Perfect partial reconstructions for multiple simultaneous sources

TL;DR

This paper mathematically proves that the signal apparition method provides the largest possible regions in frequency-wavenumber space for exact separation of multiple seismic sources, outperforming other encoding methods.

Contribution

The paper offers a rigorous proof that signal apparition yields optimally large regions of exact source separation in seismic data, valid for any number of sources.

Findings

Signal apparition creates diamond-shaped regions of exact separation.

Other encoding methods result in smaller separation regions.

Field tests confirm the theoretical predictions.

Abstract

A major focus of research in the seismic industry of the past two decades has been the acquisition and subsequent separation of seismic data using multiple sources fired simultaneously. The recently introduced method of {\it signal apparition} provides a new take on the problem by replacing the random time-shifts usually employed to encode the different sources by fully deterministic periodic time-shifts. In this paper we give a mathematical proof showing that the signal apparition method results in optimally large regions in the frequency-wavenumber space where exact separation of sources is achieved. These regions are diamond-shaped and we prove that using any other method of source encoding results in strictly smaller regions of exact separation. The results are valid for arbitrary number of sources. Numerical examples for different number of sources (three resp.~four sources) demonstrate the exact recovery of these diamond-shaped regions. The theoretical proofs' implementation in the field is illustrated by the results of a conducted field test.