Interpolation and Denoising of Seismic Data using Convolutional Neural Networks

Seismic data processing algorithms greatly benefit, or even require regularly sampled and reliable data. Therefore, interpolation and denoising play a fundamental role as starting steps of most seismic data processing pipelines. In this paper, we exploit convolutional neural networks for the joint tasks of interpolation and random noise attenuation of 2D common shot gathers. Inspired by the great contributions achieved in image processing and computer vision, we investigate a particular architecture of convolutional neural network known as U-net, which implements a convolutional autoencoder able to describe the complex features of clean and regularly sampled data for reconstructing the corrupted ones. In training phase we exploit part of the data for tailoring the network to the specific tasks of interpolation, denoising and joint denoising/interpolation, while during the system deployment we are able to retrieve the remaining corrupted shot gathers in a computationally efficient procedure. In our experimental campaign, we consider a plurality of data corruptions, including different noise models and missing traces' distributions. We illustrate the advantages of the aforementioned strategy through several examples on synthetic and field data. Moreover, we compare the proposed denoising and interpolation technique to a recent state-of-the-art method.