In seismic processing trim statics and Radon demultiple are applied as post-migration gather conditioning and improve subsequent seismic inversion and interpretation. We present an end-to-end, deep learning-based alternative to popular classical algorithms. Our method is a unified approach to both tasks with the only difference that the same deep neural network architecture is trained on different synthetic data sets specific to the task at hand. In the case of trim statics our synthetic data models the conversion of Common Depth Point (CDP) gathers to CDP gathers with aligned primary energy. Similarly, for the demultiple process, our generated data models the transformation of CDP gathers to their multiple-free counterpart. We observe an excellent performance of our trained networks when applying them to a large diversity of complex field data. Further, the application of our parameter-free method is particularly simple because our networks are solely trained on synthetics without the need for data-specific adjustments, and since they can be interpreted as image-to-image transformations which directly produce the desired output.
Presentation Date: Monday, October 12, 2020
Session Start Time: 1:50 PM
Presentation Time: 3:55 PM
Location: Poster Station 13
Presentation Type: Poster
