In many seismic processing applications, such as filtering and interpolation, the input data is partitioned into small overlapping blocks that are processed separately and then merged to construct the final output data. Tapering the different blocks before merging is done to account for the overlap and to avoid visible artefacts related to the imprint of the blocking. While tapering is important, the design of blocking tapers is often done in a simplistic way. I propose a method to design any N-dimension blocking taper with any amount of overlap. The proposed taper has the property that its aggregate sum across the data section is one, thus fold computation and normalization is avoided. In addition, the taper has some useful spectral features so it can also be used as a pre-Fourier transformation taper when the processing involves such transformation.

You do not currently have access to this article.