| . |
Research Highlight
ModEM - A Modular System for Electromagnetic Geophysical Inversion
ModEM (Modular Electromagnetic Inversion Software) is an advanced, highly flexible computational framework designed for solving complex electromagnetic (EM) inverse problems in geophysics. The core philosophy behind ModEM (Egbert & Kelbert, 2012; Kelbert et al., 2014; Kelbert et al., 2026) is a fine level of modular granularity, ensuring that the fundamental building blocks of an inversion—such as forward modeling, sensitivity computations, search algorithms, and data functionals—are entirely interchangeable, reusable, and extensible (see Kelbert et al., 2014). By using generic interface layers and a specialized multi-transmitter/multi-receiver data structure, ModEM easily accommodates complex data types and acts as a natural platform for joint EM inversion. [full article]
Anna Kelbert1, Miles Curry2, Hao Dong3, Gary Egbert4
1Center for Astrophysics | Harvard & Smithsonian, 2Inalab Group, 3China University of Geosciences, Beijing, China, 4 Oregon State University
From HQ
AI tools are now part of every conversation about scientific software. A researcher can ask a model to draft a finite-element solver before lunch and reasonably expect to come back to a working prototype. The interesting question for an organisation like CIG is not whether this changes the landscape (of course it does), but what is our place in the new landscape. The answer is clearer than it might first appear, and I want to set it out here.
On the Underworld3 team, we recently published a candid account of trying to use AI assistants to accelerate development of the next-generation Underworld code (underworldcode.org). Our finding was that the friction we encountered was diagnostic: where the AI struggled, there were problems with our code design — inconsistent APIs, unclear conventions, missing specifications. Fixing those issues made the code dramatically better for human developers, AI assistants, and users too. As we put it at the time, “if AI tools are struggling with your code, listen: the problem is probably real, and fixing it pays off well beyond AI”. That lesson generalises.
What AI changes is the cost of producing runnable code. What AI does not change (and cannot be expected to supply on its own) is trust in that code. Confidence that a code solves the equations it claims to solve, that the methodology it encodes is proven and reliable. Confidence that it has been benchmarked against community problems is maintained by people who understand the purpose and the trade-offs. Confidence that it produces results a researcher can defend in a paper. In an environment where anyone can generate code that looks plausible, the scarce and valuable thing is verified, curated, community-trusted scientific software.
This is what CIG has always provided, and it is what I think we should now say plainly. CIG’s role is to be a trusted provider of high-quality geodynamics software. That means software that is fit for purpose for our community, that meets the standards of reproducible science, and that is built to work well in an AI-assisted research environment. AI-readiness is not a separate agenda from quality; it is the same agenda. Clear APIs, machine-readable specifications, tested behaviour, and good documentation serve human users, AI assistants, and the scientific record equally.
This restatement aligns with the priorities our community has already identified: our recent TNG workshop’s call for AI/ML integration; the NSF review of our activities emphasising how we need to communicate CIG’s broader impacts and focus on building partnerships; and the strategic work we’ve done on long-term sustainability. Being the trusted, AI-ready provider of fit-for-purpose geodynamics software is how we deliver on all three at once. It is also the most honest answer to the question of why CIG should continue to exist: verified, community-curated infrastructure becomes more valuable as generation becomes cheaper, not less.
We would like to hear from you about what this should mean concretely. What testing standards, documentation conventions, benchmark suites, and provenance practices would make our codes genuinely trustworthy and genuinely AI-ready for the science you are trying to do? CIG’s strength has always been its community, and this is a conversation that needs all of us.
Louis Moresi, Co-Director, CIG
Professor, Australian National University
News
Honors
The National Academy of Sciences announced its 2026 election of its newest members, electing 120 members and 25 international members in recognition of their distinguished and continuing achievements in original research. Congratulations to all of the newest members and a big shout out to Professor Cynthia Ebinger, the Marshall-Heape Chair in Geology, Department of Earth and Environmental Sciences, Tulane University, New Orleans. Cynthia, a leading researcher in earthquakes and tectonics, has been a long-time member of the CIG family, lending her expertise and wisdom on numerous committees.
Professor Ebru Bozdag, Colorado School of Mines, is the recipient of this year's International Union of Geodesy and Geophysics (IUGG) Vladimir Keilis-Borok Medal in mathematical geophysics. The Vladimir Keilis-Borok Medal of the IUGG Commission on Mathematical Geophysics (CMG) has been established by the IUGG Bureau in 2021, and recognizes middle career scientists who have made important contributions to the field of mathematical geophysics. Ebru is a SPECFEM developer who has contributed to the development of cutting-edge methodologies in adjoint tomography in imaging the Earth's interior. Congratulations Ebru!
Governance
Speaker Series
Join us in congratulating our 2026-2027 CIG Distinguished Speakers:- Slow and not steady: What can the spectrum of fault slip tell us about earthquake processes?
Asst. Prof. Kathryn Materna, University of Colorado, Boulder - A tour of the deep Earth through geodynamic modeling
Dr. Cian Wilson, Carnegie Science
Apply to Host a Speaker
The CIG Speaker Series seeks to promote computational modeling in geodynamics and related Earth science disciplines. The series aims to bring computational geodynamics speakers to institutions that may not otherwise have access to speakers with expertise in computational science or computational geophysics. By doing so, we aim to connect speakers and CIG with audiences from a variety of STEM domains and to broaden participation in CIG and computational geodynamics. Institutions interested in hosting a Speaker in 2026-2027 should apply by August 31, 2026.
See the website for more information.Events
Exploring Thermodynamic Computational Models in Geodynamics
The tectonic and magmatic evolution of the solid Earth is fundamentally coupled to thermodynamic and petrologic processes occurring at vastly different scales. With the increasing sophistication of computational models in both mineral physics and geodynamics, our ability to simulate the coupled feedbacks between these processes has revealed new dynamics and set the stage for significant advances over the next decade.
In this special half-day webinar on Monday, May 11, from 8 AM to 12 PM PDT, we bring together authors of thermodynamic modeling codes to outline the fundamental principles of the basis of their software and describe its structure to better enable incorporation in geodynamic models.
Register here. For more information see our website.
2027 CIG Community Meeting - Call for Organizing Committee Members
The 2025 CIG Community meeting was a great success and there has been overwhelming enthusiasm for regular geodynamic
community meetings. So please save the date for the next 2027 CIG Community Meeting, to be held in person in Santa Fe, New Mexico
in week of 26 July 2027. We are looking for community volunteers to help plan the meeting.
If you are interested in helping organize the plenary talks, tutorials and breakouts,
please contact us.
Join Co Chairs Louis Moresi and Alice Gabriel in developing an exciting event
to engage the community in the latest research and cutting edge topics.
Workshops and Meetings
2026| June 8-12 | Crustal Deformation Modeling Workshop |
| June 21-16 | Ada Lovelace Workshop on Modeling of Mantle and Lithospheric Dynamics |
| July 11-18 | Rayleigh Hackathon |
| July 20-31 | ASPECT Hackathon |
The Computational Infrastructure for Geodynamics is supported by the National Science Foundation under NSF 2149126.