Expediting surface wave dispersion curve picking with ML

Apply

Project Description

Surface waves carry useful information about the subsurface, especially of the shear-wave velocities in the near subsurface (upper 100s of meters). Techniques such as the Multichannel analysis of surface waves have been successfully applied in many geological settings on both active and passive seismic recordings. Despite its maturity and ease of use, MASW requires the picking of dispersion curves from dispersion panels; this task is generally automated (eg Allmark), however it requires QC that can be very time consuming. In this project, we aim to investigate the use of deep learning techniques for the field of computer vision to accomplish this task. More specifically, we will train a neural network to learn the direct mapping from the seismic data to their associated dispersion curve by-passing the creation and picking of dispersion panels. The accuracy of our method will be evaluated on both synthetic and real datasets.

Program - Earth Science and Engineering

Division - Physical Sciences and Engineering

Faculty Lab Link - https://mrava87.github.io

Field of Study - Geophysics

About the
Researcher

Matteo Ravasi

Desired Project Deliverables

The candidate will be tasked with: - Creating synthetic datasets containing surface waves using different modeling techniques (analytical solutions, FD-modelling, FE-modelling). - Develop a Machine Learning modelfor the automatic picking of dispersion curves directly from seismic data and compare its performance with other state-of-the-art methods that require picking on dispersion panels - Apply the newly developed method to a field dataset (e.g., USArray)