Conventional stacking velocity analysis is able to estimate the RMS velocity with sufficient accuracy (2%-3%) if the layers are nearly horizontal and if the incidence angles are small. In this case, we can simply apply Dix's formula to get the interval velocities, which is the information we need to identify anomalous variations of the wave velocity.
Although the accuracy of stacking velocity is sufficient, the space and time resolution of the velocity spectra may be quite limited. In stacking velocity analysis, a spatial average is carried out over the acquisition spread, whose length in our case is 750 m (half of the streamer length). Consequently, we are not able to measure directly sharp lateral velocity variations. Along the time dimension, the coherency values are usually averaged within time windows whose length is comparable to that of the seismic wavelets. Seismic wavelets are band-limited signals; therefore, their direct comparison by semblance or cross-correlation carried out in stacking velocity analysis necessarily provides a band-limited estimate.
Tomographic inversion of traveltimes is able to improve the resolution significantly for two main reasons: