P/S-wave separation of multicomponent seismic data at the land surface based on deep learning

Abstract

P/S wave separation is a key step for data processing in multi-component seismic exploration. The conventional methods rely on either the prior information of near-surface elastic properties or the carefully selected parameters to estimate the polarization directions of the P- and S-modes when arriving at the geophones. In case of complex wave propagation or the near-surface elastic parameters unavailable, the conventional methods fail to achieve a high-quality P/S separation. The P/S wave separation of multi-component seismic data recorded at land surface can be regarded as a nonlinear problem of point-by-point image prediction, and addressed by utilizing the powerful ability for feature extraction of a deep neural network. To obtain an accurate, abundant and representative labelled dataset, a variety of elastic models are constructed through geologically reasonable data augmentation, followed by the elastic forward modeling and Helmholtz decomposition. A convolutional neural network is trained with these labelled data to predict scalar P- and S-wave recordings from the multi-component data in a data-driven manner. Eventually, a novel approach of P/S wave separation is established for multi-component seismic imaging with the help of deep learning, which can work without the prior knowledge of the near-surface elastic parameters. Numerical examples show that, by constructing an augmented labelled dataset for training, the proposed method significantly improves the generalization of the deep neural network, and thus enable the trained neural network to achieve more accurate P/S wave separation than the conventional model-driven method for the target data.