Matrix imaging as a tool for high-resolution monitoring of deep volcanic plumbing systems with seismic noise

2023/11/02 by Elsa Giraudat, Arnaud Burtin, Giraudat, Elsa +9 · 1 citation
Computer Science · Earth and Planetary Sciences · #Applied Physics (physics.app-ph) #FOS: Electrical engineering #FOS: Physical sciences #Geophysics (physics.geo-ph) #Image and Video Processing (eess.IV) #Seismic Imaging and Inversion Techniques #Seismic Waves and Analysis #Seismology and Earthquake Studies #electronic engineering #information engineering

paper · pdf · doi:10.48550/arxiv.2311.01296

openalex publication_date 2023/11/02 · openalex created_date 2023/11/04 · openalex updated_date 2026/08/03

Abstract

# Passive Seismic Matrix Imaging of La Soufrière volcano of Guadeloupe ----------------------------------------------------------------------------------------------------------------------------------------------------- ## General Notes: This repository contains the code for the article entitled "Unveiling the deep plumbing system of a volcano by a reflection matrix analysis of seimic noise" Authors: Elsa Giraudat, Arnaud Burtin, Arthur Le Ber, Mathias Fink, Jean-Christophe Komorowski, Alexandre Aubry Year: 2023 Preprint arXiv:2311.01296 - All post-processing was done with Matlab (R2022b) on a workstation with 2 processors @2.20GHz, 128Go from RAM, and a GPU with 48 Go of dedicated memory. It should be noted that the computation time is significantly reduced when a GPU is used. If a GPU is not available, the code can still be run but you must remove each occurrence of the "gpuArray()" function used to send the data to the GPU in Matlab. By default, they should normally already have been removed in this version of the code. ----------------------------------------------------------------------------------------------------------------------------------------------------- ## How to use this code The code MainCode_Rrr_Rcorr₁Z allows to compute the corrected confocal image at one depth of interest that can be choosed in the first lines of the file. (1) Download the experimental data corresponding to the cross correlation "NCFs_all₁-20Hz.mat" computed from the seismic noise recordings at La Soufrière de Guadeloupe (data reference in the paper) available on zenodo (https://zenodo.org/records/10066910) In the main script: (2) line #6 : choose the depth to be imaged in the variable "depth" . In the accompanying paper the three depths that are displayed for illustration in figures 2, 3 &4 are z= 1663 m, z= 6923 m and z = 9123 m. (3) line # 11 : change "data_dir" accordingly to your own data path where the experimental datas are stored. (4) line # 12 : change "code_dir" accordingly to your own working path with the main script and the subfunctions. (5) Then all other lines in the the main code can be executed without any change. ----------------------------------------------------------------------------------------------------------------------------------------------------- ## How does this code work The code works like that: (1) It loads the cross-correlation functions averaged over 2 months ("NCFs_all₁-20Hz.mat") computed from the seismic noise recordings at La Soufrière de Guadeloupe (data reference in the paper) (2) The focused reflection matrix is computed at the depth of interest by means of confocal redatuming process in the code section "CONFOCAL REDATUMING AT DEPTH Z (beamforming)" // See Section S2 of the accompanying supplementary Material for details about equations. (3) The original confocal reflectivity image is computed ("ImConf.orig") // See Section S3 of the accompanying supplementary Material (4) The focused reflection matrix is first corrected in the Earth surface basis (u) in the code section "CORRECTION 1 : in the EARTH SURFACE BASIS". This correction is iterated twice at output and at input. At each iteration the corresponding corrected confocal images and aberration laws are stored respectively in the ImConf structure and in the AbLawU_z structure. // See Section S7 of the accompanying supplementary Material for details about equations. (5) The focused reflection matrix is then corrected in the (k)-space in the code section " CORRECTION 2 : Additional correction in the k-space". This correction is performed one time at output and at input. The corresponding corrected confocal images and aberration laws are stored respectively in the ImConf structure and in the AbLawK_z structure. // See Section S8 of the accompanying supplementary Material for details about equations. (6) The reflection point-spread-functions (RPSF) are computed at each correction step using the M2_RPSF function and are stored in the RPSF_corr structure. The last sections of the code allow to display the results. (7) Uncorrected and corrected confocal images at depth of interest can be displayed using the section " DISPLAY CONFOCAL IMAGES AT DEPTH Z" (8) Cumulated input aberration law in the (u) basis can be displayed using code section "DISPLAY ABERRATION LAWS AT DEPTH Z IN THE U BASIS". The code can be adapted to display cumulated output aberration law. (9) Input aberration law in the (k) space can be displayed using code section "DISPLAY ABERRATION LAWS AT DEPTH Z IN THE K BASIS'. The code can be adapted to display output aberration law. (10) RPSFs can be displayed using code section "DISPLAY RPSF AT DEPTH Z". The structure R_corr contains the original focused reflection matrix and the corrected focused reflection matrices at each correction step: field "orig" : uncorrected Rrr matrix field "out1": Rrr matrix after 1 output correction in the (u) basis field "in1": Rrr matrix after 1 input correction in the (u) basis field "out2": Rrr matrix after 2 output corrections in the (u) basis field "in2": Rrr matrix after 2 input corrections in the (u) basis field" outk": Rrr matrix after 1 output correction in the (k) basis field" ink": Rrr matrix after 1 input correction in the (k) basis The structure Imconf contains the original and the corrected confocal image at each correction step. The fields names refer to the same correction steps as in the R_corr structure. The structure AbLawU_z contains the aberration law computed in the (u) basis at each correction step. field "out1": 1st output aberration law in the (u) basis field "in1": 1st input aberration law in the (u) basis field "out2": 2nd output aberration law in the (u) basis field "out2": 2nd input aberration law in the (u) basis The structure AbLawK_z contains the aberration law computed in the (k) basis at each correction step. field "out1": output aberration law in the (k) basis field "in1": input aberration law in the (k) basis The structure RPSF_corr contains the reflection point-spread-functions computed for each correction step. The fields names refer to the same correction steps as in the R_corr structure. An additional field "RPSF_corr.Dr_z" contains the Δ_rho positions corresponding to the common mid point basis described in the supplementary section S6. The structure Prm contains various focusing and correction parameters.

Cited by

• Taking advantage of multiple scattering for Optical Reflection Tomography

Related

• Abrupt changes of hydrothermal activity in a lava dome detected by combined seismic and muon monitoring
• VOLCANIC SEISMOLOGY
• Scanning Volcanoes by Synthetic Aperture Radar
• Muon dynamic radiography of density changes induced by hydrothermal\n activity at the La Soufri `ere of Guadeloupe volcano
• AWESAM: A Python Module for Automated Volcanic Event Detection Applied to Stromboli