Ifsttar PhD subject

French version
Detailed form :

Title : Near-surface seismic imaging by waveform inversion and downscaling - Combined numerical and laboratory experimental approach

Main host Laboratory - Referent Advisor GERS - GeoEND  -  LEPAROUX Donatienne      tél. : +33 240845669

Director of the main host Laboratory DEROBERT Xavier  -

PhD Speciality Géophysique

Axis of the performance contract 2 - COP2017 - More efficient and resilient infrastructure

Main location Nantes

Doctoral affiliation UNIVERSITE DE NANTES

PhD school EGAAL - Ecologie, Géosciences, Agronomie et Alimentation

Planned PhD supervisor LEPAROUX Donatienne  -  Université Gustave Eiffel  -  GERS - GeoEND

Planned PhD co-supervisor CAPDEVILLE Yann  -  CNRS  -  LPG Nantes

Planned financing Contrat doctoral  - Ifsttar

Abstract

Context
In climatic change context and energy transition, the need for imaging and characterizing the underground media and infrastructures is increasing, for land planning dedicated to new technologies (wind turbine in off-shore or on-shore contexts) or for monitoring sites and structures or anthropogenic zones (sea dikes, natural hazard assessment in urban areas ..etc).
Among the non destructive methods developed in applied geophysics, seismic imaging techniques provide a mechanical information of the described targets. Usually based on arrival times of Body Waves, their adaptation to the near surface needs to take into account Surface Waves. The latter have long been identified as noise for deep investigations such as oil exploration.
At the same time, the last decades have seen the development of the so-called "Full Wave Form Inversion". These methods are based on the entire seismic signal to finely reconstruct the underground parameters (Virieux and Operto, 2009). This approach has been successful in deep exploration for hydrocarbons and some recent studies propose to integrate Surface Waves for near-surface auscultation. The interest is based on taking into account complex propagation phenomena and in particular mode conversions. However, this type of method is currently based on a local optimization approach by linearization of the inverse problem for computation time reasons. To be successful, such approach requires the knowledge of an initial model close to the solution, which is difficult to define in complex subsurface and civil engineering environments. Moreover, the non-unique nature of the solution makes it difficult to interpret the results.

Objective and Approach
To overcome these difficulties, an innovative alternative approach proposes a 2-step process "by inversion of the homogenized model and downscaling" (Capdeville and Methivier, 2018): the first stage is based on the result of the inversion of the waveform in terms of a homogenized model at the scales of the propagated wavelengths (Capdeville et al., 2010) and the second stage is based on a “interpretation inversion”. This second inversion is a global optimization of the reconstructed parameters designed to obtain a model distribution that can be interpreted in terms of the parameters needed for the targeted applications. The theoretical developments of this new approach as well as the first experimental 2D laboratory tests for non-attenuating media are being developed within the ANR HIWAI project (directed by Y. Capdeville, LPG). The prospects of this entirely new approach allow us to consider unsurpassed imaging capabilities for complex subsurface media if the information carried by surface waves can be integrated.
In this context, the proposed PhD topic aims to develop the feasibility of the approach of inversion by "homogenization and downscaling" for the problems of subsurface environments, generally very attenuating, including many heterogeneities with strong contrasts at surface wave propagation depths. This dual approach makes it possible to go beyond the traditional inversion of S wave velocities and assess the complete elastic tensor for then using anisotropy to provide more accurate information on the targeted targets. Here, the latter correspond to areas of alteration of the subsoil or the anthropogenic structures it contains and underground cavities. The objective is to define the potential for detection, localization and geometrical characterization of damage areas and cavities.
The proposed approach focuses on the theoretical analysis of the methodology by introducing superficial heterogeneities affecting surface waves and on the analysis of reduced-scale experimental laboratory data by taking into account not only the vertical but also the horizontal component (Bretaudeau et al., 2010 ; Pageot et al., 2017). To this end, reference models typical of the subsurface, of gradual complexity, will be defined and developed numerically and experimentally by reduced-scale resin models.

Scientific issues :
The main scientific issues of the thesis include the above-mentioned integration elements necessary for subsurface imaging that have not been explored or resolved to date, namely:
• integration of the horizontal component: contribution in numerical and experimental context
• the integration of surface heterogeneities into non-periodic homogenization
• the integration of attenuating media (involving a short propagation distance)
• experimental validation for subsurface applications in anthropogenic media

Modeling tools :
The numerical and experimental modeling tools are those developed at the LPG at the University of Nantes and at the ifsttar respectively:
• The 2D and 3D numerical codes based on the Spectral Element Method (SEM) developed in the LPG lab provide a fine simulation of the complex structures of underground media with a non-structured meshing.
• The MUSC bench developed in ifsttar provide a quantitative simulation of seismic recordings in a controlled environment. It provides both the vertical and the horizontal components.
The complementarity of both the numerical and experimental modeling has been already successfully tested in regional projects VIBRIS, PROSE and the ANR project, through the partnership of this thesis supervision teams.
Multi-component field data provided by the industrial project SR2S managed by IFSTTAR will contribute to test the method in this PhD thesis.

Required profile :
Wave propagation (acoustic, seismic, ultrasonic)
Physics of continuous media
Inverse problem
Signal and data processing
Python, Matlab or Scilab, C, fortran - or other tools for scientific calculation
An experience in numerical modeling (finite differences, finite elements, spec- elements, etc.)
traux, ...) is recommended.

Thesis supervision: Donatienne Leparoux and Yann Capdeville (thesis co-director)

Bibliography
Capdeville Y. & Métivier L. (2018) Elastic FWI based on the homogenization method... illustrations. Geophys.J.Int. 213 (2), 1093-1112.
Pageot D. et al. (2017) Improving the seismic small-scale modelling ... numerical methods. Geophys.J.Int. 211(1), 637-649.
Virieux J, OpertoS, 2009, An overview of full waveform inversion in exploration geophysics, GEOPHYSICS,VOL.74,NO.6,NOVEMBER DECEMBER2009;P.WCC127WCC152
Bretaudeau F. et al. (2011) Small-scale modeling of onshore seismic experiment... methods. Geophysics, 76(5), T101-T112.
Capdeville Y. et al. (2010) 2D nonperiodic homogenization to upscale elastic media for P-SV waves. Geophys. J. Int. 182, 903-922.

Keywords : Seismic Imaging, Invers Problem, Subsurface, non périodic homogeneization, Waveform, Numerical Modeling, experimental Modeling

Main host Laboratory - Referent Advisor	GERS - GeoEND - LEPAROUX Donatienne tél. : +33 240845669
Director of the main host Laboratory	DEROBERT Xavier -
PhD Speciality	Géophysique
Axis of the performance contract	2 - COP2017 - More efficient and resilient infrastructure
Main location	Nantes
Doctoral affiliation	UNIVERSITE DE NANTES
PhD school	EGAAL - Ecologie, Géosciences, Agronomie et Alimentation
Planned PhD supervisor	LEPAROUX Donatienne - Université Gustave Eiffel - GERS - GeoEND
Planned PhD co-supervisor	CAPDEVILLE Yann - CNRS - LPG Nantes
Planned financing	Contrat doctoral - Ifsttar
Abstract Context In climatic change context and energy transition, the need for imaging and characterizing the underground media and infrastructures is increasing, for land planning dedicated to new technologies (wind turbine in off-shore or on-shore contexts) or for monitoring sites and structures or anthropogenic zones (sea dikes, natural hazard assessment in urban areas ..etc). Among the non destructive methods developed in applied geophysics, seismic imaging techniques provide a mechanical information of the described targets. Usually based on arrival times of Body Waves, their adaptation to the near surface needs to take into account Surface Waves. The latter have long been identified as noise for deep investigations such as oil exploration. At the same time, the last decades have seen the development of the so-called "Full Wave Form Inversion". These methods are based on the entire seismic signal to finely reconstruct the underground parameters (Virieux and Operto, 2009). This approach has been successful in deep exploration for hydrocarbons and some recent studies propose to integrate Surface Waves for near-surface auscultation. The interest is based on taking into account complex propagation phenomena and in particular mode conversions. However, this type of method is currently based on a local optimization approach by linearization of the inverse problem for computation time reasons. To be successful, such approach requires the knowledge of an initial model close to the solution, which is difficult to define in complex subsurface and civil engineering environments. Moreover, the non-unique nature of the solution makes it difficult to interpret the results. Objective and Approach To overcome these difficulties, an innovative alternative approach proposes a 2-step process "by inversion of the homogenized model and downscaling" (Capdeville and Methivier, 2018): the first stage is based on the result of the inversion of the waveform in terms of a homogenized model at the scales of the propagated wavelengths (Capdeville et al., 2010) and the second stage is based on a “interpretation inversion”. This second inversion is a global optimization of the reconstructed parameters designed to obtain a model distribution that can be interpreted in terms of the parameters needed for the targeted applications. The theoretical developments of this new approach as well as the first experimental 2D laboratory tests for non-attenuating media are being developed within the ANR HIWAI project (directed by Y. Capdeville, LPG). The prospects of this entirely new approach allow us to consider unsurpassed imaging capabilities for complex subsurface media if the information carried by surface waves can be integrated. In this context, the proposed PhD topic aims to develop the feasibility of the approach of inversion by "homogenization and downscaling" for the problems of subsurface environments, generally very attenuating, including many heterogeneities with strong contrasts at surface wave propagation depths. This dual approach makes it possible to go beyond the traditional inversion of S wave velocities and assess the complete elastic tensor for then using anisotropy to provide more accurate information on the targeted targets. Here, the latter correspond to areas of alteration of the subsoil or the anthropogenic structures it contains and underground cavities. The objective is to define the potential for detection, localization and geometrical characterization of damage areas and cavities. The proposed approach focuses on the theoretical analysis of the methodology by introducing superficial heterogeneities affecting surface waves and on the analysis of reduced-scale experimental laboratory data by taking into account not only the vertical but also the horizontal component (Bretaudeau et al., 2010 ; Pageot et al., 2017). To this end, reference models typical of the subsurface, of gradual complexity, will be defined and developed numerically and experimentally by reduced-scale resin models. Scientific issues : The main scientific issues of the thesis include the above-mentioned integration elements necessary for subsurface imaging that have not been explored or resolved to date, namely: • integration of the horizontal component: contribution in numerical and experimental context • the integration of surface heterogeneities into non-periodic homogenization • the integration of attenuating media (involving a short propagation distance) • experimental validation for subsurface applications in anthropogenic media Modeling tools : The numerical and experimental modeling tools are those developed at the LPG at the University of Nantes and at the ifsttar respectively: • The 2D and 3D numerical codes based on the Spectral Element Method (SEM) developed in the LPG lab provide a fine simulation of the complex structures of underground media with a non-structured meshing. • The MUSC bench developed in ifsttar provide a quantitative simulation of seismic recordings in a controlled environment. It provides both the vertical and the horizontal components. The complementarity of both the numerical and experimental modeling has been already successfully tested in regional projects VIBRIS, PROSE and the ANR project, through the partnership of this thesis supervision teams. Multi-component field data provided by the industrial project SR2S managed by IFSTTAR will contribute to test the method in this PhD thesis. Required profile : Wave propagation (acoustic, seismic, ultrasonic) Physics of continuous media Inverse problem Signal and data processing Python, Matlab or Scilab, C, fortran - or other tools for scientific calculation An experience in numerical modeling (finite differences, finite elements, spec- elements, etc.) traux, ...) is recommended. Thesis supervision: Donatienne Leparoux and Yann Capdeville (thesis co-director) Bibliography Capdeville Y. & Métivier L. (2018) Elastic FWI based on the homogenization method... illustrations. Geophys.J.Int. 213 (2), 1093-1112. Pageot D. et al. (2017) Improving the seismic small-scale modelling ... numerical methods. Geophys.J.Int. 211(1), 637-649. Virieux J, OpertoS, 2009, An overview of full waveform inversion in exploration geophysics, GEOPHYSICS,VOL.74,NO.6,NOVEMBER DECEMBER2009;P.WCC127WCC152 Bretaudeau F. et al. (2011) Small-scale modeling of onshore seismic experiment... methods. Geophysics, 76(5), T101-T112. Capdeville Y. et al. (2010) 2D nonperiodic homogenization to upscale elastic media for P-SV waves. Geophys. J. Int. 182, 903-922.
Keywords :	Seismic Imaging, Invers Problem, Subsurface, non périodic homogeneization, Waveform, Numerical Modeling, experimental Modeling

List of topics

Applications closed