Machine Learning applications in geophysics

Graphical visualization of a horizon clearly depicting geologic features such as incised valleys, channels, salt domes, and faults formed by the salt dome. Seismic data from the National Archive of Marine Seismic Surveys (NAMSS)

Course advisor - Dr. Sergey Fomel, Dr. Zoltan Sylvester, Dr. Jacob Covault

Collaborator - Landon Lockhart

We explored the following Machine Learning applications in geophysics -

• Regression to estimate sonic well log from gamma ray, resistivity, density, and neutron logs
• Classification model for predicting lithofacies from well-logs and geologic indicators
• Artificial Neural Network approach to pick a horizon automatically based on seismic data
• Convolutional Neural Network in order to classify Earthquakes from noise signals
• U-Net for image segmentation application in seismic fault detection
• Autoencoder for extracting useful patterns in order to pick horizons
• Correlating time series: Dynamic Time Warping and 3D optimization

Many parts of the code and some datasets are confidential, and thus a Github link cannot be shared here. Contact for further information.