Network-based analysis of seismo-volcanic tremors
International audience Volcanic tremors represent one of the most important class of seismo-volcanic signals due to their abundant presence in seismic records, their wealth of information regarding magmatic systems, their use as a tool for monitoring the state of volcanoes and their potential as pre...
Main Authors: | , , |
---|---|
Other Authors: | , |
Format: | Book Part |
Language: | English |
Published: |
HAL CCSD
2023
|
Subjects: | |
Online Access: | https://insu.hal.science/insu-04167423 https://insu.hal.science/insu-04167423/document https://insu.hal.science/insu-04167423/file/Volcanic-tremor_Shapiro-Soubestre-Journeau_2023-06-20_Revision-untracked-changes.pdf |
Summary: | International audience Volcanic tremors represent one of the most important class of seismo-volcanic signals due to their abundant presence in seismic records, their wealth of information regarding magmatic systems, their use as a tool for monitoring the state of volcanoes and their potential as precursor signals to eruptions. These signals have been analyzed for several decades with single station approaches, from which empirical inferences can be made regarding their sources, generation mechanism and scaling relations with eruptions parameters. Modernisation and densification of instrumentation networks coupled with sophistication of analysis methods and enhanced computation capacities, allow to switch from single-station to full seismic network based methods. We introduce in this chapter the interstation cross-correlations methods, the estimation of the network covariance matrix and the study of its eigenvalues and eigenvectors. Such advanced methods enable to measure temporal, spatial and spectral tremor properties. They are contained in the CovSeisNet Python package which has been used for characterizing various tremor episodes, including two examples from Kilauea volcano, Hawaii and Klyuchevskoy Volcanic Group, Kamchatka presented in this chapter. These application examples illustrate the complexity of tremors and emphasize the need to continue the development of new algorithms aimed at the exploration of network covariances to better constrain the different tremor generation processes that can be multiple, simultaneous and interacting. In particular, the combination of network-based analysis with polarization and machine learning approaches may represent a new step in our understanding of the underlying phenomena. In turn, this enhanced discernment of the involved processes and the links with the properties of the volcanic system can lead to a more effective monitoring and ultimately a better apprehension of volcanic system destabilizations and anticipation of the associated unrests. |
---|