Hessian kernels of seismic data functionals based upon adjoint techniques

SUMMARY We present an extension of the adjoint method that allows us to compute the second derivatives of seismic data functionals with respect to Earth model parameters. This work is intended to serve as a technical prelude to the implementation of Newton-like optimization schemes and the development of quantitative resolution analyses in time-domain full seismic waveform inversion. The Hessian operator H applied to a model perturbation δm can be expressed in terms of four different wavefields. The forward field u excited by the regular source, the adjoint field u † excited by the adjoint source located at the receiver position, the scattered forward field δu and the scattered adjoint field δu † . The formalism naturally leads to the notion of Hessian kernels, which are the volumetric densities of H δm. The Hessian kernels appear as the superposition of (1) a first-order influence zone that represents the approximate Hessian, and (2) second-order influence zones that represent second-order scattering. To aid in the development of physical intuition we provide several examples of Hessian kernels for finite-frequency traveltime measurements on both surface and body waves. As expected, second-order scattering is efficient only when at least one of the model perturbations is located within the first Fresnel zone of the Fr´ echet kernel. Second-order effects from density heterogeneities are generally negligible in transmission tomography, provided that the Earth model is parameterized in terms of density and seismic wave speeds. With a realistic full waveform inversion for European upper-mantle structure, we demonstrate that significant differences can exist between the approximate Hessian and the full Hessian—despite the near-optimality of the tomographic model. These differences are largest for the off-diagonal elements, meaning that the approximate Hessian can lead to erroneous inferencesconcerningparametertrade-offs.ThefullHessian,incontrast,allowsustocorrectly account for the effect of non-linearity on model resolution.

[1]  B. Kennett,et al.  Sensitivity kernels for finite-frequency surface waves , 2002 .

[2]  J. Tromp,et al.  Theoretical Global Seismology , 1998 .

[3]  J. Tromp,et al.  Finite‐frequency sensitivity of body waves to anisotropy based upon adjoint methods , 2007 .

[4]  A. Fichtner.  Full Seismic Waveform Modelling and Inversion , 2011 .

[5]  Hicks,et al.  Gauss–Newton and full Newton methods in frequency–space seismic waveform inversion , 1998 .

[6]  F. Dahlen,et al.  Traveltime biases in random media and the S-wave discrepancy , 2004 .

[7]  E. R. Engdahl,et al.  Constraints on seismic velocities in the Earth from traveltimes , 1995 .

[8]  J. Woodhouse,et al.  Complex Shear Wave Velocity Structure Imaged Beneath Africa and Iceland. , 1999, Science.

[9]  O. Ghattas,et al.  A Newton-CG method for large-scale three-dimensional elastic full-waveform seismic inversion , 2008 .

[10]  Kim B. Olsen,et al.  Frechet Kernels for Imaging Regional Earth Structure Based on Three-Dimensional Reference Models , 2005 .

[11]  Carl Tape,et al.  Adjoint Tomography of the Southern California Crust , 2009, Science.

[12]  V. Sakkas,et al.  Heterogeneous crust and upper mantle across southern Kenya and the relationship to surface deformation as inferred from magnetotelluric imaging , 2007 .

[13]  Qinya Liu,et al.  Tomography, Adjoint Methods, Time-Reversal, and Banana-Doughnut Kernels , 2004 .

[14]  M. Toksöz,et al.  Diffraction tomography and multisource holography applied to seismic imaging , 1987 .

[15]  Andreas Fichtner,et al.  Full waveform tomography for radially anisotropic structure: New insights into present and past states of the Australasian upper mantle , 2009 .

[16]  P. Mora.  Elastic wave‐field inversion of reflection and transmission data , 1988 .

[17]  A. Tarantola.  A strategy for nonlinear elastic inversion of seismic reflection data , 1986 .

[18]  Albert Tarantola,et al.  Theoretical background for the inversion of seismic waveforms including elasticity and attenuation , 1988 .

[19]  W. Friederich.  Propagation of seismic shear and surface waves in a laterally heterogeneous mantle by multiple forward scattering , 1999 .

[20]  A. Curtis,et al.  Global crustal thickness from neural network inversion of surface wave data , 2007 .

[21]  G. Nolet.  Partitioned waveform inversion and two‐dimensional structure under the network of autonomously recording seismographs , 1990 .

[22]  Albert Tarantola,et al.  Inverse problem theory - and methods for model parameter estimation , 2004 .

[23]  Jeroen Tromp,et al.  Finite-frequency sensitivity of surface waves to anisotropy based upon adjoint methods , 2007 .

[24]  Carl Tape,et al.  Finite‐frequency tomography using adjoint methods—Methodology and examples using membrane surface waves , 2007 .

[25]  Jorge Nocedal,et al.  On the limited memory BFGS method for large scale optimization , 1989, Math. Program..

[26]  R. Plessix.  A review of the adjoint-state method for computing the gradient of a functional with geophysical applications , 2006 .

[27]  A. Tarantola,et al.  Two‐dimensional nonlinear inversion of seismic waveforms: Numerical results , 1986 .

[28]  Guust Nolet,et al.  A Breviary of Seismic Tomography , 2008 .

[29]  Carl Tape,et al.  Seismic tomography of the southern California crust based on spectral‐element and adjoint methods , 2010 .

[30]  Jean Virieux,et al.  Quantitative imaging of complex structures from dense wide‐aperture seismic data by multiscale traveltime and waveform inversions: a case study , 2004 .

[31]  Florian Bleibinhaus,et al.  Structure of the California Coast Ranges and San Andreas Fault at SAFOD from seismic waveform inversion and reflection imaging , 2007 .

[32]  A. Tarantola.  Inversion of seismic reflection data in the acoustic approximation , 1984 .

[33]  Peter Mora,et al.  Inversion = Migration + Tomography , 1988, Shell Conference.

[34]  A. Fichtner,et al.  Efficient numerical surface wave propagation through the optimization of discrete crustal models—a technique based on non-linear dispersion curve matching (DCM) , 2008 .

[35]  Andreas Fichtner,et al.  The adjoint method in seismology: I. Theory , 2006 .

[36]  Ru-Shan Wu,et al.  Scattering characteristics of elastic waves by an elastic heterogeneity , 1985 .

[37]  F. Santosa,et al.  Computation of the Hessian for least-squares solutions of inverse problems of reflection seismology , 1988 .

[38]  B. Kennett.  On the density distribution within the Earth , 1998 .

[39]  Jeroen Tromp,et al.  Finite-frequency sensitivity kernels for global seismic wave propagation based upon adjoint methods , 2008 .

[40]  J. Tromp,et al.  Finite-Frequency Kernels Based on Adjoint Methods , 2006 .

[41]  Joseph S. Resovsky,et al.  Probabilistic Tomography Maps Chemical Heterogeneities Throughout the Lower Mantle , 2004, Science.

[42]  B. Kennett,et al.  The Seismic Wavefield I: Introduction and Theoretical Development , 2001 .

[43]  Ying Zhou.  Multimode surface wave sensitivity kernels in radially anisotropic earth media , 2009 .

[44]  Alexandre Fournier,et al.  Spherical‐earth Fréchet sensitivity kernels , 2007 .

[45]  Luis Rivera,et al.  On the use of the checker-board test to assess the resolution of tomographic inversions , 1993 .

[46]  A. Tarantola,et al.  Waveform inversion of marine reflection seismograms for P impedance and Poisson's ratio , 1996 .

[47]  H. Marquering,et al.  Diffraction effects upon finite‐frequency travel times: A simple 2‐D example , 1998 .

[48]  J. Tromp,et al.  Constraining large-scale mantle heterogeneity using mantle and inner-core sensitive normal modes , 2004 .

[49]  Gerard T. Schuster,et al.  Wave-equation traveltime inversion , 1991 .

[50]  Jeroen Tromp,et al.  Finite-frequency sensitivity of body waves to anisotropy based upon adjoint methods , 2007 .

[51]  Detecting near-surface objects with seismic waveform tomography , 2009 .

[52]  Jeannot Trampert,et al.  Using probabilistic seismic tomography to test mantle velocity–density relationships , 2003 .

[53]  G. Chavent,et al.  Inversion of normal incidence seismograms , 1982 .

[54]  Po Chen,et al.  Full-Wave Seismic Data Assimilation: Theoretical Background and Recent Advances , 2011 .

[55]  J. Tromp,et al.  Even‐degree lateral variations in the Earth's mantle constrained by free oscillations and the free‐air gravity anomaly , 2001 .

[56]  Andreas Griewank,et al.  Algorithm 799: revolve: an implementation of checkpointing for the reverse or adjoint mode of computational differentiation , 2000, TOMS.

[57]  P. Mora.  Nonlinear two-dimensional elastic inversion of multioffset seismic data , 1987 .

[58]  R. Pratt.  Seismic waveform inversion in the frequency domain; Part 1, Theory and verification in a physical scale model , 1999 .

[59]  Peter Mora.  Inversion = Migration + Tomography , 1988, Shell Conference.

[60]  J. Virieux,et al.  Multiscale seismic imaging of the eastern Nankai trough by full waveform inversion , 2004 .

[61]  A. Devaney.  Geophysical Diffraction Tomography , 1984, IEEE Transactions on Geoscience and Remote Sensing.

[62]  Fuchun Gao,et al.  Waveform tomography at a groundwater contamination site: Surface reflection data , 2007 .

[63]  Thomas H. Jordan,et al.  Full 3D Tomography for the Crustal Structure of the Los Angeles Region , 2007 .

[64]  G. Nolet.  A Breviary of Seismic Tomography: Frontmatter , 2008 .

[65]  Jeannot Trampert,et al.  Fully nonlinear inversion of fundamental mode surface waves for a global crustal model , 2007 .

[66]  S. Operto,et al.  Seismic imaging of complex onshore structures by 2D elastic frequency-domain full-waveform inversion , 2009 .

[67]  Guust Nolet,et al.  Three‐dimensional sensitivity kernels for surface wave observables , 2004 .

[68]  Andreas Fichtner,et al.  Theoretical background for continental‐ and global‐scale full‐waveform inversion in the time–frequency domain , 2008 .

[69]  Stefan Ulbrich,et al.  Optimization with PDE Constraints , 2008, Mathematical modelling.

[70]  Andreas Fichtner,et al.  Full seismic waveform tomography for upper-mantle structure in the Australasian region using adjoint methods , 2009 .

[71]  Max A. Meju,et al.  Regularized extremal bounds analysis (REBA): An approach to quantifying uncertainty in nonlinear geophysical inverse problems , 2009 .

[72]  J. Hole,et al.  Applying waveform inversion to wide-angle seismic surveys , 2009 .

[73]  Thomas Bohlen,et al.  2D elastic full waveform tomography of synthetic marine reflection seismic data, Society of Petroleum Engineers , 2010 .