Project

Back to overview

Unlocking the near surface zone in seismic recordings: A paradigm shift in three-dimensional elastic modelling for full waveform inversion and non-linear laboratory experimentation

English title Unlocking the near surface zone in seismic recordings: A paradigm shift in three-dimensional elastic modelling for full waveform inversion and non-linear laboratory experimentation
Applicant Robertsson Johan O. A.
Number 162360
Funding scheme Project funding
Research institution Institut für Geophysik ETH Zürich
Institution of higher education ETH Zurich - ETHZ
Main discipline Geophysics
Start/End 01.01.2016 - 30.09.2020
Approved amount 463'046.00
Show all

Keywords (11)

full-waveform seismic inversion; nuclear waste disposal; modelling of seismic wave propagation; wave propagation; boundary conditions; high performance computing; finite-difference methods ; multicomponent seismic data ; near-surface zone; physical modelling; dedicated computational platforms

Lay Summary (Italian)

Lead
La modellazione della propagazione delle onde sismiche svolge un ruolo chiave in quasi ogni aspetto della sismologia ambientale, e di esplorazione. Viene utilizzata per comprendere il modo in cui le onde si propagano in un sottosouolo eterogeneo e complesso. La modellazione è utilizzata anche per la progettazione di un sondaggio, l'imaging, l’inversione e l'elaborazione dei dati sismici. Infine, la creazione di dati sintetici adeguati per la ricerca si è dimostrata estremamente preziosa per lo sviluppo e la sperimentazione di nuovi algoritmi per l’elaborazione di dati. La nostra ricerca punta a dare una svolta alla capacità di ridurre il costo computazionale della modellazione, da uno a diversi ordini di grandezza, basandosi sull'utilizzo delle cosiddette condizioni al contorno esatte. Oltre alle applicazioni menzionate in precedenza, la metodologia è essenziale anche per un nuovo tipo di laboratorio per la sperimentazione della propagazione delle onde che stiamo costruendo all'ETH.
Lay summary

Proponiamo dei metodi completamente nuovi per modellare gli esperimenti sismici, per invertire i dati sismici rispetto alle proprietà della terra, e per mappare le strutture nel sottosuolo. La nuova metodologia permette di ridurre in modo sostanziale i costi computazionali in scenari in cui è necessario ricalcolare ripetutamente la risposta sismica dopo aver alterato il modello (i costi delle ricalcolazioni sono molto bassi rispetto ai calcoli iniziali completi). La metodologia di modellazione cambia la natura dei calcoli richiesti e sarà particolarmente adatta per l'esecuzione su architetture di calcolo dedicate (per una specifica applicazione). Per questo motivo, vogliamo sfruttare il nuovo “motore” di modellazione per applicazioni di inversione e prevediamo significative riduzioni di un fattore 10-100 del costo computazionale a seconda dell'applicazione. Tale significativo sviluppo faciliterà il nostro obiettivo di effettuare inversione di dati sismici superficiali tenendo completamente in considerazione la fisica della propagazione delle onde in mezzi elastici - cosa che al giorno d’oggi è proibitiva da un punto di vista computazionale. La nuova metodologia sarà direttamente applicabile per risolvere problemi di lunga data relativi all’esplorazione di idrocarburi e minerali, alla caratterizzazione di riserve di acque sotterranee e idrotermali, allo stoccaggio di CO2, e alla sismologia dei terremoti.

 

Direct link to Lay Summary Last update: 28.09.2015

Responsible applicant and co-applicants

Employees

Publications

Publication
A consistent implementation of point sources on finite-difference grids
Koene Erik F M, Robertsson Johan O A, Andersson Fredrik (2020), A consistent implementation of point sources on finite-difference grids, in Geophysical Journal International, 223(2), 1144-1161.
Optimal finite-difference operators for arbitrarily sampled data
Koene Erik F. M., Robertsson Johan O. A. (2020), Optimal finite-difference operators for arbitrarily sampled data, in GEOPHYSICS, 85(3), F39-F51.
Accurate source wavelet estimation using Marchenko focusing functions
Mildner Constantin, Broggini Filippo, de Vos Koos, Robertsson Johan O. A. (2019), Accurate source wavelet estimation using Marchenko focusing functions, in GEOPHYSICS, 84(6), Q73-Q88.
Eliminating Time Dispersion from Visco-Elastic Simulations with Memory Variables
Koene E., Wittsten J., Robertsson J., Andersson F. (2019), Eliminating Time Dispersion from Visco-Elastic Simulations with Memory Variables, in 81st EAGE Conference and Exhibition 2019, London, UK,EAGE, Amsterdam.
A finite-difference algorithm to retrieve finite-difference modeled elastic waves at the free surface
Alumbaugh David, Bevc Dimitri, Koene Erik, Robertsson Johan (2018), A finite-difference algorithm to retrieve finite-difference modeled elastic waves at the free surface, in SEG Technical Program Expanded Abstracts 2018, Anaheim, CaliforniaSEG, Tulsa, OK.
GPU Implementation of Geophysical Algorithms
Broggini F. (2018), GPU Implementation of Geophysical Algorithms, in 80th EAGE Conference and Exhibition 2018, Copenhagen, DenmarkEAGE, Amsterdam.
Sensitivity Study of the Immersive Boundary Condition Method
Broggini F., Athanasopoulos N., Gray M., van Manen D. (2018), Sensitivity Study of the Immersive Boundary Condition Method, in 80th EAGE Conference and Exhibition 2018, Copenhagen, DenmarkEAGE, Amsterdam.
Using Sparsity to Improve the Accuracy of Marchenko Imaging of Single and Time-Lapse Seismic Given Imperfect Acquisitiont
Haindl C.M., Broggini F., Ravasi M., van Manen D.-J. (2018), Using Sparsity to Improve the Accuracy of Marchenko Imaging of Single and Time-Lapse Seismic Given Imperfect Acquisitiont, in 80th EAGE Conference and Exhibition 2018, Copenhagen, DenmarkEAGE, Amsterdam.
Eliminating time dispersion from seismic wave modeling
Koene Erik F M, Robertsson Johan O A, Broggini Filippo, Andersson Fredrik (2018), Eliminating time dispersion from seismic wave modeling, in Geophysical Journal International, 213(1), 169-180.
Removing numerical dispersion artifacts from reverse time migration and full-waveform inversion
Popovici Alexander Mihai, Fomel Sergey, Koene Erik, Robertsson Johan (2017), Removing numerical dispersion artifacts from reverse time migration and full-waveform inversion, in SEG Technical Program Expanded Abstracts 2017, Houston, TexasSEG, Tulsa, OK.
Robust Marchenko Focusing - Calibrating Surface Reflection with VSP Data
Thomsen H.R., Broggini F., van Manen D.-J., Ravasi M., Kritski A. (2017), Robust Marchenko Focusing - Calibrating Surface Reflection with VSP Data, in 79th EAGE Conference and Exhibition 2017, Paris, FranceEAGE, Amsterdam.
Immersive boundary conditions: Theory, implementation, and examples
Broggini Filippo, Vasmel Marlies, Robertsson Johan O. A., van Manen Dirk-Jan (2017), Immersive boundary conditions: Theory, implementation, and examples, in GEOPHYSICS, 82(3), T97-T110.
Target-oriented velocity analysis using Marchenko-redatumed data
Mildner Constantin, Broggini Filippo, Robertsson Johan O. A., van Manen Dirk-Jan, Greenhalgh Stewart (2017), Target-oriented velocity analysis using Marchenko-redatumed data, in GEOPHYSICS, 82(2), R75-R86.
Exact wavefield reconstruction on finite-difference grids with minimal memory requirements
Vasmel Marlies, Robertsson Johan O. A. (2016), Exact wavefield reconstruction on finite-difference grids with minimal memory requirements, in GEOPHYSICS, 81(6), T303-T309.
Exact Boundary Conditions for Local Wave Field Modelling
Broggini F., Vasmel M., Robertsson J.O.A., van Manen D.J. (2016), Exact Boundary Conditions for Local Wave Field Modelling, in 78th EAGE Conference and Exhibition 2016 - Workshops, Vienna, AustriaEAGE, Amsterdam.
A model-independent finite-difference method for removal of free-surface generated multiples
Vasmel Marlies, Robertsson Johan O. A., Amundsen Lasse (2016), A model-independent finite-difference method for removal of free-surface generated multiples, in GEOPHYSICS, 81(2), T79-T90.

Awards

Title Year
EAGE Guido Bonarelli award 2020
Geophysics Reviewer of the Year 2020 award from the journal Geophysics. 2020
ETH silver medal 2017
SEG best student paper award 2017

Associated projects

Number Title Start Funding scheme
157684 A wave propagation laboratory for immersive experimentation 01.12.2014 R'EQUIP
144303 Unlocking the near surface zone in seismic recordings: A paradigm shift in three-dimensional elastic modelling for full waveform inversion and non-linear laboratory experimentation 01.01.2013 Project funding

Abstract

A fundamental goal of the research in the ETH Exploration and Environmental Geophysics (EEG) group is to better understand the Earth’s shallow near-surface properties and structure (upper 10’s to 100’s of meters) and the effects that this “near surface corrupting lens” has on seismic reflections from deeper structures. The near-surface zone is characterized by strong variations in elastic properties, such as at the free surface, transitions from dry to fluid to gas saturated media, and the sharp discontinuities between different sediments, soils and rock types, all of which lead to very complex seismic records contaminated by coherent noise. The research approach conducted in the EEG group to tackle these problems is based on three research programmes: theory and methodology development, field data experimentation and laboratory experimentation. The project described in this proposal is well integrated within these pillars with a long-term vision of developing new wave-equation based elastic modelling and inversion methods that fully explain the recorded data.We propose new, efficient and potent (in terms of the physics of wave propagation) methods for modelling seismic wave propagation and undertaking full waveform inversion (FWI) of seismic data. The novel methodology is based on the application of new boundary conditions and the projection of real or modelled data from the surface to the target of interest. In particular, the methodology promises to substantially reduce computational cost for resynthesizing the seismic response after model alterations (cost of recomputations are very small compared to the full upfront computations). Moreover, the modelling methodology changes the nature of the required computations (small scale finite-difference computations in combination with pure cross-correlation operations) and will be particularly suited to run on dedicated (application-specific) computer architectures. We therefore want to exploit the new modelling engine for FWI applications and foresee significant reductions in computational cost by one or several orders of magnitude, depending on the application. Such a significant advance will facilitate our goal of carrying out 3D elastic FWI of surface seismic data on a full scale - something that is prohibitively computationally expensive today. The new elastic FWI methodology will aid the resolution of long-standing problems in hydrocarbon and mineral exploration (including the near-surface problem), groundwater and hydrothermal reservoir characterization, CO2 sequestration and earthquake seismology. Finally, many aspects of wave propagation in real Earth media and the physics of scattered waves are still poorly understood. We are in the process of building a new wave propagation laboratory in our group for such studies. The methodology for modelling and inversion outlined in this proposal is central to the concept of the wave propagation laboratory where a physical modelling experiment is fully immersed within a projected virtual environment.
-