Conclusions

The proposed velocity estimation method performs well in the test cases presented. Estimates are robust for areas of reasonable data quality. In these areas enough information is available for the median stacking routine to eliminate poor data points, and gently smoothing the estimate in time gives a good estimate of velocity. In poor data quality areas, or at early times with little offset information, estimates are less reliable and unstable. Although estimates are not always exact in these areas, they do provide a decent starting estimate of velocity without velocity spectra analysis or manual picking.