We have trained deep convolutional neural networks (DCNs) to accelerate the computation of seismic attributes by an order of magnitude. These results are enabled by overcoming the prohibitive memory requirements typical of 3D DCNs for segmentation and regression by implementing a novel, memory-efficient 3D-to-2D convolutional architecture and by including tens of thousands of synthetically generated labeled examples to enhance DCN training. Including diverse synthetic labeled seismic in training helps the network generalize enabling it to accurately predict seismic attribute values on field-acquired seismic surveys. Once trained, our DCN tool generates attributes with no input parameters and no additional user guidance. The DCN attribute computations are virtually indistinguishable from conventionally computed attributes while computing up to 100 times faster.

Deep learning is increasingly being used as a component of geoscience workflows for processing and interpreting seismic data. Training a supervised deep learning network is a data-hungry task: Lots of data examples are needed and they must include labels. The data examples and their labels must have consistent patterns for the deep learning network to learn. Too few examples and/or poor-quality labels can lead to poor deep learning training results. One method to provide large quantities of training examples with high-quality labels is to create synthetic data. We discuss our techniques and experiences with our ongoing use of synthetic seismic data. We share our techniques as an open-source project concurrent with this paper at https://github.com/tpmerrifield/synthoseis. We hope that the geoscience community will share our enthusiasm for developing deep learning geoscience tools and for including synthetic seismic data in supervised deep learning training. We invite contributions from the geoscience community using the open-source model to collectively reduce the realism gap between synthetic data and field seismic data.

No abstract