A Multidirectional Deep Neural Network for Self-Supervised Reconstruction of Seismic Data

Abstract

Seismic studies exhibit gaps in the recorded data due to surface obstacles. To fill in the gaps with self-supervised deep learning, the network learns to predict different events from the recorded parts of data and then applies it to reconstruct the missing parts of the same dataset. We propose two improvements to the task: a rearrangement of the data, and a new deep-learning approach. We rearrange the traces of a 2D acquisition line as 3D data cubes, sorting the traces by the source and receiver coordinates. This 3D representation offers more information about the structure of the seismic events and allows a coherent reconstruction of them. However, learning the structure of events in 3D cubes is more complicated than in 2D images while the size of the training dataset is limited. Thus, we propose a specific architecture and training strategy to take advantage of 3D data samples, while benefiting from the simplicity of 2D reconstructions. Our proposed multidirectional convolutional neural network has two parallel branches trained to perform 2D reconstructions along the vertical and horizontal directions and a small 3D part that combines their results. We use our method to reconstruct data gaps resulting from several missing shots in a benchmark synthetic and a real land dataset. Compared to a conventional 3D U-net, our network learns to reconstruct the events more accurately. Compared to 2D U-nets, our method avoids the discontinuities that arise from the 2D reconstruction of each trace of the missing shot gathers.