GP005-04
Advances in Magnetotelluric modelling and inversion with SimPEG

Monday, 14 December 2020: 10:23
Virtual
Lindsey Justine Heagy1, Joseph Capriotii2, Johnathan Kuttai3, Devin Cowan2, Fernando Perez4, Joseph Hamman5, Anderson Banihirwe6 and Kevin Paul5, (1)University of California Berkeley, Department of Statistics, Berkeley, CA, United States, (2)University of British Columbia, Department of Earth, Ocean and Atmospheric Sciences, Vancouver, BC, Canada, (3)Dias Geophysical, Vancouver, BC, Canada, (4)University of California, Berkeley, Statistics, Berkeley, CA, United States, (5)National Center for Atmospheric Research, Boulder, CO, United States, (6)National Center for Atmospheric Research, Boulder, United States
Recorded Presentation
Abstract:
The SimPEG project was started in 2013 with the aim of being a flexible toolbox for researchers working with numerical simulations and inversions. Since then, there has been substantial growth in the Python and Jupyter ecosystems for scientific computing. The Pangeo project has been a driving force for the development and improvement of packages for handling large multidimensional datasets (Xarray) and performing parallel computation (Dask) on cloud resources and HPC centers. Technologies in the Jupyter ecosystem have continued to advance as well -- for example, the extension framework in JupyterLab allows for custom components such as rich visualizations to be combined with computation.

In this presentation, we will provide an update on recent advances in natural source electromagnetic methods in SimPEG that benefit from and build upon developments in the open-source Python and Jupyter ecosystems. These include using Dask to parallelize forward simulations, implementing adaptive OcTree meshes to reduce the size of forward simulations, and developing JupyterLab extensions that display visualizations that enable the user to monitor the progress of an inversion.

This work is a part of the broader Jupyter meets the Earth project and is supported by the NSF EarthCube Program under awards 1928406, 1928374.

<< Previous Abstract | Next Abstract >>