Next: Velocity-Depth Trend
Up: Alvarez: Velocity inversion
Previous: Inversion Constraints
The top left panel of Figure 2 shows a comparison
between the synthetic trace obtained from the real sonic log,
and the one obtained with the genetic algorithm using the L1 norm. The
top right panel shows a similar comparison between the original and the inverted
log. Although the match of the seismic traces is not perfect, the main features
in the original trace were remarkably well recovered. The result for the sonic
logs, however, is not so auspicious and the lack of the general
velocity-depth trend makes it very difficult to evaluate the ``goodness'' of the
match. If we forget about the trend and examine the details in the logs, we can
see that most features of the log were indeed recovered, but the overall
result is not satisfactory. The bottom panels show the same comparisons
using the L2 norm. The results are similar: good match of the seismic traces
and poor match of the sonic logs because of our inability to recover the
velocity-depth trend.
L1L2_inv_no_trend
Figure 2 Comparison of the synthetic
trace and the inverted trace (top left panel) and the original log and the
inverted log (top right panel) using the L1 norm. The bottom panels show
similar comparisons using the L2 norm.
Next: Velocity-Depth Trend
Up: Alvarez: Velocity inversion
Previous: Inversion Constraints
Stanford Exploration Project
11/11/2002