring, integrating translational, rotational, and DDSS measurements, is particularly advantageous in complex volcanic environments where scattering, anisotropy, and mixed source processes complicate traditional analyses. This work highlights the potential of such integrated approaches for improving the detection, characterization, and interpretation of volcanic seismicity and motivates their broader application in future volcano monitoring strategies.

How to cite: Izgi, G., Currenti, G., Eibl, E. P. S., Vollmer, D., Pellegrino, D., Pulvirenti, M., Alparone, S., Larocca, G., and Jousset, P.: More Than Shaking: What Rotations and Strain Reveal About Volcanic Unrest at Mt. Etna, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-10412, https://doi.org/10.5194/egusphere-egu26-10412, 2026.

This study consists of one branch of an ongoing push to understand, locate, and leverage distributed local events on volcanoes with the dual goal of segmenting volcanic and non-volcanic activity and directly imaging the shallow volcanic edifice. Typically, locating small events on volcanoes is particularly difficult due to their emergent appearance and lack of discernable ballistic waves, precluding any metrics related to travel time. Furthermore, edifice imaging through passive approaches is generally limited due to a lack of stable frequency information above ~1 Hz, truncating surface wave sensitivity to mid-crustal scales. Here, we show a two-pronged approach to tackling these problems: 1) The scattering structure of the volcano is studied in detail using active sources and lava lake eruptions coupled with Monte Carlo Radiative Transfer simulations, permitting a full understanding of seismogram envelopes for a given source location. 2) Coda correlations of distributed icequakes and eruptions have been shown to yield very high-quality Green’s functions at frequencies up to 10 Hz. Beyond cutting scatter-based imaging, these can be used to greatly extend the frequency range of dispersion curves, and thus yield valuable upper edifice information that can be coupled with matrix-based scattering imaging effort.

How to cite: Chaput, J., Galvan, I., Reisinger, R., Aster, R., and Grapenthin, R.: Towards detecting, classifying, locating and leveraging distributed events in strongly scattering media, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-22899, https://doi.org/10.5194/egusphere-egu26-22899, 2026.

Understanding geodynamic processes between Tenerife and Gran Canaria is essential for assessing seismic and volcanic hazards in the Canary Islands. The GUANCHE project characterizes seismicity, subsurface structure, and related phenomena through an integrated approach, combining land-based and ocean-bottom seismic networks.

The land-based campaign ran from April 2023 to December 2024, with 13 temporary stations across Gran Canaria transmitting real-time data to the IGN analysis center. Three high-quality sites were upgraded to permanent IGN stations after the campaign, ensuring continued seismic monitoring of the area. The marine component, using ocean-bottom seismometers (OBS), was deployed in January 2024, with data collected in June 2024. Observations from these temporary networks were integrated with the existing permanent network in Tenerife and Gran Canaria to provide a comprehensive dataset for seismic investigations.

This integrated network improves detection and localization of low-magnitude seismic events. A 3D velocity model derived from project data was applied to refine earthquake locations, providing the basis for clustering, which reveals distinct seismogenic zones and a complex pattern of activity at multiple depths. Focal mechanisms were determined for the largest earthquakes (Mw > 3.5) using TMS inversion, offering additional constraints on active faulting and regional stress.

These results highlight the value of integrated seismic monitoring for understanding seismicity patterns and geodynamic processes. This study is a collaborative effort between the Instituto Geográfico Nacional (IGN) and the Instituto de Ciencias del Mar (ICM), combining expertise in seismic monitoring and marine seismici

How to cite: Domínguez Cerdeña, I. F., Villaseñor, A., del Fresno, C., Bartolomé, R., Suárez, E. D., Barco, J., Alonso, E., Pérez-Frías, F. M., Martínez, I., Carrasco, J. M., Gómez-Liste, B., Rechcigyer, V. P., Manzanedo, M. V., Pereda de Pablo, J., and Martín Silván, A.: Integrated Seismic Monitoring of Tenerife and Gran Canaria: Insights from OBS and Land-Based Networks, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-13331, https://doi.org/10.5194/egusphere-egu26-13331, 2026.

In 2021, an eruption began in the Geldingadalir valley in southwest Iceland, lasting six months. This eruption exhibited a constantly changing eruption dynamic recorded as volcanic tremor of varying duration and amplitude. In May 2021 a transition occurred, from continuous tremor in the early phase of the eruption, to minute-long tremor episodes. Throughout this period the vent featured an active lava lake. On 2 July, the lava lake drained and several ash-rich plumes rose from the crater between 3:00 and 5:00 am. The plumes were accompanied by several transient seismic and acoustic signals. Following these events, the volcanic tremor shifted from minute-long to hour-long episodes. 

Here, we use a multidisciplinary approach combining video footage with seismic and acoustic data to investigate the source process and its potential link to the observed tremor transition. We performed a source inversion of the seismically strongest event using seismometers within 6-8 km distance from the active crater. We tested different source models and compared the simulated waveforms to those that were observed to constrain the source. In addition, we calculated the Volcanic Acoustic–Seismic Ratio (VASR) using seismic and acoustic tremor recordings. The VASR reveals a decrease over time. The local webcam footage provides an insight into surface processes including inner crater collapses preceding several ash-rich plumes. This observation suggests a potential link between shallow collapses, plume generation and seismic and acoustic signals. These collapses may have modified the shallow conduit and caused the transition from minute-long to hour-long episodes.

How to cite: Joachim, A., Heimann, S., Lamb, O. D., Eibl, E. P. S., Barnie, T., Gudnason, E. Á., Ágústsdóttir, T., Thordason, T., Hersir, G. P., Winder, T., Rawlinson, N., Fischer, T., Doubravová, J., and Burjánek, J.: Seismic, acoustic, and visual observations of ash-rich plume events during the Geldingadalir eruption, Iceland, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-10750, https://doi.org/10.5194/egusphere-egu26-10750, 2026.

Unrest has been ongoing on the Reykjanes Peninsula, Iceland, since 2019, with inflation in the Fagradalsfjall and Eldvörp-Svartsengi volcanic centres resulting in a series of volcanic eruptions beginning in 2021. We have operated a permanent broadband seismometer network on the peninsula since June 2020, complemented by networks run by several other groups. Recently, these were supplemented by 24 three-component nodes for two months starting in September 2025, which provided improved coverage in the western part of the peninsula, and further enhanced both the spatial footprint and density of the combined arrays.

Using this new dataset, and taking advantage of a period of relative volcanic and seismic quiescence, a new 3D shear wave velocity model for the peninsula is constructed from inter-station surface wave dispersion curves extracted from ambient seismic noise cross-correlations. The dense node deployment also allows analysis of shallow crustal anisotropy, thus helping to pinpoint magmatic storage regions and areas of shallow fractures. The final model spans the shallow crust from the surface to 8 km depth, with lateral model resolution approaching 1 km above the brittle-ductile transition. This allows imaging of the Reykjanes, Fagradalsfjall and Eldvörp-Svartsengi volcanic systems, as well as of geothermal fields on the peninsula.

How to cite: Pizii, W., Rawlinson, N., Winder, T., White, R. S., Brandsdóttir, B., Ágústsdóttir, T., Doubravová, J., and Burjánek, J.: New images of volcanic systems on the Reykjanes Peninsula, Iceland, from ambient noise tomography using a regional node array , EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-11833, https://doi.org/10.5194/egusphere-egu26-11833, 2026.

Piton de la Fournaise volcano on La Réunion island is a shield volcano that showed annual eruptive behavior since 2014. Following the last eruption of this eruptive cycle in 2023, only seismic crises were detected, but no eruption occurred until the time of writing, making the analysis of the 2023 eruption especially important. The 2023 eruption of Piton de la Fournaise volcano began on 2 July with two fissure openings on the northeastern flank, followed by a third eruptive vent on the southeastern flank that lasted until 10 August. We analyze data from a temporary seismic array on the western flank within Enclos Fouqué Caldera and the permanent network from the Volcanological Observatory of Piton de la Fournaise (OVPF-IPGP) to investigate the eruption dynamics.

The tremor frequency range varies slightly between the three fissure activity periods but is mostly concentrated between 0.8 and 4 Hz. Tremor amplitude and GNSS measurements at the summit crater show similar changes for the start and towards the end of the eruption as previously observed at the volcano. While the network analysis provides highly accurate locations for the three distinct fissures, we only obtain well fitting back azimuths (BAz) for specific times from the seismic array. Slowness results from the array however help distinguish the tremor signal into surface and body waves, and for certain phases even indicate the existence of two distinct tremor sources.

The deviating array back azimuths that are observed for the surface waves are interpreted to be related to the medium heterogeneity within the crater region including topographic effects. Our preliminary results, combining two different methods allow the determination of two tremor signals for one fissure site that exhibit different frequency ranges and amplitudes and possibly originate from both subsurface and surface sources. We assume that surface activity is dominating the analysis, but once decreased, a weaker tremor signal at depths becomes visible.

How to cite: Ohrnberger, M., Vesely, N. I. K., Eibl, E. P. S., Journeau, C., Duputel, Z., Vollmer, D., Brunet, C., Lauret, F., and Ferrazzini, V.: First insights into the 2023 Piton de la Fournaise eruption: Revealing two distinct tremor signals, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-10978, https://doi.org/10.5194/egusphere-egu26-10978, 2026.

Seismic tremor is widely monitored for eruption forecasting, yet its use requires improved understanding of its source processes, which remain debated. Tremor is commonly attributed to magma transport or fluid-induced resonance within volcanic plumbing systems. However, alternative studies suggest that fluids may not be required: weak, unconsolidated edifice materials geomechanically near the brittle–ductile transition can undergo diffusive brittle failure at room temperature, producing numerous low-amplitude, small-stress-drop seismic events that merge into tremor. Minor stress perturbations—caused by magma flow, gas influx, or gravitational loading—may be sufficient to trigger such dry mechanical failure.

Here, we investigate episodic high-frequency tremor (10–20 Hz) recorded at the summit of Mt. Etna during a dense seismo-acoustic deployment in summer 2022. Despite strong attenuation and scattering at these frequencies, we show that variations in the seismo-acoustic energy ratio across tremor episodes reveal differing conditions under which tremor is produced. Using multi-array beamforming and 3D grid-search techniques, we locate tremor sources in multiple regions, including both degassing-related and non-degassing areas. Synthetic tests indicate that some tremor episodes likely comprise multiple simultaneous sources, consistent with diffusive brittle failure. Frequency–magnitude analyses further support a model in which tremor arises from sequences of small-magnitude, very low stress-drop events merging into tremor due to the cumulative scaling observed and comparison with previous numerical work on seismic event population. Together, our results indicate that volcanic tremor does not necessarily require fluid movement and may also be generated by dry brittle failure processes.

How to cite: Weber, M., Bean, C., Tary, J. B., Soubestre, J., Lokmer, I., De Angelis, S., Zuccarello, L., and Smith, P.: Implications for existing tremor generation models on volcanoes through newly observed high-frequency tremor on Mt. Etna, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-13585, https://doi.org/10.5194/egusphere-egu26-13585, 2026.

Continental volcanic arcs are driven by melting in the mantle wedge between a subducting oceanic plate and overriding continental plate. These melts are emplaced within the continental crust where they then fractionate and evolve, producing silicic volcanic rocks. From observations of these systems globally, we understood them to have magmatic plumbing networks organized into transcrustal systems, and while the geometries of these systems are somewhat constrained, the links between magma storage regions, melt migration and surface unrest remains poorly understood. At Laguna Del Maule (LdM) in central Chile, where volcanic unrest is being currently monitored, we use two densely deployed seismic datasets of nodal and broadband seismometers to study these connections in detail. Here, we present interpretations using well constrained earthquake locations and high resolution crustal-scale seismic imaging using Receiver Functions (RFs), to infer the magma plumbing network and interconnectivity within this modern arc volcano setting.

Over 4300 events were detected within the periods between 2015-2018 and 2022-2024 and are divided into shallow and deep groups. Shallow seismicity is separated into clusters consistent with prior observations that link fault activity to shallow magma intrusion. Those events occurring within the deep crust (~12-30 km) are a new observation, containing a mixture of high and low frequency earthquakes. Through Frequency Index Analysis, we classify those deep events with low frequencies as Deep Long Period earthquakes (DLP). These have been observed in other volcanic arcs, but this data contains the first evidence of DLP seismicity within the Andes. The deep higher frequency events are provided in a pronounced one-day swarm of activity in 2018, all with similar magnitude and frequency index. The swarm has a vertical extent between ~21-26 km depth, and we interpret this activity to be a Volcano-Tectonic swarm (VT) related to magma migration within the middle-lower crust. In the RF images, the VT swarm is located between the top of a low velocity zone (LVZ) in the lower crust, and the base of an upper crust LVZ. The lower crust LVZ likely represents an area of deep magma storage that intermittently incubates the upper crust system with batches of basic magmas. RF images of the upper crust LVZ are consistent with prior geophysical estimates of the geometry and approximate spatial extent of LdMs shallow magma chamber.

Three months following the deep VT swarm, vertical surface uplift in the local GPS record accelerates. We therefore infer that the VT swarm was driven by the delivery of a new batch of magma from lower to upper crustal magma reservoirs. This applied additional pressure to the base of the upper crustal reservoir, leading to a surficial response in a lag-time consistent with the systems hydraulic diffusivity (~20 m²/s). Since this inflation rate has been maintained at least until 2020, the VT swarm may represent the establishment of a new preferred magma ascent path. These results indicate that volcanic unrest is preceded months in advance by seismic activity occurring within the middle-lower crust, applying bottom-up reservoir pressurization in arc volcanoes. 

How to cite: Bradford, J., Mahanti, S., Kiser, E., Beck, S., Fernandez, M., Porter, R., Maharaj, A., Howe, H., Ortiz, G., and Saez, M.: Magma Migration at Laguna Del Maule, Chile, Using Well Constrained Seismicity and Receiver Function Imaging, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-15346, https://doi.org/10.5194/egusphere-egu26-15346, 2026.

The Eifel region (southwest Germany) is an intracontinental volcanic field of distributed explosive centers, with the last major eruption occurring at the actively degassing Laacher See volcano in 13ka. The temporary Eifel large-N seismic network was deployed here between September 2022 and August 2023, which significantly increased station coverage allowing for a detailed study of the subsurface magmatic system. We use local earthquakes to study the attenuation properties of  the subsurface structure, targeting Laacher See volcano and its surroundings. 

Starting with an automatically derived earthquake catalog, we first perform a number of quality checks to ensure we only use the cleanest waveform traces and picks. Then, we estimate scattering from peak-delay measurements, i.e. the delay of the maximum energy after the S-wave. Our preliminary findings suggest the presence of two upper crustal structures beneath Laacher See, outlined by areas of high scattering, which are typically associated with the presence of small-scale heterogeneities or mechanical discontinuities (e.g. fractures, faults) and range from the surface to 8 and 13 km below sea level. Their proximity to Laacher See suggests that the rise of fluids may be facilitated by these structures. Further analysis will have to show whether the data can also be used to quantify absorption, which would shed even more light on fluid pathways.

How to cite: Bramwell, L., Reiss, M., and De Siena, L.: Preliminary imaging of the Eifel Volcanic Field from seismic scattering , EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-18644, https://doi.org/10.5194/egusphere-egu26-18644, 2026.

The Tatun Volcano Group (TVG), adjacent to the densely populated Taipei metropolitan region in northern Taiwan, is an active volcanic system where a large number of earthquakes have been observed. Although the previous studies have reported that volcanic fluids exist beneath the surface, how these fluids change over time and influence local earthquakes has remained unclear. In this study, we examined more than 12,000 earthquakes recorded between 2014 and 2021 to explore how the behavior of earthquakes and the physical properties of the seismogenic zone vary with time. By analyzing patterns in frequency-magnitude distribution of earthquakes and seismic wave velocities within the seismogenic zone, we found that the triggering mechanisms for earthquakes in the TVG shift over time, possibly due to the varying influence of volcanic gases, hydrothermal waters, and stress. This study deciphers the dynamic nature of the TVG and improves our understanding of the volcanic risk near the Taipei metropolis.

How to cite: Pu, H.-C., Lin, C.-H., Lai, Y.-C., and Shih, M.-H.: Temporal Variations in Earthquake Triggering Mechanisms in the Tatun Volcano Group, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-2705, https://doi.org/10.5194/egusphere-egu26-2705, 2026.

Strokkur geyser in the Haukadalir valley in south Iceland is an erupting hot spring that allows studying hydrothermal processes. Since March 2020, we have continuously monitored Strokkur using three seismometers located ~40 m from the conduit. This long-term dataset has enabled the creation of a high‑resolution catalogue containing more than 760,000 individual water‑fountain events, which has previously been used to investigate eruption types, driving mechanisms, and the influence of air temperature and wind on geyser dynamics.

In this contribution, we present a striking change in Strokkur’s behaviour that occurred on 18 October 2024 at 18:00. Following this moment, the geyser began producing a larger number of water fountains per eruption, more water fountains per hour and exhibited a markedly shorter recharge cycle. Simultaneously, several neighbouring hot springs activated or increased their activity. Because the onset of this transition was captured seismically, the dataset offers a rare opportunity to examine the triggering mechanism and its implications for subsurface fluid pathways.

By analysing the spatio‑temporal evolution of seismic signals associated with this behavioural shift, we explore the underlying processes driving the system’s reorganisation. The study highlights the value of dense seismic monitoring and detailed event catalogues for understanding hydrothermal dynamics, and it provides insights into geothermal systems and their time‑dependent changes.

How to cite: Eibl, E. P. S., Izgi, G., Walter, T. R., Heimann, S., Hersir, G. P., Guðnason, K. J., and Kristjánsson, V.: A high‑resolution water fountain catalogue reveals an abrupt hydrothermal change at Strokkur geyser, Iceland, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-10564, https://doi.org/10.5194/egusphere-egu26-10564, 2026.

In 2010, the station IS42 was the first infrasound station to be installed in the Azores, located on Graciosa Island in the central group of the Azores archipelago, in the middle of the North Atlantic. This station integrates the International Monitoring System (IMS) of the Preparatory Commission for the Comprehensive Nuclear-Test-Ban Treaty Organization (CTBTO). Although the mission of the CTBTO is to put an end to nuclear tests, the long-term infrasound data recorded by IMS have proven to be very valuable for monitoring and understanding natural phenomena, including seismo-volcanic activity in the Azores.

With the aim of monitoring the 2022 seismo-volcanic crisis on São Jorge Island, a first portable infrasound array was deployed to complement the data recorded from the permanent IMS station and enrich the archipelago’s monitoring network. A second portable array was subsequently deployed on Terceira Island, and an additional array is planned for deployment on Faial Island later this year, further strengthening the Azores infrasound monitoring network.

This study analyses the performance and behaviour of the portable arrays, using IS42 as a reference station. We applied a multi-channel correlation analysis in the time domain to evaluate the influence of background noise on the recorded signals and assess the impact of station location and environmental conditions on the detections. Root-mean-square (RMS) noise analyses were combined with source direction estimates based on the detections’ back azimuths. Seasonal analyses of the detections revealed a strong influence of atmospheric conditions on noise levels and, consequently, on back azimuth directions. These results highlight the importance of noise characterisation of integrated infrasound observations in oceanic islands.

How to cite: Silva, L. I., Matos, S., Marchetti, E., and Wallenstein, N.: Azores Infrasound Network: Analysis of background noise, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-13765, https://doi.org/10.5194/egusphere-egu26-13765, 2026.

Infrasound has been widely applied to the remote monitoring of explosive volcanic eruptions. Although no active volcanoes are currently present within South Korea, explosive activity in neighboring regions can still generate transboundary hazards that require effective remote monitoring. In this context, we present a quantitative assessment of volcanic eruption detectability using the Korea Infrasound Network (KIN).

The KIN has been operated for more than two decades and was originally established to monitor local and regional acoustic sources and to discriminate between natural and anthropogenic signals. The network consists of Chaparral M2 infrasound sensors, each of which has a flat response from 0.1 to 200 Hz. Since 2011, eight infrasound arrays with apertures ranging from 0.15 to 1.68 km have been fully operational. We evaluate the detectability of eruptions with VEI ≥ 3 that have occurred since 2011, examining detection characteristics as a function of distance, azimuth, and atmospheric propagation conditions. Detection was performed using the Progressive Multi-Channel Correlation (PMCC) algorithm to identify coherent infrasound signals.

Many eruption signals recorded by the KIN extend into frequencies below the nominal flat-response bandwidth and are often obscured by persistent microbarom noise. Despite these limitations, volcanic eruptions were conditionally detected depending on eruption size and atmospheric propagation conditions. The analyzed cases include the 2022 Hunga Tonga–Hunga Haʻapai eruption (VEI 5), the 2020 Taal and 2021 Fukutoku-Oka-no-Ba eruptions (VEI 4), and several VEI 3 eruptions such as Asosan, Kirishimayama, and Raikoke.

Our results indicate that automated eruption detection using KIN is feasible, particularly at the TJIAR array in central South Korea. A long-term PMCC detection catalog spanning approximately 15 years (since 2011) was compiled for TJIAR and compared with independent eruption records from the Tokyo Volcano Ash Advisory Center and the Global Volcanism Program to assess detection reliability. This study represents the first long-term assessment of volcanic infrasound detectability based on the KIN. In addition, low-frequency infrasound sensors (MB3d) with an extended dynamic range were collocated at one of the KIN arrays in 2025 to improve low-frequency detectability. Ongoing work focuses on assessing improvements in eruption detectability through comparisons between legacy and upgraded sensor configurations, with implications for the development of an infrasound-based automated eruption detection and long-term monitoring of explosive volcanic eruptions in East Asia.

How to cite: Park, I. and Che, I.-Y.: Quantitative Assessment of Remote Volcanic Eruption Detectability Using the Korea Infrasound Network, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-8960, https://doi.org/10.5194/egusphere-egu26-8960, 2026.