Automation in full-waveform inversion: A way forward

Abstract

Summary We have introduced a new full-waveform inversion (FWI) workflow that combines advanced enhanced template-matching (ETM) objective function with automated data selection and decision making based on quality control (QC) attributes such as travel-time, least-squares, mean time-delay, mean cross-correlation coefficient, etc. QC attributes are derived internally at each step of FWI. QC attributes can be instrumental in driving the automated data selection and providing additional control in implementing relevant objective function parts (travel-time or least-squares) of ETM. As a proof of concept, we have demonstrated the workflow using 2D synthetic Marmousi data and marine narrow azimuth (NAZ) seismic data from the East Mediterranean Sea, offshore Egypt. The results from the synthetic data example show that the workflow can achieve high-quality results with minimum user intervention. The results from the field data example show a robust and geologically consistent velocity update, which further validates the competency of this workflow. The workflow uses minimally pre-processed full record seismic input data and significantly reduced user intervention. The ETM-FWI workflow in conjunction with QC attributes can accelerate the FWI workflow significantly and provide more accurate and data-driven velocity updates in geologically complex areas.