Chapter 9: Seismic Imaging

Published:January 01, 2007
In this chapter, we describe quantitatively the actual asymptotic inverse process for diffractionstack migration – namely, isochronstack demigration. We show that the diffractionstack integral (DSI) and the isochronstack integral (ISI) constitute an asymptotic transform pair that is interlinked well by the duality theorems that were derived in Chapter 5. This transform pair can be used to solve a multitude of trueamplitude targetoriented seismicimaging (or imagetransformation) problems, including, e.g., the dynamic counterparts of the kinematic maptransformation examples that Chapter 2 discussed qualitatively. All imagetransformation problems can be addressed by applying the two stacking integrals in sequence, whereby the macrovelocity model, the measurement configuration, or the ray code of the considered elementary reflections may change from step to step. Alternatively, the two stacking procedures can be combined mathematically into a single process, which leads to weighted (Kirchhofftype) summations along certain stacking surfaces (or inplanats). This chapter provides the general formulas for the stacking surfaces and the trueamplitude weights of these new Kirchhoff processes. To demonstrate the value of the proposed imaging theory, which is based on analytically chaining the two stacking integrals, we solve the trueamplitude configurationtransform and remigration problems for the case of a 3D isotropic, laterally inhomogeneous medium.
In this important chapter of our book, we present the theoretical background for the geometrically motivated mapping and imaging concepts that were discussed in Chapter 2. The diffractionstack theory, as presented in Schleicher et al. (1993a) and discussed in detail in Chapter 7 of this book, serves as our point of departure for presenting the theory of trueamplitude isochronstack demigration along very similar lines. These Kirchhofftype migration and demigration operations provide the basis for the unified approach to seismicreflection imaging (Hubral et al., 1996a; Tygel et al., 1996) that is the central subject of this book. In the last section of this chapter, we show how to chain the DSI and the ISI to solve various seismicimagetransformation problems. As in the papers just cited, we present this unified theory here in the time domain. Bleistein and Jaramillo (2000) and Bleistein et al. (2001) discuss an equivalent frequencydomain theory.
Figures & Tables
Contents
Seismic TrueAmplitude Imaging
From Volume Editor’s Preface by Robert H. Stolt
Seismic TrueAmplitude Imaging is a raytheoretical exposition of seismic imaging processes, unapologetic in that it bristles at any suggestion that ray theory is not wave theory. This attitude is justified in part by being right – ray theory does come from an asymptotic approximation to the wave equation – and by the fact that similar asymptotic assumptions tend to sneak like viruses into socalled fullwavetheoretical processes as well.
On other issues, the book is more accommodating. After noting that the original Kirchhoff diffraction integral was devised only for forwardwave propagation, the authors concede that the term Kirchhoff migration has passed into common usage, and they learn to live with it. Similarly, although the term true amplitude is a lexicological minefield, it is widely applied to processes that seek to faithfully preserve amplitude information. The authors are comfortable with that, although they apply a very specific definition of the term in their text.
This book uses the term imaging in the widest possible sense, which was an eyeopener for me. I have tended to use imaging as a synonym for migration, but I can’t do that anymore. The authors note that seismic data in any stage of processing is likely to contain discrete events or images, perhaps geometrically distorted but nevertheless pictures or images of the earth’s interior. This observation is facilitated by the raytheoretical point of view, because it is largely in asymptopia that such images form. In consequence, any operation that affects or changes the images in the data can be considered an imaging process. This brings under the imaging umbrella a variety of processes, including partial and residual migration, conversion of one experimental configuration into another, and so on.
Strictly, I suppose, Kirchhoff imaging employs a raytheoretical model for propagation and a farfield diffractive model for reflections. It is best suited for a medium composed of regions where earth properties change slowly, divided by reflecting surfaces where properties change abruptly. Where those conditions are met imperfectly, the method may tend to impose this model, because this is how the method views the world. That is not necessarily bad – geophysicists tend to view the world that way too.
Kirchhoff depth migration has enjoyed a long period of preeminence. Advances in computer power in recent years have allowed contemplation of fullwaveequation depthmigration algorithms, but the authors of Seismic TrueAmplitude Imaging show no deep concern for the future of raytheoretical imaging. In that, they are probably justified, for several reasons. First, migration methods, once established, never die. The inverse seismic problem is so tough and complex that no single technique, however powerful, can be universally successful. New techniques may be added to the arsenal but are not likely to completely displace proven methodologies.
Second, the asymptotic approximation is very well met under most circumstances by seismic data, and even when that is not the case, raytheoretical techniques often are able to extract useful information. Third, ray theory is well suited to trueamplitude processing because amplitudes can be computed explicitly at every point and related back to the underlying earth properties. Fourth, Kirchhoffbased imaging is uniquely suited to composite operations (e.g., demigration using one velocity structure followed by migration using another velocity structure) because, in asymptopia, a composite operation can be condensed into a single operation.
For extremely complex geology, ray theory might have difficulty providing a complete description of wave propagation, even where it is technically possible to do so. A fullwaveequation method, in contrast, may provide a complete description without extra effort. That might give some advantage to a fullwave method, but the blessings are mixed. In ray theory, one knows exactly which waves are where. If some portion of the complete waveform (perhaps a multiple or a converted wave) does not contribute positively to the desired image, a raytheoretical method might eliminate it. Even where a fullwave method can produce the better image, one would likely want to run a Kirchhoff algorithm concurrently to aid in analysis.
Seismic TrueAmplitude Imaging provides a clear, readable, and reasonably complete presentation of Kirchhoff imaging theory. Although subjects such as beam forming and multipath imaging are not presented in detail, the tools to deal with them are present. For those of us not steeped in ray theory, the book provides a good introduction and tutorial, then digs deeply and profoundly into a theory of generalized imaging.