Interactive velocity analysis

Series Investigations in Geophysics Öz Yilmaz http://dx.doi.org/10.1190/1.9781560801580 ISBN 978-1-56080-094-1 SEG Online Store

With the availability of powerful workstations, efficiency in seismic data analysis at large has increased enormously. Applications that involve numerically intensive computations and large input-output operations are performed using multiprocessor servers, and results are viewed and evaluated using high-performance graphics workstations. Interactive data analysis enables efficient parameter testing that is needed for many of the steps in a processing sequence, such as filtering, deconvolution, and gain. Additionally, interactive analysis provides efficiency in picking events — first breaks on shot records for refraction statics, reflection times on migrated sections, and velocity functions from velocity spectra.

Figure 3.2-26 shows a CMP gather and its velocity spectrum displayed in color. Note the distinct velocity trend with changes in vertical gradient at 600, 1400, and 2000 ms. Color display of velocity spectra is used for interactive picking of semblance peaks, whereas the contour and gated row plots are used for the traditional paper display of velocity spectra. Figure 3.2-27 shows the same velocity spectrum as in Figure 3.2-26 with the velocity picks associated with primary reflections. The hyperbolic traveltime trajectories that correspond to the velocity picks are superimposed on the CMP gather to observe any discrepancy between the modeled and the actual traveltimes, and thus verify the accuracy of the velocity picks. Further verification of the picks can be made by applying moveout correction and examining the flatness of events on the CMP gather as shown in Figure 3.2-27. The undercorrected event at about 800 ms is the water-bottom multiple; this event is represented in the velocity spectrum by the isolated peak to the left of the velocity trend.

Figure 3.2-28 shows the same gather and velocity spectrum as in Figure 3.2-27 but with a velocity function erroneously picked along a high-velocity trend to cause undercorrection of the reflections. Note the discrepancy between the modeled and actual travel-times on the CMP gather before moveout correction, and the misalignment of the reflections after moveout correction. The case of overcorrection caused by erroneously low velocities is demonstrated in Figure 3.2-29.

A localized mispick along a velocity trend yields a physically implausable interval velocity value as ill-strated in Figures 3.2-30 and 3.2-31. The strategy for picking velocities from velocity spectra is based on tracking the velocity trend that coincides with semblance peaks associated with the primary reflections. Whether these reflections are associated with key geological markers or not is irrelevant — as many picks as necessary should be picked so as to honor changes in vertical velocity gradients and thus obtain the best stack. However, picks at too close time intervals can yield anomalous interval velocities from Dix conversion (Section J.4); therefore, they should not be used for deriving interval velocities. (Earth modeling in depth, in addition to Dix conversion, is devoted to several techniques for estimating interval velocities.)

To derive plausable interval velocities from the picked rms velocity functions, first intersect the time horizons picked from the time-migrated volume of data with the velocity functions at analysis locations and extract horizon-consistent rms velocity functions. Then, perform spatial interpolation to derive horizon-consistent rms velocity profiles along line traverses or maps over the survey area. Next, perform Dix conversion of the horizon-consistent rms velocities to interval velocities vint [1]:

 ${\displaystyle v_{int}={\sqrt {\frac {v_{n}^{2}t_{n}-v_{n-1}^{2}t_{n-1}}{t_{n}-t_{n-1}}}},}$ (24)

where vn and vn−1 are the rms velocities at the layer boundaries n and n − 1, respectively; and tn and tn−1 are the horizon times at these layer boundaries. An alternative method for deriving horizon-consistent rms velocities is presented next.