Seismic data denoising and deblending using deep learning
This addresses the need for more automated seismic data processing for geophysical applications, but it is incremental as it builds on existing deep learning methods.
The paper tackled the problem of removing noise and interference from seismic data, a time-consuming step in processing, by using a deep learning approach with a U-net model incorporating ResNet architecture, achieving moderate success on datasets from different parts of the world.
An important step of seismic data processing is removing noise, including interference due to simultaneous and blended sources, from the recorded data. Traditional methods are time-consuming to apply as they often require manual choosing of parameters to obtain good results. We use deep learning, with a U-net model incorporating a ResNet architecture pretrained on ImageNet and further trained on synthetic seismic data, to perform this task. The method is applied to common offset gathers, with adjacent offset gathers of the gather being denoised provided as additional input channels. Here we show that this approach leads to a method that removes noise from several datasets recorded in different parts of the world with moderate success. We find that providing three adjacent offset gathers on either side of the gather being denoised is most effective. As this method does not require parameters to be chosen, it is more automated than traditional methods.