Challenges in Velocity-Model Building With 3D P-Cable Data
M. Gloeckner, J. Walda, D. Gajewski, T. Roth, D. Klaeschen and C. Berndt
Event name: 81st EAGE Conference and Exhibition 2019
Session: Diffraction Imaging and Modelling
Publication date: 03 June 2019
Info: Extended abstract, PDF ( 281.75Kb )
Price: € 20
P-cable data have the special characteristic of short source-receiver offsets and a high frequency source. These characteristics lead to high resolution data with issues in velocity-model building. Conventional methods can not be applied due to an insufficient source-receiver offset coverage. Point diffractions have the potential to allow velocity-model building without appropriate offset information available. This is the case due to their scattering nature, their moveout in midpoint and offset direction is equivalent. Therefore, velocity information of point scatterer can be extracted in midpoint direction. We present a diffraction based velocity-model building approach with the support of a machine learning algorithm, namely k-means. Multi-dimensional stacking operators, event separation, and wavefront tomography using machine learning for quality control are the key steps in our suggested method. We apply our velocity-model building workflow to 3D P-cable data. A comparison of the inverted velocities with and without machine learning support, shows an improved velocity model with our proposed method. Furthermore, we verify the improved velocity model with a 1D ocean-bottom seismometer velocity profile.