Supplemental material and extensive references for

"Making the Most Out of the Least (Squares Migration)"



Due to space limitation, very few references and only a brief introduction can be accommodated in our article in the special issue on Migration and Imaging, TLE, Sept. 2014.  Here, more extensive references and background material are provided. Only the text that includes such citations, references, and background material are included in this webpage.  We refer the readers to our TLE article for the rest of the text and the figures. 

Theory and background

Least-squares migration (Nemeth et al., 1999; Duquet et al., 2000) is an imaging algorithm (an excellent overview can be found in Schuster (2011)) equivalent to linearized waveform inversion (Lailly, 1984) that seeks to find the best reflectivity image  given a fixed smooth background velocity model. This is achieved by minimizing an objective function


where  is the recorded reflection data,  represents the linearized forward modeling operator, and the regularizer (Wang and Sacchi, 2007) encourages the solution to adopt specified characteristics of . The operator  represents the solution of the wave equation for a smooth background velocity model embedded with a sharp reflectivity distribution represented by m.


The Levenberg-Marquardt solution to equation 1 is formally given as, whereis a damping parameter. Unfortunately, the Hessian inverse  is too expensive to compute or store, so either an iterative solution in the data domain (Nemeth et al., 1999) or an approximate inverse in the image domain (Hu et al., 2000; Yu et al., 2006; Tang, 2009) is employed. In the image domain, the Hessian inverse  is approximated with the migration deconvolution (MD) operator (Hu et al., 2000; Yu et al., 2006). Then the least-squares migration (LSM) image is given by, where the contribution from the regularizer is dropped for brevity. The benefits of the image domain approach include: (1) once the approximate inverse Hessian is obtained, the LSM image is computed at the cost of about one migration; and (2) MD can be computed in a target-oriented manner where the computation is limited to a small target volume of interest. The limitation is that the MD operator is an approximation that assumes a local layered medium around the image point of interest.


In the data domain, on the other hand, the data misfit drives the iterative updates of the trial reflectivity model, such as


where  represents the iteration index,  is the step length, and  is the preconditioner. If this preconditioner is inadequate, a conjugate gradient or quasi-Newton method is used instead for the iterative update. When  and the initial model  is assumed to be smooth (or a reflectivity model where only the direct wave is generated), then  is the migration image computed by standard migration (Lailly, 1984; Mora, 1987). Further iterations will adjust the reflectivity model (not the migration velocity model) to enhance resolution and partly mitigate migration artifacts due to trace aliasing, finite aperture effects, geometric spreading, defocusing, and ringiness in the source wavelet. The aforementioned artifacts in the migration image are largely suppressed by LSM because such artifacts will not lead to minimization of the normed residual. The rest of this article focuses on the data domain implementation of LSM.


LSM can be implemented with Kirchhoff migration, one-way wave equation migration (Kuehl and Sacchi, 1999), or reverse time migration (Plessix and Mulder, 2004; Guitton et al., 2006; Dai et al., 2010; Yao and Jakubowicz, 2012). Typically, Kirchhoff migration and one-way wave equation migration produce fewer artifacts because they only smear received energy along the tentative reflectors coincident with the migration ellipse. In contrast, reverse time migration (RTM) automatically generates upgoing reflections from reflectors, and so the received energy is also smeared along the reflection wavepaths (known as the “rabbit ears”) to give rise to low frequency artifacts (Zhang and Sun, 2008). One way to suppress such artifacts is to smooth the migration velocity model. In the context of least squares reverse time migration (LSRTM), Born modeling is employed which assumes a weak scattering approximation.


Curbing the computational cost


 To reduce the cost of LSM, shot gathers can be encoded and blended together to form one supergather or several sub-supergathers (Romero et al., 2000; Tang and Biondi, 2009; Neelamani et al., 2010; Huang and Schuster, 2012; Dai and Schuster, 2013). To this end, there are at least five strategies for source-encoding: (1) random time series (Romero et al., 2000; Neelamani et al., 2010), (2) random polarity (Krebs et al., 2009), (3) random time shift (Dai and Schuster, 2009; Tang and Biondi, 2009), (4) random frequency division (Huang and Schuster, 2012), and (5) plane-wave encoding (Zhang et al., 2005; Etgen, 2005; Dai and Schuster, 2013; Wang et al., 2013).




Dai, W. and G. T. Schuster, 2009, Least-squares migration of simultaneous sources data with a deblurring filter: 79th Annual International Meeting, SEG, Expanded Abstracts, 2090--2994.

Dai, W., C. Boonyasiriwat, and G. T. Schuster, 2010, 3D multi-source least-squares reverse time migration: 80th Annual International Meeting, SEG, Expanded Abstracts, 3120--3124.

Dai, W. and G. T. Schuster, 2013, Plane-wave least-squares reverse-time migration: Geophysics, 78, S165--S177.

Duquet, B., K. J. Marfurt, and J. A. Dellinger, 2000, Kirchhoff modeling, inversion for reflectivity, and subsurface illumination: Geophysics, 65, 1195--1209.

Dutta, G., K. Lu, X. Wang, and G. T. Schuster, 2013, Attenuation compensation in least-squares reverse time migration using the visco-acoustic wave equation: 83rd Annual International Meeting, SEG, Expanded Abstracts, 3721--3725.

Etgen, J. T., 2005, How many angles do we really need for delayed-shot migration: 75th Annual International Meeting, SEG, Expanded Abstracts, 1985--1988.

Guitton, A., B. Kaelin, B. Biondi, 2006, Least-square attenuation of reverse time migration artifacts: 76th Annual International Meeting, SEG, Expanded Abstracts, 2348--2352.

Hu, J., G. T. Schuster, et al., 2000, Prestack migration deconvolution: 70th Annual International Meeting, SEG, Expanded Abstracts, 984--987.

Huang, Y. and G. T. Schuster, 2012, Multisource least-squares migration of marine streamer data and land data with frequency-division encoding: Geophysical Prospecting, 60, 663--680.

Huang, Y., X. Wang, and G. Schuster, 2014, Non-local means filter for trim statics: 84th Annual Meeting, SEG, Expanded Abstracts (in press).

Krebs, J. R., J. E. Anderson, D. Hinkley, R. Neelamani, S. Lee, A. Baumstein, and M.-D. Lacasse, 2009, Fast full-wavefield seismic inversion using encoded sources: Geophysics, 74, WCC177--WCC188.

Kuehl, H., and M. D. Sacchi, 1999, Least-squares split-step migration using the Hartley transform: 69th Annual International Meeting, SEG, Expanded Abstracts, 1548--1551.

Lailly, P., 1983, The seismic inverse problem as a sequence of before-stack migrations: Conference on inverse scattering: theory and application, 206--220.

Mora, P., 1987, Nonlinear two-dimensional elastic inversion of multioffset seimic data: Geophysics, 52(9): 1211--1288.

Neelamani, R., C. Krohn, J. Krebs, J. Romberg, M. Deffenbaugh, and J. Anderson, 2010, Efficient seismic forward modeling using simultaneous random sources and sparsity: Geophysics, 75, WB15--WB27.

Nemeth, T., C. Wu, and G. T. Schuster, 1999, Least-squares migration of incomplete reflection data: Geophysics, 64, 208--221.

Plessix, R.-E., and W.A. Mulder, 2004, Frequency-domain finite-difference amplitude-preserving migration: Geophysical Journal International, 157, 975--987.

Romero, L. A., D. C. Ghiglia, C. C. Ober, and S. A. Morton, 2000, Phase encoding of shot records in prestack migration: Geophysics, 65, 426--436.

Schuster, G. T., 2011, Seismic imaging, Overview. In: H. Gupta (Eds.), Encyclopedia of solid earth geophysics. Berlin: Springer, pp. 1121--1134.

Tang, Y., 2009, Target-oriented wave-equation least-squares migration/inversion with phase-encoded Hessian: Geophysics, 74, WCA95--WCA107.

Tang, Y., and B. Biondi, 2009, Least-squares migration/inversion of blended data: 79th Annual International Meeting, SEG, Expanded Abstracts, 2859--2863.

Wang, J., and M. Sacchi, 2007, High-resolution wave-equation amplitude-variation-with-ray-parameter (AVP) imaging with sparseness constraints: Geophysics, 72, S11--S18.

Wang, X., W. Dai, and G. Schuster, 2013, Regularized plane-wave least-squares Kirchhoff migration: 83rd Annual International Meeting, SEG, Expanded Abstracts, 3242--3246.

Yao, G. and H. Jakubowicz, 2012, Least-squares reverse-time migration: 82nd Annual International Meeting, SEG, Expanded Abstracts, 1--5.

Yu, J., J. Hu, G. T. Schuster, and R. Estill, 2006, Prestack migration deconvolution: Geophysics, 71, S53--S62.

Zhang, Y., and J. Sun, 2008, Practical issues of reverse time migration-true-amplitude gathers, noise removal and harmonic-source encoding: 70th EAGE Conference & Exhibition.

Zhang, Y., J. Sun, C. Notfors, S. H. Gray, L. Chernis, and J. Young, 2005, Delayed-shot 3D depth migration: Geophysics, 70, E21--E28.