← Back to Sessions

adjTomo: An Open-Source, Python Toolkit for Adjoint Tomography and Full Waveform Inversion

Description: 

Adjoint tomography is a seismic imaging technique that minimizes differences between observed and synthetic waveforms to generate high-resolution images of Earth structure. It is computationally and algorithmically complex, and often requires substantial programming investment to reduce the amount of redundant human effort required during a workflow. Tasks such as wavefield simulations, seismic data processing and interactions with high performance computers may be repeated hundreds to thousands of times during a tomographic inversion. Currently, closed-source software, lack of documentation or loss of active code maintainers make it difficult to leverage existing workflow tools. In an effort to provide a unified set of tools for the entire adjoint tomography community, we present adjTomo: open-source, Python-based tools for adjoint tomography. adjTomo tools tackle the tasks of seismic data manipulation, compute-system interaction, and workflow automation, working alongside existing 2D and 3D numerical solvers (e.g., SPECFEM). The software emphasizes documentation, tutorials and unit testing, which ensure the software can be used, understood and modified by future users and developers. The aim is to foster community research codes which can be adopted and improved by the seismological community. adjTomo packages have been used for published studies, including a real data inversion of an active subduction zone, and numerous synthetic inversion experiments. Here we showcase the software, their supporting material, and associated ongoing research projects.

Session: Numerical Modeling in Seismology: Developments and Applications [Poster]

Type: Poster

Date: 4/20/2023

Presentation Time: 08:00 AM (local time)

Presenting Author: Bryant Chow

Student Presenter: No

Invited Presentation: 

Authors