A stable and fast automatic workflow for smooth velocity model retrieval is of interest for subsequent depth migration and as a source of initial models for full-waveform inversion. At the same time, seismic data enhancement by means of stacking is widely performed. The idea of utilizing kinematic wavefield attributes representing an extra stacking output as input data for a fast smooth velocity model inversion was first proposed in a paper of Duveneck (2004b). Common-reflection-surface (CRS) stack attributes (Mann et al., 1999; Jäger et al., 2001) form a data vector for nonlinear least-square optimization. Recently, this method of zero-offset wavefront tomography has found its application to passive seismic source localization (Schwarz et al., 2016) and diffraction imaging (Bauer et al., 2017). The inverse problem was formulated similarly to stereotomography (Lambaré, 2008) with velocity model, diffractors and scattering angles as unknowns. This is a redundant set for the zero-offset wavefront tomography because in contrast to the pre-stack stereotomography the CRS processing produces zero-offset travel times to diffractors before the inversion. To fully exploit this fact we developed a novel kinematic wavefield attributes inversion workflow. It performs an overall minimization of diffracted and normal incidence point waves geometrical speading at emergence time. We refer this procedure as dynamic focusing. During the optimization the wavefield attributes remain constant and serve as initial conditions and travel time for kinematic and dynamic ray tracing. Transition to velocity as the only unknown significantly decreases tomographic matrix dimension and improves a ratio of data points number and a number of unknowns. After a velocity model is retrieved reflectors and diffractors are localized with ray tracing. Fréchet derivatives together with adjoint-state method gradient are adduced. The algorithm was tested on a full of diffrated energy marine dataset from the Levantine Basin (Netzeband et al., 2006).
Presentation Date: Wednesday, September 27, 2017
Start Time: 2:15 PM
Location: Exhibit Hall C, E-P Station 1
Presentation Type: EPOSTER
