Inversion Processing
Figure 1 illustrates the difference between forward and inverse modeling.
Figure 1
In forward modeling, we use sonic log data to calculate reflection coefficients, which
we then convolve with a wavelet to produce a synthetic seismogram.
In inverse modeling, we reverse this process. We begin with the real seismic trace,
calculate the observed spike train and convert this series of reflection coefficients to
a sonic log format.
We do this by solving the reflection coefficient equation defined earlier for velocity:
(1)
If we know the first velocity value in the series, we can calculate all subsequent
velocities iteratively. Once we have calculated velocities for the entire seismic trace,
we will have produced a "sonic log." And if we repeat this procedure across an entire
seismic section, we will generate a synthetic sonic log section. Figure 2 shows this
conversion of a seismic trace to a synthetic sonic log.
Figure 2
In reality, however, the result does not truly represent an actual sonic log for several
reasons. First, the seismic trace has a density component which is factored into our
final data. So, the equation above becomes:
(2)
Our end result is really an acoustic impedance log, which incorporates both velocity
and density data, rather than a sonic log, which incorporates only velocity data.
Second, we must consider the effects of attenuation and geometric spreading.
Because the earth and the recording instruments act as high-frequency filters, our
inversion data are of lower frequency than are the data of a true sonic log.
Third, there is a very low-frequency component of velocity that the sonic log records,
but the seismic method does not. We see this in the sonic log as a steady drift of the
curve toward higher velocities ( Figure 3 ).
Figure 3
In the inversion technique, we must add this low-frequency component into the
synthetic sonic traces before we can truly match the real sonic log curves. Without
the addition of this low-frequency component, the synthetic sonic traces indicate
relative velocity changes rather than absolute changes. We say that the logs are
measuring residual impedance. By adding the low-frequency component we obtain
the total impedance.
And finally, to completely invert our seismic data, we must resolve all of the
components that influence the seismic response, including the shape of the wavelet.
But, it is extremely difficult to eliminate all the propagation effects that contribute to
the seismic trace, and therefore, artifacts of the wave will remain and will
contaminate our synthetic sonic results.
In spite of all these problems, the seismically derived sonic section is a valuable
exploration tool which we can use to see stratigraphic features we could not easily
see on the seismic section.
Synthetic Sonic Display
We usually display synthetic sonic log data in color. This adds dimension to the
normal black and white displays. Generally, we plot the synthetic curves as black
lines on a white background ( Figure 1 ).
Figure 1
Because the curves are closely spaced, we can make detailed correlations from curve
to curve. Contours of isotransit time can also be drawn by computer with a high
degree of confidence. Figure 2 is an example of an isotransit plot.
Figure 2
We then color the area within the contours according to a chosen color scheme, with
different colors corresponding to different velocity values. The use of color increases
our visual dynamic range up to twenty times over a black and white display.
However, we must be careful when we define our color scheme; we must not try to
make the plot "pretty." The purpose of the color scheme is to make the velocity
variations more evident. In doing so, the section may not look as pretty as it would
had we chosen another color scheme, but an "ugly" section often enables us to see
more of the velocity variations that exist in the data.
Our next step is to relate velocity values to lithology. By now we know that lithology
and velocity are involved in a highly complex relationship. By correlating a lithologic
log interpretation to the seismically derived acoustic impedance trace, we can
overlay lithologic symbols on to the color sections ( Figure 3 ).
Figure 3
In this display, we can see a low-velocity shale, indicated by the shale symbol and
colored green, a higher-velocity sand, indicated by the sand symbol and colored
yellow, and an even higher-velocity limestone, indicated by the limestone symbol
and left white.
By overlaying lithologic symbols on to the color sections, we can more accurately
interpret our seismic section.