Aliasing in the tau-p transform and the removal of spatially aliased coherent noise

Abstract

The tau-p transform is a discretized Radon transform. The choice of discretization parameters is a very important part of performing the transform. Insufficient sampling in the tau direction leads to aliasing problems equivalent to those encountered in any one‐dimensional time series. A simple graphical method illustrates that too coarse sampling in the p direction results in reconstructions containing data duplicated incorrectly at different spatial positions. The spacing of these duplications is dependent on the temporal frequency of the data. Insufficient spatial sampling of the original seismic data causes events to plot at multiple p values in tau-p domain, again dependent on temporal frequency. Therefore, to velocity filter spatially aliased noise efficiently, multiple p values must be filtered. The use of appropriate filters in the p-f domain can substantially improve the noise rejection capabilities of velocity filters on spatially aliased noise while having little effect on desired reflection signals.