This session welcomes contributions addressing open questions in the study and modelling of volcanic processes and associated hazards, including but not limited to field and satellite data analysis, physico-mathematical formulations of natural processes, probabilistic forecasting, data assimilation and data fusion, and the development and application of numerical methods. We particularly encourage interdisciplinary contributions that bridge traditional volcano monitoring with emerging innovations in computational science, statistical analysis, Machine Learning (ML), and Artificial Intelligence (AI).
The objectives of the session include: (i) expanding knowledge of complex volcanic processes and their spatio-temporal dynamics; (ii) advancing methods for monitoring, modelling, and forecasting of volcanic phenomena; (iii) assessing the robustness of models through validation against real case studies, analytical solutions, and laboratory experiments; (iv) quantifying uncertainty propagation through both forward (sensitivity analysis) and inverse (optimisation/calibration) modelling; and (v) exploring the potential of AI- and ML-driven techniques to integrate and process multidisciplinary datasets for improved volcanic hazard assessment, risk reduction, mitigation strategies, and decision-support applications.
Forecasting sismo-volcanic events and their evolution in time and space requires a detailed understanding of magma plumbing systems in terms of their geometry, connectivity, and physico-chemical properties.
The MPGF’s multidisciplinary approach, developed over the last decades on several active volcanoes, integrates petrochemical reconstruction of the plumbing system with detailed geochemical characterization and high-frequency monitoring of gas emissions. This framework allows us to constrain magma evolution and dynamics within a volcano’s plumbing system over a wide range of pressures, temperatures, and compositions, as well as across various timescales and eruption frequencies.
Here, we review key insights gained from active volcanic systems in the Indian Ocean (La Réunion, Mayotte, and the Comoros) that have formed in different geodynamic settings (intraplate and plate boundary) and exhibit highly contrasting eruption rates, volumes, and dynamics. Among the most significant findings, we highlight:
i) The role of lateral shifts in deep magma ascent paths relative to eruptive sites, and
ii) The coexistence of both evolved (phonolite to trachyte) and mafic (basalt to basanite) melts over a broad depth range, from the mantle to the crust.
Effective long-term monitoring is achieved by focusing on the deepest parts of the plumbing system (often located on the volcano flanks) which enables the identification and tracking of new magma inputs that may lead to lateral magma drainage at shallower levels. We emphasize the importance of detecting deep silicic and variably degassed melts—sometimes already present in the mantle and near the Moho—alongside mafic, volatile-rich melts. This approach provides a robust foundation for geochemical and petrological monitoring and for sound integration between geochemical and geophysical datasets.
How to cite: Di Muro, A.: The tight link between magma plumbing system and volcano monitoring: a contribution from the multidisciplinary petrological and geochemical framework (MPGF), EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-13974, 2026.
The Villarrica volcano in southern Chile is one of the most persistently active basaltic systems in South America, characterized by continuous open-vent degassing, sustained tremor, and episodic lava bursts. These conditions generate a complex seismic environment where traditional event-based analyses may overlook subtle changes in system behaviour. This study focuses on the period between December 2018 and September 2019, during which multiple eruptive pulses were documented by Villarica Volcano Observation Project (POVI) during the austral summer, autumn, and mid-winter, followed by a quieter interval in August and renewed activity in September. The identification and delimitation of the study period is based on long- and very long-period classifications and visual observations, but these data were not considered as analytical variables. This natural alternation between eruptive and calm phases provides an ideal framework for evaluating temporal patterns in seismic and deformation signals.
Continuous broadband seismic data at 100 Hz are segmented into 3-minute windows (18,000 samples), producing thousands of high-resolution segments per day across several stations and components. From each window, several statistical and spectral features are extracted using the tsfresh package (python), creating a high-dimensional representation of signal variability. In parallel, an eight-station GNSS network (2012–2024) provides deformation context to interpret the analysed interval within Villarrica’s broader inflation–deflation behaviour.
Unsupervised learning methods are applied to the feature space to identify latent patterns without imposing predefined classes. Preliminary results indicate that feature-based representation captures clear differences between eruptive and quiescent intervals, suggesting that changes in the seismic signal statistical structure may reflect shifts in fluid dynamics and conduit conditions. The method also reveals intermediate states that do not coincide directly with eruptive pulses, pointing to possible transitions in the underlying system.
This work presents an integrative framework linking high-frequency seismic variability, eruptive observations, and GNSS-derived deformation. The results highlight the potential of unsupervised learning to identify transitions in volcanic behaviour and to support future multiparametric monitoring strategies at Villarrica and similar open-vent systems.
How to cite: Santori, C., Potin, B., Ruiz, S., and González-Vidal, D.: A multiparametric and unsupervised-learning approach to characterize seismic and deformation variability during Villarrica’s eruptive cycle (2018–2019), EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-886, https://doi.org/10.5194/egusphere-egu26-886, 2026.
In rift settings, lateral magma propagation is commonly observed, as extensional tectonics tend to favor dyke opening perpendicular to the minimum compressive stress aligned with rift axis. Whether such intrusions propagate vertically or laterally within the crust toward eruption depends on the competition between buoyancy-driven ascent and stress-controlled fracture propagation. However, the role of the mechanical properties of the host rock, magma buoyancy, tectonic stress and surface loading in dyke propagation remain insufficiently quantified. To better understand these controlling mechanisms, a series of analogue experiments are performed by injecting a finite volume of silicone oil, analog to viscous magma, into gelatin of different compositions, analog to elastic crust subjected to surface loading and an extensional stress field. The physical properties of gelatin (density and rigidity) are measured and the shape and position of the oil crack are tracked over time using cameras. These experimental observations are further compared with stress fields computed using finite element numerical models implemented in COMSOL Multiphysics based on the experimental boundary conditions associated with applied extension and surface loading. The results indicate that, within the propagation plane, the direction of propagation consistently aligns with the direction of the maximum pressure gradient, depending on both buoyancy and the external stress field, rather than being strictly vertical as if it were entirely controlled by the buoyancy effect. The close agreement between experimentally observed trajectories and numerically derived stress-gradient paths highlights that at shallow depths, the influence of the edifice's load dominates tectonic extensional stresses at a radial distance from the volcanic summit on the order of the edifice's radius; beyond this distance, the extensional stress dominates the stress induced by the edifice's load on magma propagation. The presented findings are very important for rift volcanoes, like Nyiragongo volcano in the East African Rift, where lateral magma migration under extensional stress is potentially hazardous to the densely populated cities of Goma in the Democratic Republic of Congo (DRC) and Gisenyi in Rwanda.
How to cite: Barayagwiza, S., A. Meriaux, C., and Pinel, V.: Controls of tectonic extension and surface loading on dyke propagation: insights from analogue experiments and numerical modelling, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-5593, 2026.
Dislocation models, whether analytical or numerical, are widely used to interpret surface displacements induced by magmatic intrusions. Deep learning methods developed for interpreting InSAR data are also trained using synthetic data produced by dislocation models. While these models are well suited to describing the deformation and stresses induced by the emplacement of magma within a planar structure, they neglect buoyancy, which is the main driving mechanism for magma ascent within the crust. We use analog experiments to highlight the effect of buoyancy on dike-induced surface deformation. Finite volumes of air or silicone oil are injected into gelatin, which is characterized by elastic behavior. The fluid-filled crack rises vertically through the gelatin due to buoyancy. Its position, shape, and orientation are tracked by side cameras, and the surface displacement of the gelatin is measured simultaneously by photogrammetry and pixel shift tracking from images acquired by four synchronized cameras located at the top of the experimental setup. We compare the displacement field estimated from the dislocation model with the recorded displacement field in the laboratory. We show that while dislocation models with realistic opening distributions are able to reproduce the displacement field profile fairly accurately, they systematically underestimate vertical displacement in the near field and overestimate horizontal displacements for a buoyant ascending crack. Our study shows that buoyancy of dikes triggers upward displacement of the Earth’s surface, which is not accounted by dislocation models. We discuss in more detail the potential consequences of this bias in dislocation models for the interpretation of geodetic data in volcanic areas.
How to cite: Pinel, V. and Galland, O.: Limitations of dislocation models for quantifying surface displacement induced by the buoyancy of ascending dikes, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-5871, 2026.
Highly concentrated geogenic CO2 emissions have been reported worldwide. Although atmospheric CO2 dispersion is the most common occurrence, specific topographic and meteorological conditions can lead to surface accumulation in the form of “CO2 rivers”. Although catastrophic events such as the deadly limnic eruption of Lake Nyos in 1986 are well documented, the behavior of these CO2 rivers is not well understood. This lack of understanding poses challenges for hazard assessment and mitigation. While computational models such as computational fluid dynamics (CFD) and integral models provide analytical insights, their practical application in risk management is limited by computational cost and accuracy constraints. To address these limitations, we simulate the behavior of CO2 rivers using TWODEE, a depth-averaged numerical model that is a computationally efficient alternative for simulating dense flows. We test the model at the Syabru-Bensi Hydrothermal System (SBHS) in central Nepal, where high seismic activity and significant CO2 degassing have been observed. In the field, we measure the airborne CO2 concentration, wind velocity and direction using autonomous sensors at 0, 50, 150, and 300 cm above the ground at each measurement point, as well as surface CO2 flux using the accumulation chamber method. Our results demonstrate the robustness of the statistical approach by providing well-constrained maps of CO2 concentration in the lowest atmospheric layers over large distances from the emission source. This method can be applied to other non-volcanic and volcanic sites. Additionally, we assess the impact of the 2015 Mw 7.9 Gorkha earthquake in Nepal, which triggered additional CO2 degassing vents and changes in surface CO2 flux across the SBHS. Our work aims to improve our understanding of how dense gases disperse in the lower atmospheric layers. We are developing an operational hazard assessment tool with potential applications in real-time risk management. This tool will quantify the CO2 budget of CO2 rivers in various geodynamic contexts and estimate health hazards in volcanic and non-volcanic environments.
How to cite: Robert, M.-M., Girault, F., Carazzo, G., Bhattarai, S., Pokharel, T., Bhattarai, M., Adhikari, L. B., and Jha, M.: Dispersion of geogenic CO2 in the lower atmosphere: Statistical analysis and application to the Syabru-Bensi Hydrothermal System in the Nepal Himalaya, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-11150, 2026.
The present work investigates advanced methodologies for detecting and quantifying volcanic carbon dioxide (CO₂) emissions from degassing plumes and fumarole fields using hyperspectral data acquired by satellite and airborne sensors. Building upon well-established algorithms such as the Continuum Interpolated Band Ratio (CIBR), Matched Filter (MF) and Imaging Mapping Differential Optical Absorption Spectroscopy (IMAP-DOAS), this study introduces innovative approaches that exploit the high spectral resolution of modern hyperspectral satellite sensors. In particular, the analysis leverages data from the Italian Space Agency (ASI) satellite mission PRecursore IperSpettrale della Missione Applicativa (PRISMA) launched in 2019, and operating in the Visible and Near-InfraRed (VNIR) and Short-Wave InfraRed (SWIR) spectral ranges. The quantification of the CO₂ columnar content is achieved through the application of the Matched Filter algorithm to CO₂ absorption features in the 1900–2200 nm spectral interval. The MF approach is designed to maximize the detection of the CO₂ spectral signature while suppressing background variability associated with surface reflectance and atmospheric effects. We provide the first examples of high-resolution maps of CO₂ concentration and flux from two actively degassing, quiescent volcanoes (Campi Flegrei and Vulcano Island), hence contributing to volcanic monitoring efforts. Our results provide new insights into volcanic degassing processes and their potential implications for the regional and global carbon cycle and for the climate system.
How to cite: Sapia, E., Aiuppa, A., Buongiorno, F., and Romaniello, V.: Analysis of Volcanic CO₂ Emissions Using Next-Generation Satellite Hyperspectral Data, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-12806, 2026.
Volcanic activity can pose a serious threat to nearby populations, as continuous gas emissions remain dangerous even in the absence of eruptions. Nyiragongo and Nyamuragira volcanoes located in the East African Rift are among the largest global emitters of SO2. Given the various environmental, climatic and health impacts of SO2, studying its dispersion is important. In our study we use FALL3D model (Folch et al., 2020), an Eulerian atmospheric dispersal model that solves the advection-diffusion-sedimentation equation, combined with ensemble-based data assimilation technique to reduce uncertainties in eruption source parameters to simulate SO2 dispersion during both eruptive and passive degassing phases.
Satellite observations from TROPOMI are processed using a trained AI algorithm based on machine learning that automatically detects and quantifies volcanic SO2 emissions in near real-time filtering out non-volcanic sources (Corradino et al., 2024). The meteorological data used are from ERA5 reanalysis dataset.
Literature studies (e.g.Mingari et al., 2022) show that the inclusion of the satellite data in the model greatly improves the dispersion forecasts. Building on these results we aim to improve the dispersion forecasts of SO2 from Nyiragongo volcano and develop probabilistic hazard maps of SO2 exposure enabling an uncertainty informed assessment of potential impacts on populations and infrastructure surrounding the volcanoes. Our study will demonstrate the potential of combining observational data, numerical modeling, and ensemble-based data assimilation to improve volcanic hazard monitoring.
How to cite: Uwinema, C., Meriaux, C., Costa, A., Folch, A., Massaro, S., Corradino, C., and Mingari, L.: Analysis of the persistent gas dispersion from Nyiragongo and Nyamuragira volcanoes using numerical modeling and satellite data, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-4889, 2026.
Sulfur dioxide (SO₂) represents one of the most important volcanic gases released by magma degassing in the shallow crust. Its monitoring provides information on magma ascent rates, conduit dynamics, and eruption style and intensity, thereby supporting volcano monitoring and hazard assessment.
The TROPOMI instrument onboard Sentinel-5 Precursor, launched in 2017, is the most recent sensor which delivers daily measurements of atmospheric SO₂ column densities at an unprecedented spatial resolution of 5.5 km × 3.5 km at nadir.
The aim of the present study is to improve the current SO₂ detection capabilities by combining TROPOMI products with data from the MSG‑SEVIRI radiometer, which offers higher spatial resolution (~3 km × 3 km at nadir) and revisit times of 15 minutes, or 5 minutes in Rapid Scan mode.
To enhance SO₂ retrieval capabilities, a data-driven AI model was implemented to estimate SO₂ vertical column densities at SEVIRI spatial and temporal resolution, using TROPOMI observations as reference. In particular, a multilayer perceptron was designed and trained, consisting of two hidden layers with 128 and 64 neurons, respectively, followed by a single linear output neuron. The model was trained for up to 200 epochs and optimized by minimizing the Mean Squared Error, with an early-stopping strategy applied to prevent overfitting. Model performance was then evaluated on the test set using the Mean Absolute Error, which measures the average absolute difference between predicted and observed SO₂ Vertical Column Density (VCD) values and provides a reliable indication of the prediction accuracy.
This approach allows SEVIRI data to inherit the sensitivity of TROPOMI while preserving their native high-frequency coverage. The method substantially increases measurement density and improves spatial detail, enabling more refined and continuous monitoring of volcanic degassing.
The methodology is applied to Mount Etna (Sicily, Italy), an open‑conduit volcano characterized by persistent degassing sustained by shallow convecting magma, with typical SO₂ fluxes ranging from 500 to 5000 t/day. The satellite‑based results are quantitatively validated against measurements from ground‑based monitoring networks.
Results show that the AI-enhanced SEVIRI-based SO₂ VCDs differ from the original TROPOMI values by 5–10%, confirming the robustness and reliability of the approach.
This integrated technique offers a promising tool for rapid and robust volcanic hazard assessment, introducing improvements to current retrieval methods, and enhancing early warning capabilities for aviation safety, as well as studies of climate impacts from volcanic emissions.
How to cite: Dozzo, M.: AI-Powered Volcanic SO₂ Retrieval using MSG-SEVIRI and Sentinel-5P TROPOMI, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-1051, https://doi.org/10.5194/egusphere-egu26-1051, 2026.
The detection and monitoring of volcanic clouds are critical for hazard assessment and aviation safety. In this study, we present the application of a neural network (NN) model trained on Spinning Enhanced Visible and Infrared Imager (SEVIRI) data to detect the volcanic cloud produced during the Mount Etna eruption of 27 December 2025. The analysis focuses on evaluating the model’s ability to generalize across satellite instruments by extending its application to data acquired by the Flexible Combined Imager (FCI) onboard the Meteosat Third Generation (MTG) platform. Furthermore, a volcanic cloud quantitative analysis was conducted by applying the Volcanic Plume Removal (VPR) algorithm, using the neural network–based volcanic cloud detection as input.
The primary objective of this work is to demonstrate the cross-instrument applicability of the neural network model, highlighting its robustness and adaptability to next-generation geostationary sensors. The results show that the model effectively identifies volcanic cloud structures in both SEVIRI and FCI observations, emphasizing the potential of artificial intelligence techniques for reliable volcanic cloud detection.
The second objective of this study is to present the first volcanic cloud quantitative analysis using FCI data and to compare the results with those derived from SEVIRI observations. The results demonstrate a higher sensitivity of FCI compared to SEVIRI, which can be attributed to the advanced sensor technology and the improved spatial resolution of the instrument.
This approach represents a significant step toward the development of a near-real-time monitoring system, enabling automated detection and subsequent quantification of volcanic clouds. Such a system has significant implications for operational volcano monitoring and hazard mitigation, enabling the timely and consistent delivery of information during eruptive events.
How to cite: Naranjo, C., Guerrieri, L., Corradini, S., Picchiani, M., Merucci, L., and Stelitano, D.: Neural Network–Based Detection of the Etna Volcanic Cloud: From MSG-SEVIRI to MTG-FCI, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-12849, 2026.
Geysers provide natural laboratories for studying eruptive dynamics, analogous to those observed at volcanoes, offering a safe and accessible setting in which processes can be observed at high spatial and temporal resolution. However, despite their easy accessibility and reliable activity, there is a lack of research regarding the electrical signals they generate. In this study, we investigate the source of electrical signals recorded by a Biral Thunderstorm Detector (BTD) in close proximity to Strokkur Geyser in Iceland. We focus on the effect known as shielding, where a moving conductive object connected to the ground distorts the electric field lines, inducing a current in a conductor, in our case, the BTD antenna. To test whether this effect is the source of the recorded signals, Finite Element Method Magnetics (FEMM) models are used to model the rising fountain. The results show that the induced charge and current are dependent on the fountain height, radius, and atmospheric potential gradient. We determine the atmospheric potential gradient using an electric field mill, colocated with the BTD, and measure the fountain height using video recordings. After deriving an empirical equation from the FEMM results, we can reproduce the measured BTD signal with the model by inverting for the fountain radius. Due to the high coherence between the signals and the good agreement between observed and calculated fountain radius, we conclude that the shielding effect is mostly responsible for the electric signals measured close to a geyser. This result is a significant contribution to understanding electric signals from various natural phenomena, including lava fountain activity and discharge generation in volcanic plumes.
How to cite: Gestrich, J., Cimarelli, C., Bennett, A. J., Klein-Schiphorst, S., Capponi, A., and Poetsch, C.: Linking Fountain Dynamics to Electric Field Variations: A Study of Strokkur Geyser, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-12187, 2026.
Submarine calderas remain some of the least explored volcanic systems on the planet, even though the recent Hunga Tonga-Hunga Ha’apai event has demonstrated their capacity to generate significant geohazards, including tsunamis, damage to seafloor infrastructure, and atmospheric disturbances. Their global identification has long been limited by sparse bathymetric coverage and operational constraints. In this work, we apply a machine‑learning caldera detection algorithm (CDA) to global bathymetric datasets, enabling a systematic search for previously unrecognised submarine calderas. We identify 78 potential calderas spanning a broad range of water depths (down to 5,600 m), diameters (up to 20 km), and tectonic environments (divergent, convergent, and intraplate). Among these, eight shallow‑water calderas, mostly located in volcanic arcs, were highlighted as high‑priority targets due to their elevated hazard potential. This new global dataset addresses a major observational gap and provides a reproducible, extensible framework for submarine volcano characterisation, hazard evaluation, and deep‑sea exploration. The results emphasise the importance of incorporating submarine calderas into future global hazard models and monitoring strategies.
How to cite: Verolino, A., Lee, C., Jenkins, S. F., Jutzeler, M., and Switzer, A. D.: Calderas Beneath the Waves: AI-Powered Exploration of Subaqueous Volcanism, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-7642, 2026.
Quantifying topographic changes in volcanoes provides important information about volcanic deposits and mass-wasting processes, with implications for forecasting volcanic hazards. High-resolution Digital Elevation Models (DEMs) acquired over time are a powerful tool to develop time-series of topographic changes. Here we use EarthDEM/ArcticDEM DEMs, derived from Maxar satellites stereo-optical data, and DEMs derived from bistatic TerraSAR-X/TanDEM-X data to study topographic changes in different volcanoes placed worldwide. These volcanoes experienced different volcanic eruptions, generating a wide range of volcanic deposits and mass-wasting features. The high resolution of these DEMs allowed us to detect many topographic changes not visible with lower resolution DEMs, also in difficult environmental conditions, as long as height changes are ≥0.5-2 m, which is the range of vertical data errors. Pre-eruptive DEMs used to process bistatic data can affect volume estimates, while clouds and artifacts often affect EarthDEM/ArcticDEM. Nevertheless, high-resolution DEMs remain a valuable tool to quantify volcanic deposits and can be combined with other remote sensing data (thermal, InSAR) to better understand the volcanic activity in poorly monitored volcanoes. Acquisition of high resolution DEMs on a more frequent basis could significantly improve our ability to document time-dependent topographic changes at volcanoes worldwide.
How to cite: Galetto, F., Miller, S. M., Barris, R., Lobos Lillo, D., Shevchenko, A., and Pritchard, M.: High-resolution remote sensing data to measure topographic changes in volcanoes: successes, challenges and future perspectives, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-12470, 2026.
The Sea and Land Surface Temperature Radiometer (SLSTR), aboard Sentinel-3A/3B satellites, thanks to SWIR (shortwave infrared), MIR (medium infrared) and TIR (thermal infrared) bands, and a temporal resolution up to about 12 hours, may be used to detect, monitor and characterize thermal volcanic activity. In particular, the SWIR bands (500 m spatial resolution) may enable a more accurate identification of high-temperature volcanic features (e.g., lava flows/lava lakes), which could be then quantified also in terms of radiative power. Recently, the NHI (normalized hotspot indices) system, originally developed to map these features on a global scale through the analysis of Sentinel-2 and Landsat 8/9 imagery, has been extended to SLSTR SWIR observations to monitor active volcanoes in near real time. In this work, we present the updated NHI system, along with the outcomes of first months of operation. The results show the successful identification of several eruptive activities with a very low false positive rate, in both daylight and night-time conditions, as well as their effective characterization in terms of relative intensity level. The study demonstrates that SLSTR SWIR observations may provide valuable support to the surveillance of active volcanoes from space.
How to cite: Marchese, F., Mazzeo, G., ciancia, E., pietrapertosa, C., pergola, N., and filizzola, C.: A SLSTR (Sea and Land Surface Temperature Radiometer)-based system for the near-real-time monitoring of active volcanoes on a global scale from space., EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-16919, 2026.
Heat transfer at the surface in volcanic environments is an ongoing phenomenon representing the dynamic balance between the magma chamber and the adjacent rocks. In volcanoes, part of the magma’s energy drives fluid circulation, resulting in increased ground temperatures. Heat is primarily transferred through conduction, convection, and radiation, each detectable using specific techniques. Convection is evident in fumaroles and areas of diffuse degassing while moderate thermal anomalies indicate conductive heat transfer. Radiative fluxes can be measured using multispectral instruments. On Vulcano Island (Italy), the continuous monitoring network has recorded transient variations in heat flow from the active cone, associated with increased seismicity and ground deformation. Based on the generated time series, three volcanic thermal states have been defined (Background, Minor Crisis, and Unrest) corresponding to distinct thermal behaviors observed at the La Fossa crater. Building on these observations, we propose a two-stage methodology for forecasting volcanic thermal states using Artificial Intelligence applied to satellite remote sensing data. In the first stage, Long Short-Term Memory (LSTM) neural networks predict future values of time series derived from multi-sensor satellite imagery. In the second stage, a Semi-Supervised Generative Adversarial Network (SGAN), trained on the same satellite observations, classifies the LSTM-predicted series into volcanic thermal states. Input time series include established satellite-based monitoring products, such as the Normalized Thermal Index (NTI) and Volcanic Radiative Power (VRP) from VIIRS sensor, and environmental indices NDVI, NDWI, and NDMI from Sentinel 2 MSI sensor. This framework leverages the strengths of LSTM models for temporal forecasting and SGANs for robust classification with limited labeled data, enabling the prediction of volcanic thermal state evolution solely from Earth Observation data. Preliminary results indicate that the LSTM–SGAN framework can successfully forecast and classify thermal states at multiple future horizons. This work was supported by the 'Space It Up' project (code CUP I53D24000060005) funded by the Italian Space Agency, ASI, and the Ministry of University and Research, MUR, under contract n. 2024-5-E.0.
How to cite: Spina, F.: Forecasting Volcanic Thermal States with LSTM and SGAN Using Multi-Source Satellite Time Series: The Vulcano Case Study (2016–2024), EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-541, https://doi.org/10.5194/egusphere-egu26-541, 2026.
Numerical modeling is a powerful tool for predicting the most likely paths that lava flows may follow during an ongoing eruption. In particular, 2D models represent an excellent compromise between execution time and accuracy in simulating lava flows. These models can assimilate, as input data for numerical simulations, physical parameters provided by remote sensing or field observations such as the time-averaged discharge rate (TADR), the vent position, and the extent of the active lava field.
However, uncertainties associated with these parameters, combined with the simplifications inherent in the adopted numerical approaches, make it challenging to define the optimal conditions that best reproduce the actual lava flow and to make a reliable forecast of its evolution. Furthermore, simple 2D modeling struggles to accurately reproduce composite lava fields, which are generated from the overlap of multiple lava flow units that induces changes in the original topography and from unpredictable eruptive dynamics (e.g., opening of new vents, formation of lava tubes, and fluctuations in effusion rates). More complex eruptive dynamics can be addressed by simulating the different lava flow units through a multistep approach that includes multiple vents; however, this strategy increases the dimensionality of the parameter space required to run the model, leading to higher computational costs.
In this regard, an optimization strategy is fundamental to identify the best-fit solution by exploring the parameter space within a relatively short time. In this study, two methods are adopted: i) the Metropolis–Hastings approach, part of the Markov Chain Monte Carlo (MCMC) family, which performs a sequential refinement of the input parameters; and ii) the Nelder–Mead approach, a direct search method that minimizes a nonlinear objective function. The two methods differ in their goals and outcomes. The Metropolis–Hastings approach is designed to fully explore the multidimensional parameter space and to provide probability distributions of the parameters, whereas the Nelder–Mead approach aims to identify a single optimal solution that minimizes the mismatch between simulated and observed lava flows. The latter method significantly reduces computational costs compared to the MCMC approach; however, its performance may be affected by the presence of local minima, potentially preventing convergence toward the global minimum.
Both methods are tested on two recent effusive Mt. Etna (Italy) eruptions: the 27 February–1 March 2017 eruption, during which a single lava flow unit was emplaced over three days, and the 13 May–14 June 2022 eruption, characterized by multiple lava flows emitted from several vents that opened sequentially during the eruptive activity. The development of workflows based on these methods represents an important step towards the accurate, near-real-time reproduction of lava flows, which is essential for rapid hazard forecasting during volcanic crises and can be a powerful tool in assisting the mitigation of volcanic risk.
How to cite: Zuccarello, F., Bilotta, G., Cannavò, F., Cappello, A., Di Biasi, M., and Ganci, G.: Improvements to a lava flow simulation workflow with statistical and deterministic optimizations, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-10051, 2026.
Pyroclastic density currents (PDCs) are among the most hazardous particle-laden gravity currents on Earth, yet their runout and depositional footprints remain difficult to predict reliably. Accurate forecasting requires models that correctly represent flow density, which is strongly controlled by particle sedimentation rates. Recent high-fidelity Euler–Lagrangian simulations of polydisperse sedimenting particles have motivated a modified drag law that accounts for particle clustering, a common feature of highly mass-loaded flows such as PDCs. These simulations show enhanced settling of fine particles and hindered settling of coarse particles relative to isolated particle behavior. While this represents an important advance toward improved large-scale predictions, such as runout distance, the drag law constitutes only one component of a coupled, nonlinear system when embedded in depth-averaged hazard models such as IMEX_SfloW2D. Here, we apply adjoint-based sensitivity analysis to the clustering-aware drag law within IMEX_SfloW2D to quantify the influence of individual drag-law terms and model parameters on key quantities of interest, including deposition thickness and mean runout distance.
How to cite: Beetham, S. and Breard, E.: How Drag Laws Shape PDC Hazards: Adjoint Sensitivity in Depth-Averaged Models Applied to the Taupō 232 CE Eruption, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-8052, 2026.
With the onset of the Fagradalsfjall 2021 eruption, Reykjanes Peninsula entered a new eruptive period after 781 years break. Such periods last decades and can activate multiple volcanic systems on the Peninsula, including some that intersect the capital area. Eruptions from these volcanic systems have the potential to affect up to 75% of the Icelandic population (~ 285,000) either by compromising essential infrastructure and/or inundate inhabited areas. Between 2021-2025, twelve eruptions have occurred.
Therefore, the long-term lava hazard assessment began in 2024, which is a part of the volcanic hazard and risk assessment for the Reykjanes Peninsula, led by the Icelandic Meteorological Office on behalf of the Icelandic government. It is the first lava hazard assessment comprising the entire Peninsula reaching east to the South Iceland Seismic Zone at the Ölfusá river, comprising six overlapping volcanic systems.
The long-term lava hazard assessment is divided into three parts. Firstly, an assessment of spatial distribution of vent opening probability based on geological mapping of eruptive fissures (subglacial and subaerial), faults, geothermal areas and the plate boundary axis using MatHaz (Bertin et al., 2019). Secondly, lava flow simulations for four different eruption scenarios were performed on a 5m/pixel digital elevation model using the probabilistic code MrLavaLoba (de’Michieli Vitturi and Tarquini, 2018) covering a 200 square metre grid in areas with a vent opening likelihood > 0. In total nearly 200,000 simulations were executed on the supercomputing facilities of the Icelandic Research e-Infrastructure (IREI). This national high-performance computing (HPC) system were critical to achieving the resolution and duration required for the study. After post-processing, the likelihood of lava inundation can be assessed for the entire peninsula for each of the four eruption scenarios. Finally, the combined results of the likelihood of vent opening and lava inundation are assessed with respect to inhabited areas and essential infrastructure: water supply, power supply, and roads. The results are intended for urban planning and serve as a knowledge base for emergency response plans. They will be published in reports, a web-map and data repository.
Here we present key findings and discuss challenges in this long-term lava hazard including i) complex study area with multiple volcanic systems and with sparse geological information, ii) performing multiple eruption scenarios and iii) additional considerations needed when providing both static and online/dynamic maps.
How to cite: Pedersen, G. B. M., Pfeffer, M. A., Moreland, W. M., Óladóttir, B. A., Hjartardóttir, Á. R., Karlsson, Þ. Á., Wallevik, J. E., and Björnsson, B. B.: Advancing long-term lava hazard assessment of the Reykjanes Peninsula, SV Iceland, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-9575, 2026.
Forecasting volcanic eruptions remains challenging due to the scarcity of long-term monitoring data, the diversity of volcanic systems, and the difficulty of distinguishing subtle precursory signals from background variability. Here we proposed two methodological advances that offer complementary pathways to improve both scientific skill and operational decision-making: (i) machine-learning transfer forecasting based on ergodic seismic precursors, and (ii) socio-economic valuation of forecasts using the Potential Economic Value (PEV) framework.
First, we show that seismic precursors exhibit ergodic behavior, enabling machine-learning models trained on multi-volcano datasets to forecast eruptions at completely unseen, data-limited volcanoes. Using 73 years of continuous seismic data from 24 volcanoes and 41 eruptions, transfer-learning models identify statistically recurrent time-series features that strengthen prior to eruptions and can be effectively transferred between systems with distinct eruptive characteristics. Out-of-sample tests show forecasting skill comparable to tailored local models and exceeding benchmarks based on seismic amplitude. These results indicate that cross-volcano precursor patterns can provide robust forecasting capability even where local eruption histories are sparse, supporting global applicability of generalized forecasting tools.
However, forecast skill alone does not guarantee societal value. To address this gap, we introduce the potential economic values (PEV) framework to quantify the operational benefits of these forecasts by balancing the manageable costs of false alarms against the catastrophic consequences of missed eruptions. Retrospective analyses at Whakaari (2019) and Ontake (2014), combined with hypothetical high-impact scenarios, shows that even imperfect ML forecasts can reduce avoidable losses by 30–90%. PEV reveals that forecast value is maximized not by optimizing statistical accuracy, but by minimizing missed eruptions—highlighting the asymmetric socio-economic impacts of forecast errors. Optimal operational thresholds emerge within a stable range across volcanoes and cost assumptions, underscoring transferability of the framework.
By combining cross-volcano transfer learning with cost-based evaluation, our integrated framework advances two frontiers in volcanic hazard science: (1) improving eruption forecasting capability at data-limited volcanoes using ergodic precursor patterns, and (2) enabling monitoring agencies to select operational thresholds that maximize societal benefit rather than statistical performance alone. This approach supports more transparent, defensible, and economically efficient decision-making during volcanic unrest and provides a scalable pathway toward next-generation, globally transferable hazard-forecasting systems.
How to cite: Ardid, A., Dempsey, D., and Cronin, S.: Integrating Transfer Learning and Socio-Economic Value Metrics to Improve Eruption Forecasting and Decision-Making at Data-Limited Volcanoes, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-1509, 2026.
During eruptive crises, one of the key elements in emergency management is assessing whether and how magma is propagating, especially in cases of potentially dangerous lateral intrusions. A crucial issue is predicting in near-real time whether dyke propagation is likely to arrest or continue toward the lower flanks, where towns and villages are commonly located.
Magma ascent typically generates an extensional stress field around the dike propagation path, associated with earthquakes displaying normal focal mechanisms. The occurrence of events with reverse focal mechanisms, indicative of compressional regime, is rare in such settings. However, analysis of several eruptive episodes at Mt. Etna, from the 1989 crisis through the 2002 eruption, up to the eruptions of 2008 and 2018, reveals a consistent picture: the terminal portion of lateral intrusions that do not reach the surface is systematically characterized by the appearance of reverse focal mechanisms, which are absent during the initial propagation phases. According to the study, the appearance of reverse focal mechanisms is linked to a change in the stress field, likely associated with the magma's cooling and solidification processes, which favour compressive conditions. What emerges is a simple yet extremely effective indicator: reverse focal mechanisms are not an anomaly, but a key signal that allows us to recognize the potential arrest of a dike in near-real time, providing valuable constraints for operational decisions-making during eruptive crises.
How to cite: Giampiccolo, E., Bonaccorso, A., and Musumeci, C.: Near-real time recognition of dike arrest at Mt. Etna from reverse focal mechanisms, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-6738, 2026.
The Serracozzo cave is one of the most famous and fascinating lava tubes of Etna volcano. The cave formed in about one month during the final phase of the well-known 1971 flank eruption. It is a morphologically complex volcanic cavity because of its dual-nature: the upper section developed directly within the eruptive fissure, while the lower part formed within a ravine by sealing of an arterial ‘a‘ā lava flow, resulting in a distinctive lava tube. The features of the cavity formed within the eruptive fissure reveal a pulsating emplacement of the feeder dike, with alternation of magma pressure build-up and release. The dike emplaced a structural weakness along the Etna’s NE flank, which was frequently intruded by several other dikes during previous historical lateral eruptions. During its way to the surface the dike progressed by pulses, expanding laterally and then upwards. The same pulses occurred during its propagation down slope, where we observed wide chambers alternated to narrow passages. The effusive vents at the top of the fissure became skylights that acted as pressure release valves, forming short pāhoehoe overflows around the vents with upper level cavities partially merging with the main one during lava drain back. To characterize the volcanic features of the cave and reconstruct its spatiotemporal evolution, we conducted a traditional geological field survey integrated with old photos taken during the eruption, historical topographic maps and previous field surveys. Furthermore, high-resolution digital models of both the internal and external environments were generated using Terrestrial Laser Scanning (TLS) and Unoccupied Aerial Systems (UAS). This study realized with a multi-proxy approach is extremely valuable for hazard assessment because it allows us to constrain the timing necessary for the development and growth of the internal and external lava tube features and of how they evolved with time.
How to cite: Cavallaro, D., Calvari, S., Giudice, G., Messina, D., Self, S., Puglisi, G., De Beni, E., Cantarero, M., Morgavi, D., and Maugeri, R.: The Serracozzo cave, Etna Volcano, Italy: A peculiar cavity developed during the 1971 flank eruption within an eruptive fissure and an arterial lava flow, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-20443, 2026.
September 2021 on Vulcano island (Aeolian Islands, Sicily) was marked by the beginning of a new phase of volcanic unrest, during which the volcano underwent a dramatic increase in geophysical and geochemical parameters: a notable radial ground deformation centred around the Gran Cratere della Fossa crater occurred alongside increased seismicity, soil CO2 flux, plume SO2 flux, and fumarole outlet temperatures.
However, this period of volcanic unrest did not occur in isolation but was, in fact, preceded by two minor periods of ground deformation and increased soil CO2 flux occurring in 2018 and subsequently 2019. Global Navigation Satellite System (GNSS) and tilt data recorded during March – August 2018 have been used to create an analytical model of the deformation source, which shows a deeper source of inflation with respect to the source proposed for the 2021 unrest. The position of the 2018 source model is potentially indicative of a deeper recharge of the plumbing system beneath Vulcano, which reached progressively shallower levels before ultimately triggering the hydrothermal crisis of 2021.
This approach offers not only insight into the temporal evolution of a complex volcanic system before unrest periods, but also further implications as to the role played by tectonics. This is especially important in light of Vulcano’s position within a pull-apart-type structure along the Aeolian-Tindari-Letojanni fault system, a transtensional fault system which has shed light on the complex interplay between regional geodynamics in the southern Tyrrhenian Sea and volcanic activity at Vulcano.
How to cite: Bolam, A., Bruno, V., Messina, D., Mattia, M., and Ferlito, C.: Evolution of volcanic sources at Vulcano Island preceding the 2021 unrest from GNSS data, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-21224, 2026.
During the summer of 2024, Mt. Etna was characterized by a sequence of six powerful paroxysmal events originating from the Voragine summit crater. This activity marked a significant departure from the preceding years, when paroxysms at the Southeast Crater prevailed. The high dynamicity of this period required continuous and precise monitoring, mapping, and quantification, which were achieved through frequent Unoccupied Aerial System surveys. Using the difference of digital elevation models, which compared the pre-eruptive surface of April 29, 2024, against the post-eruptive surface of September 12, we clearly demonstrated a pattern dominated by net accumulation over the Voragine, with the greatest vertical accretion reaching over 108 m. This substantial growth, attributable to repeated lava effusion and pyroclastic deposition, established Voragine as the new peak of Mt. Etna, reaching an elevation of 3403 m a.s.l. We then analyzed the effect of topography changes on gravimetric terrain corrections, which is important for computing the complete Bouguer anomaly, and the impact of changes in the nearest topography on the prediction of vertical gravity gradients. This interdisciplinary work provides a detailed quantification of the eruptive products from Mt. Etna's 2024 volcanic sequence and highlights the critical impact of the resulting morphological changes on high-precision gravimetric surveying, thus emphasizing the need for up-to-date digital terrain models.
How to cite: De Beni, E., Proietti, C., Cantarero, M., Greco, F., Papčo, J., Zahorec, P., Vajda, P., Carbone, D., Mirabella, L. T., and Messina, A.: The dynamic summer of 2024 at Etna volcano documented by UAS: morphological changes and their gravimetric effects, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-18743, 2026.
Nyiragongo volcano, located in the western branch of the East African Rift (DR Congo), is among the world's most active and hazardous volcanoes. Its proximity to the densely populated city of Goma and to Lake Kivu, a CO2- and CH4-rich hydrogeological reservoir, makes pre- and rapid syn-eruptive hazard assessment critical, as demonstrated by the devastating impact of the January 2002 eruption on people and infrastructure.
On May 22, 2021, a new eruption produced extensive lava flows and significant co-eruptive surface deformations, prompting urgent assessment of the underlying magmatic and tectonic dynamics. We considered multi-orbit Sentinel-1 InSAR data to map the two-dimensional co-eruptive deformation field. Night-time Landsat thermal infrared imagery was used to quantify pre- and post-eruptive heat flux and assess its association with eruptive dynamics and deformation patterns. Interferograms reveal spatially heterogeneous deformation, indicating multiple deformation sources that reflect the combined influence of shallow magmatic intrusions, regional tectonic adjustments, and surface fracture propagation. Field observations by the Goma Volcano Observatory corroborate the satellite-detected displacements, confirming the location, orientation, and evolution of major fractures in eastern Goma.
The ultimate objective of this research is to establish an integrated, multi-sensor volcanic monitoring framework. By quantitatively linking InSAR-derived deformation with thermal infrared observations, we are able to capture the dynamic interplay between magmatic processes and surface responses. This synergistic approach provides a robust tool for hazard assessment and risk mitigation in densely populated regions of the East African Rift, bridging observational data with actionable decision-support for emergency management. Finally, this synergistic approach, combined with recent and upcoming advances in sensor capabilities, would open new avenues for space-based early-warning systems.
How to cite: sambo gloire, K. and Tolomei, C.: Capturing Nyiragongo’s Dynamics: Synergistic InSAR and Thermal Infrared Observations for Volcanic Hazard Assessment across the 2021 Nyiragongo Eruption, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-446, https://doi.org/10.5194/egusphere-egu26-446, 2026.
