The paths of lava flows are well predictable following the gradient of the terrain, however, at near flat topography flow reconstruction is challenging. This problem stems from the importance of small barriers and topographic complexities affecting low-gradient terrain lava flows. Lava flows emplaced on low-gradient surfaces can propagate in a wide range of directions, making hazard forecasting highly sensitive to initial conditions and topographic representation. These challenges are compounded by long repose times between eruptions, which often result in poorly constrained pre-eruptive surfaces that must be reconstructed to achieve meaningful comparisons between simulated and observed lava flows.
The San Francisco Volcanic Field (SFVF), one of the largest in the continental United States, poses a significant lava-flow hazard due to eruptions such as Sunset Crater (~1085 CE) and the SP lava flow (5.5–6 ka). Assessing lava flow hazards at the SFVF is inherently challenging due to uncertainties in the spatio-temporal distribution of future eruptive vents and the strong sensitivity of flow trajectories to subtle variations in slope and aspect, particularly on the gentle terrain.
To this aim we combined a remote sensing and numerical modeling approach to constrain the emplacement dynamics of the SP lava flow, and perform lava flow simulations on gentle slopes. We first analyze a high-resolution drone-derived digital elevation model (DEM) and orthomosaic to map lava flow outlines, surface structures (e.g., channels, levees, and folds), and topographic features such as grabens and fluvial incisions. This mapping is complemented by automated surface-texture classification using Sentinel-2 multispectral satellite data, enabling reconstruction of a sequential emplacement history, comprising two main pulses of extrusion. These observations inform the reconstruction of the pre-eruptive surface, incorporating inferred tectonic and fluvial features.
The reconstructed surface is then compared to the present topography to estimate lava flow volume and thickness distributions, and to constrain model parameters for lava flow simulations. We use the MrLavaLoba model, which includes an inertia-like parameter well suited for simulating lava propagation on gentle slopes. Multiple eruptive scenarios, including single-pulse and two-pulse eruptions, are simulated and quantitatively compared with thickness distribution observed in our derived DEM.
Our results demonstrate that detailed reconstruction of pre-eruptive topography significantly improves model–data agreement for lava flows emplaced on gentle terrain. We propose a best-practice workflow for integrating remote sensing data and lava flow modeling in distributed volcanic fields, with direct implications for future lava flow hazard assessments at the SFVF and similar volcanic systems worldwide.
How to cite: Brauner, J., Tarquini, S., Walter, T. R., Liu, C., Germa, A., Smekens, J.-F., and Vanderkluysen, L.: Controls on Lava Flow Emplacement on Low-Gradient Terrain: Insights from the SP Lava Flow (San Francisco Volcanic Field, USA), EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-5545, 2026.
In 2018, Shinmoedake volcano, Japan, produced a rapidly emplaced, pancake-shaped andesitic lava dome within approximately three days. The lava filled the summit crater and slightly overflowed from the crater rim. Because collapse of lava domes can generate hazardous pyroclastic flows, understanding the behavior of lava overflow is essential for hazard assessment. To investigate the controls on overflow behavior, we conduct numerical simulations of lava extrusion within the Shinmoedake crater, using a depth-averaged Bingham-fluid model. The comprehensive simulations show that the overflow direction is primarily controlled by lava viscosity. When the viscosity is lower than 10^9 Pa s, lava overflows from the western side. In contrast, when the viscosity exceeds 10^9 Pa s, lava overflows from the eastern side. This difference in overflow direction is explained by a geometric effect: at higher viscosity, reduced lateral spreading leads to thicker lava accumulation, allowing the flow to overcome the higher eastern crater wall. By comparing the numerical results with SAR observations capturing the detailed evolution of dome morphology, we further constrain the lava viscosity to values lower than 10^9 Pa s and estimate the corresponding yield strength during the 2018 eruption. Using these rheological parameters, we discuss implications for predicting the extent of future lava emplacement and associated pyroclastic-flow hazard areas at Shinmoedake volcano.
How to cite: Maruishi, T. and Kozono, T.: Numerical simulation of pancake-shaped lava dome overflow from the summit crater: the 2018 Shinmoedake eruption, Japan, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-9010, 2026.
Rapid and accurate forecasting of volcanic eruptions remains a central challenge for volcano surveillance agencies. Traditionally, forecasting efforts have focused on recognizing recurrent patterns in geophysical or geochemical signals to detect unrest and assess its evolution; however, translating these precursory signals into clear, easy-to-interpret eruption probabilities remains challenging. A promising signal in the context of probabilistic eruption forecasting is seismic tremor, as it often exhibits recognizable patterns (e.g., amplitude escalation, frequency shifts, and spectral variations) during the run-up to eruptions. This raises the following question: Can seismic tremor patterns be used operationally to produce objective eruption probabilities? To address this question, we developed a supervised machine learning-based framework built upon the Dempsey et al., 2020 [https://doi.org/10.1038/s41467-020-17375-2], Ardid et al., 2023 [https://doi.org/10.21203/rs.3.rs-3483573/v1], and Girona and Drymoni, 2024 [https://doi.org/10.1038/s41467-024-51596-z] approaches, and tested it retrospectively on 13 paroxysmal events at Shishaldin Volcano, Alaska, that occurred between July and November 2023. Specifically, our framework extracts statistical features from continuous tremor data, such as dominant frequency, amplitude, kurtosis and Shannon entropy, and applies a Random Forest classifier to quantify the similarity between ongoing tremor and previously recorded pre-eruptive tremor; this similarity can, in turn, be interpreted as an estimate of the probability of an eruption occurring within a specific time window. To mimic operational conditions, models were retrained on progressively larger datasets, using only data available prior to each Shishaldin paroxysm; and forecasts targeted seismic amplitude peaks and the onset of ash emissions for 1, 6, 12, and 24-hour windows. Results show that, in most cases, probabilities increased in the lead-up to the paroxysms, indicating that our approach captured evolving tremor patterns associated with imminent explosive activity. Although evaluated retrospectively, the findings highlight the potential of seismic tremor–based probabilistic forecasts to support volcano monitoring and decision-making during volcanic crises. The framework is fully retrainable, automatically updating as new paroxysms occur and additional data become available, thereby enhancing its suitability for near-real-time operational use and enabling straightforward extension to other volcanic systems.
How to cite: Burgos, V. and Girona, T.: Toward Operational Probabilistic Eruption Forecasting Using Machine Learning and Seismic Tremor: A Retrospective Study of the 2023 Shishaldin Paroxysms (Alaska), EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-9178, 2026.
The increasing availability of free and global satellite Interferometric Synthetic Aperture Radar (InSAR) data, combined with the development of automatic InSAR processing chains operating at regional to global scales makes it possible to obtain dense and regularly updated spatio-temporal measurements of ground deformation over hundreds of active volcanoes worldwide. This growing volume of InSAR time series offers new opportunities for operational monitoring, but raises significant challenges for automated analysis and interpretation. Surface deformation reflects magmatic and hydrothermal processes associated with magma storage, pressurization, migration and withdrawal within volcanic plumbing systems and may constitute a precursory signal during the early stages of volcanic unrest. From an operational perspective, automatic detection of the onset of deformation in SAR time series is required to support early warning strategies, but it remains a major methodological challenge. Early-stage deformation signals are low-amplitude, spatially heterogeneous, and temporally non-stationary, while InSAR observations are affected by atmospheric delays, temporal decorrelation, and topography-related noise. These effects significantly reduce the detectability of deformation, particularly at active volcanoes characterized by low signal-to-noise ratios, raising the question of how early deformation can be detected with statistical confidence. Machine learning approaches based on convolutional neural networks (CNNs) rely on spatial pattern recognition to detect deformation signals on individual interferograms. CNNs require the use of extensive training datasets across many volcanoes, and often do not consider temporal information. As a result, the approach is more suitable fo scenarios with large signal-to-noise ratios. Additionally, independent component analysis (ICA) exploits both spatial and temporal information. However, it requires long-duration and complete time series to separate persistent deformation signals from noise and relies on the assumption of statistical independence between deformation and noise components. Here, we propose an operational detection framework that jointly exploits the spatial and temporal structure of InSAR data, enabling the identification of coherent deformation signals while explicitly accounting for their spatio temporal evolution. This study investigates detection strategies for the automatic identification of volcanic deformation in synthetic SAR time series coupling deformation signals and noise sources. Synthetic deformation scenarios representative of different volcanic processes, including linear, exponential, and transient inflation or deflation driven by analytical models (Mogi, Okada), are generated and embedded within spatially and temporally correlated atmospheric noise fields, providing a ground-truth framework to evaluate detection performance under varying deformation regimes and noise conditions. Recursive filtering techniques, such as Kalman filters, are considered to improve signal-to-noise ratio and enable continuous tracking of deformation in the presence of irregular acquisitions. Probabilistic change-point detection methods are investigated to identify transitions in deformation regimes and assess the likelihood of deformation onset, particularly at early stages. In parallel, cumulative detection statistics are examined, based on persistent exceedances relative to background noise variance, including the spatio-temporal CUSUM method, in order to exploit both the temporal persistence and spatial consistency of deformation signals. By comparing and combining these methods, the framework aims to identify which detection strategies are most appropriate for different unrest scenarios and noise environments.
How to cite: Bouygues, P., Albino, F., and Pinel, V.: Testing Automatic Detection Algorithms of Volcanic Unrest in SAR time series using Synthetic data, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-10989, 2026.
Effective volcano monitoring relies on the timely detection and correct classification of diverse, time-dependent geophysical signals associated with magmatic and hydrothermal processes, including volcanic tremor, long-period and volcano-tectonic earthquakes, deformation transients, gas release, and thermal anomalies. Artificial Intelligence and Machine Learning (AI/ML) methods have emerged as powerful tools to automate event detection, classification, and forecasting in operational volcano observatories. Consequently, the number of peer-reviewed studies applying AI/ML to volcano monitoring has increased exponentially in the past decade.
Despite this rapid development, we suggest that the effective operational uptake of AI/ML in volcano monitoring remains limited due to four structural challenges. First, the lack of standardised, community-accepted benchmarking datasets and evaluation protocols prevents meaningful comparison of algorithm performance across studies, volcanoes, and datatypes. Second, differing implementation, training, and testing practices limit reproducibility. Third, many AI/ML-based monitoring methods remain deterministic, with limited or no uncertainty quantification. This favours overconfident models and complicates their integration into probabilistic, risk-based decision frameworks that are central to operational volcanology. Finally, the relative novelty of AI/ML in volcano monitoring has resulted in a fragmented research landscape with limited coordinated community infrastructure.
We propose a community-driven initiative to address these limitations through the design of a modular, open validation framework for AI/ML methods in volcano monitoring. The framework should integrate curated, benchmark-quality multi-parameter datasets that capture real-world variability in volcanic behaviour. Standardised training, testing, and evaluation protocols will enable fair, transparent, and reproducible comparison of both classical and emerging AI/ML approaches and the inclusion of uncertainty quantification, allowing performance to be assessed not only in terms of accuracy but also in terms of reliability and decision relevance.
By establishing shared benchmarks and open evaluation practices, we aim to accelerate methodological development, improve reproducibility, and support the responsible transfer of AI/ML tools into operational volcano observatories. We will present a prototype as a starting point and invitation to the volcanological and data science communities to help design and implement this validation framework.
How to cite: Behr, Y., Hammer, C., and Ohrnberger, M.: Towards a community-driven validation framework for AI/ML methods in volcano monitoring, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-15238, 2026.
Monitoring active volcanoes requires the integrated analysis of multidisciplinary datasets to constrain magma migration and its temporal evolution. Here we present a multidisciplinary study combining geochemical observations, seismic–volcanic signals, and thermal satellite data to investigate magmatic processes within Mt. Etna’s shallow plumbing system. The integrated dataset allows us to assess magma ponding, convection, and degassing dynamics, with particular emphasis on halogen gas emissions.
Halogen fluxes are used to evaluate the efficiency of magma residence and the steadiness of magma supply in the shallow system. From January to April 2023, persistent thermal anomalies, low infrasound activity, stable volcanic tremor, and sustained halogen degassing indicate a steady-state degassing regime and efficient magma rejuvenation at shallow levels. From late May 2023, an increase in SO₂ emissions not accompanied by a proportional increase in HCl emissions, together with enhanced infrasound activity, increased tremor amplitude, and sporadic thermal anomalies, suggests a decoupling between deep gas ascent and magma ascent within the main conduit.
These observations indicate that halogen flux monitoring, owing to the high solubility of halogens in silicate melts, provides a sensitive indicator of changes in magma supply rate and near-surface magma residence time in basaltic volcanic systems.
How to cite: La Spina, A., Sciotto, M., Corradino, C., Salerno, G., Di Grazia, G., Bonfanti, P., and Del Negro, C.: Shallow magma ponding and degassing beneath Mt. Etna summit craters inferred from multi-parameter survey, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-21241, 2026.
