Выпуск: 5 , Том: 30 , Год издания: 2021

We present a novel workflow to build a reliable initial velocity depth model from challenging seismic data. This workflow is based on automated 3D grid reflection tomography that utilizes coherent poststack and prestack reflection events in the data domain. The workflow consists of two parts: data preconditioning and nonlinear tomographic inversion. Data preconditioning is underpinned by robust data-driven prestack data enhancement in the form of 3D nonlinear beamforming. Operating directly in the data domain, we obtain robust NMO velocities and pick main reflection events on stacked time images. Ray-based tomographic inversion fits prestack traveltimes approximated by hyperbolae using the engine of standard grid reflection tomography. Powerful prestack enhancement, combined with regularization of observed traveltimes by hyperbolae, delivers a robust and computationally efficient approach to reconstructing the velocity depth model directly in the data domain during the early stages of seismic processing. The new approach enables iterative depth processing critical for low signal-to-noise ratio data such as land seismic with small field arrays or single sensors. We present the tomographic workflow details and showcase the method's capabilities using synthetic and real data examples.