To streamline fast-track processing of large data volumes, we have developed a deep learning approach to deblend seismic data in the shot domain based on a practical strategy for generating high-quality training data along with a list of data conditioning techniques to improve the performance of the data-driven model. We make use of unblended shot gathers acquired at the end of each sail line, to which the access requires no additional time or labor costs beyond the blended acquisition. By manually blending these data, we obtain training data with good control of the ground truth and fully adapted to the given survey. Furthermore, we train a deep neural network using multi-channel inputs that include adjacent blended shot gathers as additional channels. The prediction of the blending noise is added in as a related and auxiliary task with the main task of the network being the prediction of the primary-source events. Blending noise in the ground truth is scaled down during the training and validation process due to its excessively strong amplitudes. As part of the process, the to-be-deblended shot gathers are aligned by the blending noise. Implementation on field blended-by-acquisition data demonstrates that introducing the suggested data conditioning steps can considerably reduce the leakage of primary-source events in the deep part of the blended section. The complete proposed approach performs almost as well as a conventional algorithm in the shallow section and shows a great advantage in efficiency. It performs slightly worse for larger traveltimes, but still removes the blending noise efficiently.@en

Deep learning has shown a considerable potential to significantly improve processing efficiency but has not yet been widely deployed to production projects of seismic signal separation such as seismic interference attenuation. The main reasons are: First, the industry has high standards for signal fidelity, which are critical for the success of subsequent seismic imaging, and deep neural network methods have not yet matched the required level; second, the network's interpretability issue has affected many geophysicists and sponsors’ trust in the deep learning technique. To develop deep neural network methods towards the end of benefiting real-world production, we first attempt to better understand their performance, especially in how they make use of local and global features of the data. A novel quantitative research of the overall network model behaviour on synthetic data is conducted. We simulate three types of coherent seismic data components in the shot domain, blend them together and then train a network to separate them. In this process, random noise, a component having only learnable local features, is selectively injected into the network's training pairs. Three network models sharing the same architecture are trained individually, and they show distinctive behaviours when applied to the same test data. Step-by-step analysis of each of them reveals that training the network with additional random noise injected into both the input and the output channel of the desired signal can lead to a decent prediction of the coherent noise based on good learning of the global features and, in the meantime, preserve almost all the data information from being lost. We propose this key lesson we learnt as a new method to improve the network's signal fidelity for shot-domain seismic interference attenuation, which is essentially a signal separation task. Its effectiveness is demonstrated on field data from Africa with a comparison to a conventional physics-based seismic interference attenuation method used in production.@en

To separate seismic interference (SI) noise while ensuring high signal fidelity, we have developed a deep neural network (DNN)-based workflow applied to common-shot gathers (CSGs). In our design, a small subset of the entire to-be-processed data set is first processed by a conventional algorithm to obtain an estimate of the SI noise (from now on called the SI noise model). By manually blending the SI noise model with SI-free CSGs and a set of simulated random noise, we obtain training inputs for the DNN. The SI-free CSGs can be either real SI-free CSGs from the survey or SI-attenuated CSGs produced in parallel with the SI noise model from the conventional algorithm depending on the specific project. To enhance the DNN’s output signal fidelity, adjacent shots on both sides of the to-be-processed shot are used as additional channels of the input. We train the DNN to output the SI noise into one channel and the SI-free shot along with the intact random noise into another. Once trained, the DNN can be applied to the entire data set contaminated by the same types of SI in the training process, producing results efficiently. For demonstration, we applied our DNN-based workflow to 3D seismic field data acquired from the northern Viking Graben of the North Sea and compared it with a conventional algorithm. The studied area has a challenging SI contamination problem with no sail lines free from SI noise during the acquisition. The comparison finds that our DNN-based workflow outperformed the conventional algorithm in processing quality with less noise residual and better signal preservation. This validates its feasibility and value for real processing projects.@en