4D full-waveform inversion detects time-lapse reservoir changes at King Field, Gulf of Mexico, USA

Abstract We applied a novel 4D full-waveform inversion (FWI) approach to improve the 4D signal and overall imaging of a Miocene sandstone reservoir at King Field, Mississippi Canyon, Gulf of Mexico. To demonstrate the feasibility of 4D FWI, we produced a proof-of-concept benchmark by applying conventional 4D processing as well as 4D FWI to a pair of closely matching narrow-azimuth (NAZ) base and monitor surveys. We conducted 4D FWI in a multiscale manner, progressively increasing the frequencies from 4 to 50 Hz. Through testing various 4D FWI methods, joint 4D FWI with a modified enhanced template-matching objective function was found to provide the most robust framework for minimizing acquisition-related artifacts and detecting time-lapse signals. This approach produced results very similar to legacy conventional 4D processing but with several potential advantages. For example, 4D FWI rendered better stratigraphic resolution, FWI-derived reflectivity (FDR) volumes improved imaging of events near steep salt flanks, and processing times were shortened by 30% to 50%. In addition, we applied 4D FWI to ocean-bottom node (OBN) and NAZ streamer surveys in the same field to examine the viability of this approach for detecting time-lapse signals between nonmatching seismic surveys. For comparison with conventional 4D processing of these nonrepeatable surveys, we employed an optimal selection workflow to reduce OBN azimuths and traces to better match the NAZ data. The resulting 4D difference maps are comparable, suggesting that 4D FWI is a viable methodology for revealing time-lapse changes across different survey types. Furthermore, this study reveals that 4D FDR volumes have the potential to enhance 4D signal detection in steeply dipping subsalt reservoir regions where conventional migration-based methods encounter significant challenges.