Seismic Synthetics with Devito and GemPy

Transform 2021 workshop Thursday, April 22 (09:00 - 11:00 BST):

• Presented by: Edward Caunt, Imperial College London
• Conference website: https://softwareunderground.org/events/transform-2021

Devito is a domain specific language (DSL) and compiler for finite difference schemes and other stencil computations, with a primary user base among the geophysics community for use in seismic modelling and processing applications. This workshop will provide attendees with an introduction to using GemPy (an open-source 3D geological modelling package) to create synthetic seismic models, which can then be used for wave propagation in Devito. The intention is to highlight the possibilites of creating sharable, editable, and reproducable models for seismic modelling experiments using freely-available tools. A breakdown of the agenda is as follows:

• Creating a geological model in GemPy with seismic properties assigned to each unit
• Bridging the geological model data to Devito
• Forward acoustic propagation using Devito
• Wavefield visualisation in 3D using PyVista

Attendees who would like a more hands-on experience with Devito can follow along on their own local machine. Installation instructions for Devito can be found here, however, the notebook contains the required pip commands for installing any non-standard modules, so you should be able to simply have have a fresh python environment called transform2021_devito with an ipython kernel installed.

You can alternatively follow along using Google Colab (recommended for Windows users). Note that you will need to uncomment and run a couple of additional cells to make the 3D visualisations work, and prevent the notebook crashing.

Workshop prerequisites:

• Basic Python programming knowledge
• Basic geological background
• Basic knowledge of finite differences
• A Google account (if using Colab)

To access the workshop material, simply open the notebook titled creating_synthetics.ipynb in a new python venv (instructions below) or in Google Colab. All necessary installations are contained within the notebook.

Python environment setup:

• In a bash terminal, create the virtual environment with python3 -m venv $PATH_TO_DIR/transform2021_devito where $PATH_TO_DIR is the path to the directory where you want to put your python environment.
• Now activate the environment with source $PATH_TO_DIR/transform2021_devito/bin/activate
• Update pip with pip install --upgrade pip
• Install ipykernel with pip install ipykernel
• Install the kernel into the venv with ipython kernel install --user --name='transform2021_devito'
• Start the notebook with jupyter notebook

Contents:

• creating_synthetics.ipynb - The main notebook which will be covered in the tutorial
• further_reading_fwi.ipynb - Further reading, using a slice of the model created in a toy full-waveform inversion (FWI) example