Domain adaptation-based acoustic impedance estimation with seismic data constraint

Abstract

Acoustic impedance is one of the important seismic attributes for reservoir characterization. Recently, with the successful implementation and high performance of the machine learning (ML) in various fields, it has inspired geoscientists to carry out various studies to predict impedance using ML algorithms. Both previous ML-based and conventional pre-stack impedance inversion methods usually use well log data to predict impedance. Therefore, with those impedance inversion techniques, the results in the area around the well have high accuracy. However, the results far from the well have relatively low accuracy. To overcome this limitation, we have proposed a domain adaptation-based impedance inversion method. Domain adaptation is a ML method that can be applied not only to the source domain (domain with labeled data) but also to the target domain (domain without labeled data). Therefore, we have applied the domain adaptation technique to predict the acoustic impedance more accurately in the area not only around the well but also away from the well. Furthermore, to generalize the ML model, the reconstruction process of seismic data was added as a constraint and a pseudo labeling strategy was applied. The proposed method was validated using field data acquired from Carnarvon basin in Western Australia. The developed ML model showed promising results for impedance predictions both near well and away from well outperforming the conventional pre-stack inversion method. Therefore, the developed ML method using domain adaptation can be applicable in the preliminary assessment of reservoir in the areas far from the existed wells.