• Intro.
• New Technology
  • Seismic Experiment
  • Phase Shift Migration
  • Reverse Time Migration
• Classic Seismic
  • Physical Background
  • Seismic Acquisition
  • CDP, Stacking
    • Hyperlink 4.3.1
    • Hyperlink 4.3.2
    • Hyperlink 4.3.3
• Hyperlink 5

Reverse Time Migration: Principles

Suppose: we have a reflection at T time.

We reversed our seismic trace and pushed back to the ground by a t_pushed time. There is still a t_togo time remaining, until we reach the time zero.

The T time is unknown and t_pushed can be anything. If we shortened the trace, or extended it formerly, the t_pushed will be different. But it does not matter. What matters, it is that for any t_pushed time, there is an associated t_togo time. Any wave, which went down by the (unknown) t1 time has still t_togo time to come up to the surface. More precisely, it should be the inverse, but it will not change the principles.

.....

So, this is what we have to do:

First thing: we reverse the seismic trace in time.

Next thing: we start to push back the trace into the ground at the physical location of the receiver by small dt time steps. The result will be a depth image of the wave field.

We store the depth image, together with the formerly introduced t_togo time. We continue this pushing and storing procedure, until we reach the zero time.

Next step: we start a source wave at the source position and let it to propagate into the ground. We use the same dt time steps, as we were using during the stepping of the reversed seismic trace. This also gives us a depth image. Each step has its proper timing. Let's call it t_source

Now, we search in the previously stored images. We look for the one, where the stored t_togo time corresponds to the present t_source time. The image generated by the source wave shows, where the wave could go. The stored image of the seismic trace shows, where the wave could have come back from. Matching the two image together we will obtain the possible location or the reflecting point.

Imaging condition

What does it mean, "matching" the two images?

This question opens a wide range of discussions. We will come back to this later on, but one can imagine several ways to do the job. For example to multiply the two images. Or to produce small cross correlations. There might be several ways, one can find them described in the literature. It is a nice topic for academic discussions.

Please notice: the following images are quite large, your browser might be slow ... please wait !