Preprints
https://doi.org/10.5194/nhess-2022-164
https://doi.org/10.5194/nhess-2022-164
 
14 Jun 2022
14 Jun 2022
Status: this preprint is currently under review for the journal NHESS.

Multi-station automatic classification of seismic signatures from the Lascar volcano database

Pablo Salazar1,2,3, Franz Yupanqui4, Claudio Meneses4, Susana Layana1,2,3, and Gonzalo Yañez5 Pablo Salazar et al.
  • 1Núcleo de Investigación en Riesgo Volcánico-Ckelar Volcanes, Universidad Católica del Norte, Antofagasta, Chile
  • 2Departamento de Ciencias Geológicas, Universidad Católica del Norte, Chile
  • 3Centro de Investigación para la Gestión Integrada del Riesgo de Desastres (CIGIDEN), Santiago, Chile
  • 4Departamento de Ingeniería de Sistemas y Computación, Universidad Católica del Norte, Chile
  • 5Departamento de Ingeniería Estructural y Geotécnica, Pontificia Universidad Católica de Chile, Chile

Abstract. This study was aimed to build a multi-station automatic classification system for volcanic seismic signatures. This system was based on a probabilistic model made using transfer learning, which has, as the main tool, a pre-trained convolutional network named AlexNet. We designed four experiments using different datasets with data that was real, artificial, and two different combinations of these (combined 1 and combined 2). The experiment presented the highest scores when a process of data augmentation was introduced into processing sequence. Thus, the lack of real data in some classes (imbalance) dramatically affected the quality of the results, because the learning step (training) was over-fitted to the more numerous classes. To test the model stability with variable inputs, we implemented a k-fold cross-validation procedure. Under this approach, the results were more than optimal, considering that only the percentage of recognition of the tectonic events (TC) class was partially affected. The most valuable benefit of using this technique was that the use of volcano seismic signals from multiple stations provided a more generalisable model which, in near future, can be extended to multi-volcano database systems. The impact of this work is significant in the evaluation of hazard and risk by monitoring the dynamic evolution of volcanic centres, which is crucial for understanding the stages in a volcano’s eruptive cycle.

Pablo Salazar et al.

Status: open (until 23 Oct 2022)

Comment types: AC – author | RC – referee | CC – community | EC – editor | CEC – chief editor | : Report abuse

Pablo Salazar et al.

Pablo Salazar et al.

Viewed

Total article views: 208 (including HTML, PDF, and XML)
HTML PDF XML Total BibTeX EndNote
162 41 5 208 4 2
  • HTML: 162
  • PDF: 41
  • XML: 5
  • Total: 208
  • BibTeX: 4
  • EndNote: 2
Views and downloads (calculated since 14 Jun 2022)
Cumulative views and downloads (calculated since 14 Jun 2022)

Viewed (geographical distribution)

Total article views: 191 (including HTML, PDF, and XML) Thereof 191 with geography defined and 0 with unknown origin.
Country # Views %
  • 1
1
 
 
 
 
Latest update: 06 Oct 2022
Download
Short summary
The acquisition of more generalisable models, using machine learning techniques, creates a good opportunity to develop a multi-volcano probabilistic model for volcanoes worldwide. This will improve the understanding and evaluation of the hazards and risks associated with the activity of volcanoes.
Altmetrics