4D FWI: Strategies for addressing the inversion noise

Abstract 4D full-waveform inversion (FWI) is increasingly being used for time-lapse data processing. It provides two main advantages. First, the enhancement in image quality that FWI offers compared to traditional imaging methods is significant, particularly in regions with complex overburden and poor illumination. Second, traditional time-lapse processing workflows are cumbersome, and FWI, being an inversion process, can potentially alleviate some of this processing burden. That said, FWI is a complex, nonlinear, and rank-deficient inverse problem, making 4D FWI workflows sensitive to noise coming from the inversion process, such as the null-space. This article reviews two strategies for addressing the 4D noise originating from the null-space and smaller singular values of the inversion matrix. We discuss the advantages and disadvantages of both strategies and share results from Atlantis, a deepwater subsalt field with nearly two decades of production history.