S017
Geophysical inversion, inference and imaging in the age of machine learning
Session ID#: 80845
Session Description:
A defining feature of many geoscience problems is the need to perform inference using only indirect measurements of the system of interest, and the development of novel mathematical tools for inversion and inference remains central to progress in our knowledge of the Earth. Open challenges include: (i) proper characterization of uncertainty and non-uniqueness; (ii) handling non-linearity in an effective manner; and (iii) managing the practical challenges associated with massive data sets and expensive forward models. This session will highlight recent developments in geophysical inverse theory (broadly defined), focusing primarily on methodology. We particularly hope to explore synergies and parallels between inversion algorithms and machine learning, and the potential for fundamentally new approaches to geophysical inference. However, we welcome all abstracts that address problems in inversion, regardless of the approach adopted.
Index Terms:
0555 Neural networks, fuzzy logic, machine learning [COMPUTATIONAL GEOPHYSICS]
1942 Machine learning [INFORMATICS]
3260 Inverse theory [MATHEMATICAL GEOPHYSICS]
3275 Uncertainty quantification [MATHEMATICAL GEOPHYSICS]
Primary Convener: Andrew P Valentine, Australian National University, Research School of Earth Sciences, Canberra, ACT, Australia
Conveners: Malcolm Sambridge, Australian National University, Canberra, ACT, Australia, Anya Marie Reading, University of Tasmania, Hobart, TAS, Australia and Lion Krischer, ETH Zurich, Department of Earth Sciences, Institute of Geophysics, Zurich, Switzerland
Primary Liaison: Andrew P Valentine, Australian National University, Research School of Earth Sciences, Canberra, ACT, Australia
Seismic full waveform inversion of wide-aperture seismic data using a trans-dimensional Bayesian method (492697)
Peng Guo1, Satish Chandra Singh2, Gerhard Visser3, Venkata Abhishikth Vaddineni4 and Erdinc Saygin1, (1)Deep Earth Imaging, CSIRO, Perth, WA, Australia, (2)IPG Paris, Paris Cedex 05, France, (3)CSIRO, Deep Earth Imaging Future Science Platform, Perth, WA, Australia, (4)Institut de Physique du Globe de Paris, Paris, France
A Bayesian Approach to Geodynamic Tomography (500614)
John Keith Magali, Villeurbanne, Auvergne-Rhône-Alpes, France, Thomas Bodin, LGLTPE, Univ. Lyon 1, ENS Lyon and CNRS, Lyon, France, Navid Hedjazian, University Claude Bernard Lyon 1, Villeurbanne, France, Henri Samuel, IRAP, Toulouse, France; CNRS, Paris Cedex 16, France and Suzanne Atkins, Utrecht University, Utrecht, 3584, Netherlands
Machine Learning in the Investigation of Active Deformation Processes: Rethinking Non-linearity in Seismic Waveform Modelling (504039)
Anya Marie Reading1, Ross J Turner2, Stephen G Meyer1,3 and Stephen J Walters2, (1)University of Tasmania, Hobart, TAS, Australia, (2)University of Tasmania, Hobart, Australia, (3)Institute of Mine Seismology, Kingston, TAS, Australia
Multi-scale Deep Learning Seismic Full Waveform Inversion (510331)
Zi Xian Leong, Pennsylvania State University Main Campus, Department of Geosciences, University Park, PA, United States and Tieyuan Zhu, Pennsylvania State University, Department of Geosciences, University Park, PA, United States
A Fuzzy Inversion Approach to Earthquake Kinematic Finite-Fault Source Modelling (534440)
Navid Kheirdast, International Institute of Earthquake Engineering and Seismology, Seismology, Tehran, Iran, Anooshiravan Ansari, International Institute of Earthquake Engineering and Seismology, Tehran, Iran and Susana Custodio, IDL - Instituto Dom Luiz, Lisboa, Portugal
Up, down and around: Transformational approaches to geophysical inverse problems. (535364)
Malcolm Sambridge, Australian National University, Canberra, ACT, Australia and Andrew P Valentine, Australian National University, Research School of Earth Sciences, Canberra, ACT, Australia
An EPIC Condition for Tikhonov Regularization (536474)
Francisco H Ortega Culaciati, University of Chile, FCFM - Department of Geophysics, Santiago, Chile, Mark Simons, Seismological Laboratory, California Institute of Technology, Pasadena, CA, United States, Luis A Rivera, Institut de Physique du Globe de Strasbourg, UMR 7516, Université de Strasbourg/EOST, CNRS, Strasbourg, France and Javier A Ruiz, Departamento de Geofísica, Facultad de Ciencias Físicas y Matemáticas, Universidad de Chile, Santiago, Chile
Seismic wave propagation and inference using deep learning, numerical methods and wave physics (Invited) (549921)
Tarje Nissen-Meyer1, Alexandre Szenicer2, Benjamin Moseley2, Kuangdai Leng1 and Andrew Markham2, (1)University of Oxford, Department of Earth Sciences, Oxford, United Kingdom, (2)University of Oxford, Oxford, United Kingdom
Towards Automatic Identification of Tectonic Signals in GNSS Time Series: A Discontinuity Detector (561701)
Nicole Krumm1, Francisco H Ortega Culaciati1 and Mark Simons2, (1)University of Chile, FCFM - Department of Geophysics, Santiago, Chile, (2)Seismological Laboratory, California Institute of Technology, Pasadena, CA, United States
Outlier-Insensitive Bayesian Inference for Linear Inverse Problems (OutIBI) with Applications to Earth Science (570390)
Hang Yu, Nanyang Technological University, School of Electrical and Electronic Engineering, Singapore, Singapore, Prof. Sylvain Barbot, University of Southern California, Department of Earth Sciences, Los Angeles, CA, United States, Justin Dauwels, Nanyang Technological University, School of Electrical & Electronic Engineering, Singapore, Singapore, Teng Wang, Nanyang Technological University, Singapore, Singapore, Priyamvada Nanjundiah, Earth Observatory of Singapore, Singapore, Singapore and Qiang Qiu, University of Southern California, Department of Earth Sciences, Los Angeles, United States
Efficient Bayesian model selection strategy for tsunami source inversion (570776)
Roberto Benavente, Universidad Católica de la Santísima Concepción, Department of Civil Engineering, Concepción, Chile, Alejandro Urrutia, National Research Center for Integrated Disaster Management (CIGIDEN), Santiago, Chile, Jorge G. F. Crempien, Pontificia Universidad Católica de Chile, Departamento de Ingeniería Estructural y Geotécnica, Santiago, Chile, Jan Dettmer, University of Calgary, Department of Geoscience, Calgary, AB, Canada and Rodrigo Cienfuegos, Pontificia Universidad Católica de Chile, Departamento de Ingeniería Hidráulica y Ambiental, Santiago, Chile
3D seismic data denoising and reconstruction using total variation regularized nonlocal low-rank tensor decomposition (571841)
Ouyang Shao1, Lingling Wang1, Xiangyun Hu2 and Zhidan Long1, (1)China University of Geosciences, Wuhan, China, (2)Hubei Subsurface Multi-scale Imaging Key Lab, Institute of Geophysics & Geomatics, China University of Geosciences (Wuhan), Wuhan, China
Estimation on co- & post-seismic slip distributions through viscoelastic inversion using Reversible-jump MCMC (572079)
Fumiaki Tomita1, Takeshi Iinuma1, Ryoichiro Agata2 and Takane Hori2, (1)Japan Agency for Marine-Earth Science and Technology (JAMSTEC), Research Institute for Marine Geodynamics, Yokohama, Japan, (2)Japan Agency for Marine-Earth Science and Technology (JAMSTEC), Research and Development Center for Earthquake and Tsunami, Yokohama, Japan
Two Dimensional Acoustic Full Waveform Inversion using Discrete Cosine Transform (576612)
Sumin Kim1, Jonghyun Harry Lee2 and Wookeen Chung1, (1)Korea Maritime and Ocean University, Busan, Korea, Republic of (South), (2)University of Hawai‘i at Mānoa, Civil and Environmental Engineering, Honolulu, HI, United States
Estimation of micro-earthquake source locations and source mechanisms based on adjoint wave-equation imaging, with uncertainty analysis (590405)
Chenglong Duan, University of Texas at Dallas, Richardson, TX, United States and David E Lumley, University of Texas at Dallas, Geoscience, Richardson, TX, United States
Exploring the Model Space with Regularized Stochastic Inversion: 2D Magnetotelluric Example from Ordos, North China (591107)
Hao Dong1, Zeqiu Guo1,2 and Alan G Jones1,3, (1)China University of Geosciences (Beijing), Beijing, China, (2)Sichuan University, Chengdu, China, (3)Complete MT Solutions, Ottawa, ON, Canada
Full waveform inversion with a gradient based Markov chain Monte Carlo method and uncertainties quantification, application on Axial seamount (603445)
Zeyu Zhao, Institute for Geophysics, University of Texas Austin, Austin, TX, United States, Adrien F Arnulf, Institute for Geophysics, University of Texas at Austin, Austin, TX, United States; Institute for Geophysics, Austin, TX, United States and Mrinal K Sen, University of Texas at Austin, Institute for Geophysics, Austin, TX, United States
Understanding the encoder-decoder network for solving an unsupervised seismic inversion problem (604638)
Reetam Biswas, University of Texas at Austin, Austin, TX, United States, Debajeet Barman, Baylor University, Waco, TX, United States and Mrinal K Sen, Univ Texas Inst Geophysics, Austin, TX, United States
Mini-batch Least-Squares Reverse Time Migration In A Deep Learning Framework (613308)
Janaki Vamaraju1, Jeremy Vila2, Mauricio Araya-Polo3, Debanjan Datta4, Mohamed Sidahmed2 and Mrinal K Sen5, (1)University of Texas at Austin, Austin, TX, United States, (2)Shell Houston, Houston, United States, (3)Shell International Exploration and Production, Houston, TX, United States, (4)University of Texas at Austin, Institute for Geophysics, Austin, TX, United States, (5)Univ Texas Inst Geophysics, Austin, TX, United States
Including 3D Magnetotelluric Data into Joint Probabilistic Inversions (627401)
María C. Manassero1, Juan Carlos Afonso1, Fabio Zyserman2 and Sergio Zlotnik3, (1)Macquarie University, Sydney, NSW, Australia, (2)CONICET - Facultad de Ciencias Astronomicas y Geofisicas, Universidad Nacional de La Plata, Argentina, La Plata, Argentina, (3)UPC, BarcelonaTech, Barcelona, Spain
