Low-rank representation of extended image volumes: Applications to imaging and velocity continuation

Abstract

In this paper, we present a cost-effective low-rank factorized form of computing the full subsurface extended image volumes without explicitly calculating the receiver wavefields. The propose approach is computationally feasible, which exploits the low-rank structure of full subsurface extended image volumes organized as a matrix, thus avoiding the customary loop over sources. Using carefully selected stylized examples, we show how conventional migration images can be extracted from the low-rank factorized form. We also present a velocity continuation procedure that uses the same low-rank form and allows us to map the extended image for one velocity model to another velocity model without recomputing all the source wavefields.