Bayesian Monitoring of Seismo-Volcanic Dynamics

Rodriguez, Angel Bueno; Benítez, Carmen; Zuccarello, Luciano; Angelis, Silvio De; Ibáñez, Jesús M.

doi:10.1109/tgrs.2021.3076012

Cited by 7 publications

(8 citation statements)

References 38 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Thus, to analyze these patters we transform the original features vector into a features matrix, as highlighted in Figure 1. As Bueno et al (2021aBueno et al ( , 2021b and Cortés et al (2015) indicated, the Kurtosis and the Frequency Index can be used as indicators of the type of recorded seismic events and their evolution according changes in the volcanic system. The Kurtosis (Equation 3) evaluates how the frequencies of the signal are distributed, and the Frequency Index (Equation 4) takes into consideration the ratio of the energy content between high and low frequencies of the signal (we considered low frequencies between 1 and 6 Hz, and high frequencies between 6 and 16 Hz).…”

Section: Discussionmentioning

confidence: 99%

“…We want to emphasize that the volcanoes Mt. Etna, Bezymianny and Mount St. Helen have been selected, in addition to their interest based on their eruptive history, because they were the volcanoes studied by Bueno et al (2021aBueno et al ( , 2021b where it was observed how uncertainty could be used as a forecasting indicator.…”

Section: Data and Volcanic Scenariosmentioning

confidence: 99%

“…For this analysis we have selected four episodes of lava fountains from 2013, previously studied by Bueno et al. (2021a, 2021b) and which were forecasted by a strong volcanic tremor and absence of VT earthquakes. For Bezymianny volcano we selected two eruptive phases.…”

Section: Data and Volcanic Scenariosmentioning

confidence: 99%

See 2 more Smart Citations

Volcanic Early Warning Using Shannon Entropy: Multiple Cases of Study

Rey‐Devesa,

Benítez,

Prudencio

et al. 2023

JGR Solid Earth

Self Cite

View full text Add to dashboard Cite

The search for pre‐eruptive observables that can be used for short‐term volcanic forecast remains a scientific challenge. Pre‐eruptive patterns in seismic data are usually identified by analyzing seismic catalogs (e.g., the number and types of recorded seismic events), the evolution of seismic energy, or changes in the tensional state of the volcanic medium as a consequence of changes in the volume of the volcano. However, although successful volcanic predictions have been achieved, there is still no generally valid model suitable for a large range of eruptive scenarios. In this study, we evaluate the potential use of Shannon entropy as short‐term volcanic eruption forecasting extracted from seismic signals at five well studied volcanoes (Etna, Mount St. Helens, Kilauea, Augustine, and Bezymianny). We identified temporal patterns that can be monitored as short‐term eruptive precursors. We quantified the decay of Shannon entropy prior to eruptions, noting that changes appear between 4 days and 12 hr before. When Shannon entropy is combined with the temporal evolution of other features (i.e., energy, kurtosis, and the frequency index), we can elaborate physical models according to the occurring volcanic processes. Our results show that pre‐eruptive variation in Shannon entropy is a confident short‐term volcanic eruption monitoring tool.

show abstract

Section: Discussionmentioning

confidence: 99%

Section: Data and Volcanic Scenariosmentioning

confidence: 99%

See 1 more Smart Citation

Volcanic Early Warning Using Shannon Entropy: Multiple Cases of Study

Rey‐Devesa,

Benítez,

Prudencio

et al. 2023

JGR Solid Earth

Self Cite

View full text Add to dashboard Cite

show abstract

“…Variations in amplitude and frequency during the event are direct consequences of the source mechanism defining said event and can therefore serve to identify it. Given this clear temporal structure and the interest of the specific sequence of events occurring, several continuous VSR systems have been proposed in the literature using memory-based architectures like Hidden Markov Models [33], or the current state-of-the-art Recurrent Neural Networks [34] and Temporal Convolutional Neural Networks [35]. Figure 2 shows the common experimental methodology followed by continuous VSR systems [36].…”

Section: A Volcano-seismic Data: Modeling 1-d Temporal Structuresmentioning

confidence: 99%

Toward Knowledge Extraction in Classification of Volcano-Seismic Events: Visualizing Hidden States in Recurrent Neural Networks

Titos

Martínez

Kowsari

et al. 2022

IEEE J. Sel. Top. Appl. Earth Observations Remote Sensing

Self Cite

View full text Add to dashboard Cite

Understanding how deep hierarchical models build their knowledge is a key issue in the usage of artificial intelligence to interpret the reality behind data. Depending on the discipline and models used, such knowledge may be represented in ways that are more or less intelligible for humans, limiting further improvements on the performance of the existing models. In order to delve into the characterization and modeling of volcano-seismic signals, this paper emphasizes the idea of deciphering what and how Recurrent Neural Networks (RNN) model, and how this knowledge can be used to improve data interpretation. The key to accomplishing these objectives is both analyzing the hidden state dynamics associated with their hidden units as well as pruning/trimming based on the specialization of neurons. In this work, we process, analyze and visualize the hidden states activation maps of two RNN architectures when managing different types of volcanoseismic events. As a result, the class-dependent discriminative behavior of most active neurons is analyzed, thereby increasing the comprehension of the detection and classification tasks. A representative dataset from the Deception Island volcano (Antarctica), containing volcano-tectonic earthquakes, long period events, volcanic tremors and hybrid events, is used to train the models. Experimental analysis shows how neural activity and its associated specialization skills change depending on the architecture chosen and the type of event analyzed.

show abstract

“…Various techniques have been proposed for modelling subsurface dynamics, including Deep Learning and Machine Learning (e.g., Titos et al, 2018;Bueno et al, 2019;Bueno et al, 2021;Martínez et al, 2021), satellite remote sensing (e.g., Ganci et al, 2020), among others (e.g., Saccorotti and Lokmer, 2021). However, tomographic analysis based on seismic velocity and attenuation remains one of the best tools because it can provide direct links between changes in wave-field properties and the physical conditions of the medium (Castro-Melgar et al, 2021).…”

Section: Introductionmentioning

confidence: 99%

Small-Scale Volcanic Structures of the Aeolian Volcanic Arc Revealed by Seismic Attenuation

et al. 2021

View full text Add to dashboard Cite

We present the first two-dimensional (2-D) spatial distribution of seismic scattering and intrinsic attenuation beneath the Aeolian Islands arc. The Aeolian Islands archipelago represents one of the best examples of a small dimension volcanic island arc characterised by the alternation of different structural domains. Using the seismic wave diffusion model as the basis for the analysis, and using data from an active seismic experiment (TOMO-ETNA), we analysed more than 76,700 seismic paths marked by epicentre-seismic station pairs. Based on frequencies of 4–24 Hz, we identified high regional attenuation, comparable with other volcanic areas of the world. We used two different seismogram lengths, reflecting two different sampling depths, which allowed us to observe two different attenuative behaviours. As in most volcanic regions, scattering attenuation predominates over intrinsic attenuation, but some characteristics are area-specific. Volcanic structures present the highest contribution to scattering, especially in the low frequency range. This behaviour is interpreted to reflect the small size of the islands and the potentially relatively small size of individual magmatic feeding systems. In addition, strong scattering observed in one zone is associated with the northernmost part of the so-called Aeolian-Tindari-Letojanni fault system. In contrast, away from the volcanic islands, intrinsic attenuation dominates over scattering attenuation. We interpret this shift in attenuative behaviour as reflecting the large volume of sedimentary material deposited on the seabed. Owing to their poorly consolidated nature, sediments facilitate intrinsic attenuation via energy dissipation, but in general present high structural homogeneity that is reflected by low levels of scattering. Our results show that this region is not underlain by a large volcanic structural complex such as that beneath nearby Mt. Etna volcano. Instead, we observe dimensionally smaller and isolated subsurface volcanic structures. The identification of such features facilitates improved geological interpretation; we can now separate consolidated marine structures from independent subsurface volcanic elements. The results of this study provide a model for new research in similar regions around the world.

show abstract

Bayesian Monitoring of Seismo-Volcanic Dynamics

Cited by 7 publications

References 38 publications

Volcanic Early Warning Using Shannon Entropy: Multiple Cases of Study

Volcanic Early Warning Using Shannon Entropy: Multiple Cases of Study

Toward Knowledge Extraction in Classification of Volcano-Seismic Events: Visualizing Hidden States in Recurrent Neural Networks

Small-Scale Volcanic Structures of the Aeolian Volcanic Arc Revealed by Seismic Attenuation

Contact Info

Product

Resources

About