The primary purpose of this section is to provide a simple understanding of why certain parts of early digital processing techniques did not image all of the Earth's structures. This, in effect, is an interpretation issue. We will see that what might be considered easily imaged events are sometimes totally invisible in the seismic record. In many such cases, parts of the subsurface structure may be invisible simply because we have not applied the most accurate available technique to image it. In other cases, its absence may be due to improper noise suppression techniques applied during the preprocessing steps. Whatever the cause, the idea is always to be able to understand what approach produces the best image.
As we will see, migration can be split into four conceptual pieces. As a rule of thumb, these four pieces will help us understand what migration is and how it naturally completes the imaging process.
When the Earth's velocity has very little lateral variation, these four operations can be split apart and applied in any desired order. The most familiar order is NMO, DMO, stack, and finally migration. However, when the velocity is almost constant, it is quite possible to use the order DMO, migration, NMO, and stack. Full prestack migration can be thought to have the order DMO, NMO, imaging, and stack, but in reality the sequence DMO, NMO, and imaging is usually done in one giant process.
Initial attempts at subsurface imaging forced the geophysicist to join the Flat Earth Society. For a constant velocity medium, Figure 35 shows that Greek mathematics can be used to provide the total travel time, t, from a surface source to a flat subsurface reflector and back to a surface receiver. This is done in terms of the two-way vertical travel time, t 0, from the midpoint, M, to the reflector and back to the surface. Neither the velocity, v, nor the vertical or "zero-offset" traveltime is usually available directly, so redundant source and receiver configurations must be used to estimate the traveltimes. For most acquisition geometries, redundancy is usually sufficient to simultaneously estimate both t 0 and v. The subsurface image point, I, is usually referred to as the common-depth-point (CDP). The common depth point is the halfway point in the travel of a wave from a source to a flat-lying reflector to a receiver. When we know the velocity, the arrival at time t and offset h can be moved to time t 0. This process is usually called normal moveout correction (NMO). After NMO, all traces with a common-midpoint or CDP are summed to remove the redundancy and produce a zero-offset trace. However, for our purposes, the important thing is that this vertical time shift is the first step in formulating a prestack approach to imaging. The shift corrects to the arrival time consistent with coincident sources and receivers. After NMO, the result is as though the source, S, and receiver, R, were located at the midpoint, M.
The traces in the CDP gather of Figure 36 all have the same midpoint. When the subsurface reflectors are all flat, the hyperbolic curve in red defines the appropriate velocity to use to correct the data to zero offset time t 0. A modern computer easily fits the data and provides the graphics to estimate both the vertical traveltime and the velocity. The vertical traveltime, t 0, in this figure is extremely important. Keep this in mind as the book continues.
The shots in Figure 37 are from the pyramid model in Figure 38.
Figure 39 is a stack of the common-midpoint ordered data in Figure 37. NMO was performed using the root-mean-square velocity from the model used to generate the data. The "noise" in this data set is representative of a poor implementation of the approximations to the differential equation used to model the data. This kind of noise is related either to the fact that the differences have not been approximated well, or because damping at the boundaries is poor.
![]() |
Figure 40 shows how dip affects arrival times as a function of half-offset.
To relate the traveltime from source at m