Skip to main content

Data-intensive analysis of seismic tremors and long period events: a new paradigm for understanding transient deformation processes in active geological systems

Periodic Reporting for period 2 - SEISMAZE (Data-intensive analysis of seismic tremors and long period events: a new paradigm for understanding transient deformation processes in active geological systems)

Reporting period: 2020-07-01 to 2021-12-31

What is the problem/issue being addressed?
The SEISMAZE project is aimed at studying low frequency seismic tremors and earthquakes that form a broad class of signals generated by internal sources that are different from classical earthquakes. While the physical origin of this type of signals remains to be fully understood, they are related to slow transient energy release processes that occur in active geological systems during the accumulation of mechanical energy that is then released during catastrophic events, such as strong earthquakes or volcanic eruptions. Therefore, low-frequency seismic tremors and earthquakes represent a unique source of information that can be used to understand the physics of these ‘preparation’ processes and to design new monitoring and forecasting approaches.

Why is it important for society?
Low-frequency seismic tremors and earthquakes are manifestation of the slow processes that accommodate the mechanical evolution of geologically active systems under slow large-scale forcing prior to and after cataclysmic energy release during volcanic eruptions or strong earthquakes. They provide, therefore, a new paradigm for studying active geological processes in a variety of tectonic settings and for improving monitoring methods of seismic and volcanic hazards.

What are the overall objectives?
Modern digital seismological networks record huge numbers of tremors in different active regions, and breakthroughs can be achieved with systematic exploration of these observations that includes data analysis and physical modeling. The main objective of the SEISMAZE project is to undertake such an effort via the development of a new unified framework for the study of low-frequency seismic tremors and earthquakes. The plan to combine advanced methods for data mining, signal processing, and numerical simulations of the generating processes, to apply these to different large datasets of volcanic and tectonic low-frequency tremors and earthquakes.
Analysis of observations in targeted regions
We analysed following large datasets of continuous seismic recordings by large seismic networks:
- Four years of data from the Shikoku, Japan has been used to study tectonic low frequency earthquakes (LFE). We built a new LFE catalogue and studied their source scaling properties based on analysis of more than 130 thousand events.
- More than two years of the data from Guerrero, Mexico have been analysed to study the source scaling properties of the tectonic LFE. Their size-duration scaling has been found to be very different from the “regular” earthquakes.
- The data of the KISS seismic experiment in Kamchatka on one of the largest World’s cluster of subduction zone volcanoes have been used to build a new catalog of volcanic tremors that demonstrate how their sources are distributed over a wide range of depths form the surface to the crust-mantle boundary and are characterized by intermittency and very rapid migrations. The geographical distribution of the tremor sources follows main structural features inferred from new results of seismic tomography.
- The data from the seismic experiment on the Gorely volcano have been used to create a new catalogue of co-eruptive volcanic LFEs and to study their temporal clustering.
- The data from the permanent Kamchatka seismic network have been used to develop classification algorithms allowing to automatically separate signals generated by volcanic and tectonic earthquakes.
- More than the years of the data of the seismic network of the Piton de la Fournaise volcano (La Reunion island, France) have been used to build a new catalogue of seismo-volcanic tremors. The studied period includes multiple eruptions and the obtained results will be used to characterize the pre- and co-eruptive processes in the plumbing system and to develop machine-learning-based approaches for the seismo-volcanic monitoring.

New methods for analysis of STER seismic signals
- We continued to develop the analysis based on the network covariances (cross-correlations) to detect seismic tremors and to characterize their sources. In particular, we added a new approach to use the stability of temporal cross-correlations for detection and improved the procedure for the source location. The set of compute codes for this type of analysis is organized as an openly available python package “covseisnet” distributed via GitHub.
- We experimented with machine-learning approaches to classify seismo-volcanic low frequency earthquakes and tremors.
- We continued to develop statistical methods for analysing source properties of the LFE sources and their time-space patterns.

Physical modeling of tremor and LFE generating processes.
- We developed a new physical model of the source of volcanic LFE generated due to rapid degassing at depth and investigated its relationship with the H2O/CO2 content in the magmas.
- We developed a new model of the source region of the tectonic LFEs is subduction zone. This model considers as mane driving mechanism the pore-pressure diffusion via a subduction channel with a spatially heterogeneous and pressure-dependent permeability. We demonstrated that the model can reproduce some of the key features of the observed LFE catalogues such as strong intermittency and rapid migrations.
- We started numerical experiments to model the generation of seismic radiation from a fault zone in a visco-elastic media with damaging rheology.
- Systematic application of the analysis based on the network covariance matrix to large datasets (Kamchatka, La Reunion) provides us with un-precedent amount of information about seismic signals generated by active volcanic systems (including their time frequency properties) and about the distribution of their sources in space and time.
- We proceeded with building large catalogues of low-frequency seismic tremors and earthquakes in different seismically and volcanically active regions. These catalogues are built from the analysis of continuous seismic data based on new network-based methods. Together with relevant signals, they are used for the statistical analysis of the low-frequency seismic sources and related slow energy release processes in active geological systems.
- Systematic analysis of the source scaling properties in different regions is a very important task. To date, the comparison of the results from Japan and Mexico seems to indicate that the scaling might vary for different regions. In the future, we also plan to perform a similar source scaling analysis with the data from volcanic regions.
- We demonstrated that degassing of magma in the deep route of volcanoes can generate strong and rapid pressure changes sufficient for generation of volcanic low-frequency earthquakes.
- Results of our most recent numerical modeling show that rapid pore-pressure migration can be forcing driving mechanism driving the time-space patterns of the low-frequency earthquakes in subduction zones.
- Sources of seismic tremors detected beneath the Kyuchevskoy Volcanic Group allowed us to image very large trans-crustal active plumbing system and to observe how it is activated by fast pressure transients prior to eruptions.
Model of hydralically generated seismic tremors (Farge et al.: https://doi. org/10.1029/2021JB02189)
Volcanic earthquakes generated by magmatic degassing (Nature Comm.; doi: 10.1038/s41467-020-17759-4)