Interpolation

Using the dips estimated in the first step and the static corrections estimated in the second step, the data are interpolated onto whatever output geometry is desired. Interpolation is performed by a local slant stack, using several neighboring traces. The number of neighbors used need not be the same as in the dip estimation case. In doing the interpolation, it is possible to median filter along the stacking trajectory and/or weight the output by the semblance coefficient obtained in dip estimation (using the semblance measured at the nearest available location). I demonstrated Cole (1993) that median filtering can make the interpolation less sensitive to noise bursts on isolated traces. And the nonlinear semblance weighting can enhance coherent events.