Matching Pursuit Fourier Interpolation Using Priors Derived from a Second Data Set

Abstract

Seismic data are typically irregularly and sparsely sampled along the spatial coordinates, leading to suboptimal further processing. Matching pursuit Fourier interpolation (MPFI) is a beyond aliasing interpolation technique for single component seismic data. The anti-aliasing capabilities of the method can be improved by using priors, which are typically derived from the lower frequencies in the data, and used to de-alias the higher frequencies. In this paper we investigate using a prior derived from a separate, more densely sampled data set. Practical examples are “dense-over/sparse-under” data and time-lapse data. Tests are done by decimating an existing dataset, deriving the prior from the non-decimated data, and using the priors for the interpolation of the decimated data. It is shown that using priors from a second data set can give a significant uplift in data reconstruction compared with deriving the priors in a conventional way. In particular, some steeply dipping diffraction events are reconstructed better, and a reduction of artefacts is observed.