PS Decomposition of Isotropic Elastic Wavefields Using CNN-Learned Filters
W. Wang and J. Ma
Event name: 81st EAGE Conference and Exhibition 2019
Session: Poster: Multi-Component Seismic A / Simultaneous Sources A
Publication date: 03 June 2019
Info: Extended abstract, PDF ( 666.29Kb )
Price: € 20
Imaging of multi-component seismic data requires separation/decomposition of P- and S-waves. Traditional separation methods damage the amplitude and phase information of the input wavefield. We propose a set of spatial filters to perform PS decomposition of isotropic elastic wavefield, which preserves the vector information of the input wavefield. The spatial filters are transformed from the wavenumber domain multipliers, and tuned with a convolutional neural network to improve accuracy. The spatial filters are robust and efficient to be applied. Tests with synthetic data show satisfactory PS decomposition results.