Automatic migration velocity analysis via deep learning

Abstract

Migration velocity analysis is a crucial seismic processing step that aims to translate residual moveout (RMO) in common-image gathers (CIGs) into velocity updates. However, this is often an iterative process that requires migration and significant human effort in each iteration. To derive the RMO correction accurately and efficiently, we have developed a new method that combines a newly designed RMO normalization and RMO identification. To make training successful, the former is designed to normalize the RMO from reflectors with different slopes and different depths to a nondipping case. To replace manually picking the velocity spectrum, the latter is arranged to recognize normalized frown and smile patterns in CIGs and translate them into velocity updates via convolutional neural networks. Two numerical and field data examples demonstrate that our method can effectively and efficiently flatten CIGs. This method improves the quality of the velocity updates where there exists manual-picking error in comparison with traditional methods.