3-D seismic interpretation with deep learning: a set of Python tutorials

Here we are sharing our code, tutorials and examples used to interpret geological structures (e.g. faults, salt bodies and horizones) in 2-D and/or 3-D seismic reflection data using deep learning. The repository is organised in a series of tutorials (Jupyter notebooks) with increasing degree of difficulty. We show step-by-step how to: (1) load seismic data, (2) train a model and (3) apply the model to map different geological structures. You can find a few visual examples on our poster and more technical details in our preprint.

To get started, you don't need any special hardware, software, data or experience - just a bit of time. Check out tutorial-1/tutorial-1.ipyng.

Tutorials

Salt

Tutorial 1

• This tutorial shows you how to map salt in a 2-D seismic image using a 2-D convolutional neural network for pixel-wise classification.

Tutorial 2

• This tutorial describes how to speed up our mapping using U-Net type convolutional neural networks.

Faults

Tutorial 3

• This tutorial shows you how to map tectonic faults in a 3-D seismic volume.

Tutorial 4

• This tutorial will explain how to translate our fault mapping workflow to 3-D.

Tutorial 5

• This tutorial will hows you how to quantify uncertainty during fault mapping (not ready yet)

Horizons

Tutorial 6

• This tutorial introduces you to mapping horizons in a 3-D seismic volume.

Tutorial 7

• This tutorial shows you how to improve horizon mapping using a 2-D CNN

Tutorial 8

• This tutorial will explain how to translate our horizon mapping workflow to 3-D.

Acoustic impedance

Tutorial 9

• This tutorial will show how to invert synthetic seismic data for rock properties (not ready yet)

Citation

If you use this project in your research or wish to refer to the results of the tutorials, please use the following BibTeX entry.

@misc{deepseis2021,
  author =       {Thilo Wrona, Indranil Pan, Rebecca E. Bell, Robert L. Gawthorpe, Haakon Fossen and Sascha Brune},
  title =        {3-D seismic interpretation with deep learning: a set of Python tutorials},
  doi =          {10.5880/GFZ.2.5.2021.001},
  url =          {https://doi.org/10.5880/GFZ.2.5.2021.001},
  year =         {2021}
}