Thomas Zwinger is an experienced expert in the field of mechanical engineering, with a distinguished career of over three decades. He currently serves as a senior application scientist at the CSC – IT Center for Science Ltd. in Espoo, Finland, and as an adjunct professor in the Department of Physics at Helsinki University. Zwinger’s work is deeply rooted in high-performance computing (HPC) and its application to geophysical flows. His extensive research includes the development and optimization of continuum models for fluid flow and heat transfer, particularly in the study of ice sheets, glaciers, and avalanches. As one of the principal developers of the state-of-the-art ice sheet modeling software Elmer/Ice, Dr. Zwinger has significantly contributed to advancing the understanding of cryospheric dynamics and their broader implications.
In this interview, part of the Meet the Researcher series, we explore Dr. Zwinger’s crucial role within the ChEESE Project. He shares insights into the challenges and achievements of integrating HPC into glaciology, reflects on the inspiring journey of his research, and discusses how his work enhances the prediction and management of geological hazards across Europe.
What are the main contributions of the CSC- IT Center for Science to the project?
CSC is contributing by implementing of a new model for rapidly rising sub-glacial lake outburst flood as well as activities connected to runs of the CoE on the EuroHPC-JU installation LUMI, which is operated in our Kajaani data centre.
Reflecting on the past year, which moment in your work or the project has been particularly significant for you, and why?
It is hard to pinpoint a moment, as the project has been an interesting and inspiring journey up until now. I would say that the structure of the project pushed us to think about how to improve our code’s (Elmer/Ice) performance. If I have to define a moment, I’d say that the very inclusive attitude of our partners in the CoE towards a completely new topic of glacier hazards at the initial meeting and later conferences was very encouraging for me.
Can you explain how your work on glaciology plays a role in hazard assessment?
Glaciers are ice masses residing at high elevations. Any mass at high elevation that can be released by definition is a natural hazard. Imposed by a warming climate, for glaciers that can be in form of destabilized ice (i.e., the binding to its substrate is lost) manifesting in ice-avalanches or, in case of melting the ice in the build-up of pro- and sub-glacial lakes that can release water and cause flooding down their path. In the case of our studies in Iceland, those glacial lake outburst floods (GLOFS) are rather caused by volcanic activity. Simply understanding of these processes helps us with assessing their hazards. Building quantitative models enables us to map their risks.
Are there any particularly challenging aspects or obstacles that you’ve encountered in your research in ChEESE, and how have you overcome them?
We have a code that has been around for a long time and went through a lot of changes, starting with the introduction of MPI a few decades ago. The current challenge is to adapt the methods of the multi-physics code to utilize accelerators. That kind of challenges make HPC such an interesting field.
Can you discuss any recent achievements or milestones that your team has reached within the framework of the project?
The first year of ChEESE included a performance assessment of the code (POP-audit), which we ran on CPU-only machine and confirmed that previous work on such platforms with Elmer resulted in excellent performance. We hence have a confirmed proper baseline to move to GPU enhanced computing. Recently, we were granted support by HPE within a CoE to port Elmer to AMD GPU’s on LUMI, where the application was highly motivated by activities within ChEESE-2P. So, the project is helping us to improve the code in a larger scope than glacier hazards.
How do you foresee the impact of your research on the understanding and management of geological hazards in Europe?
The improvement on the physics of the types of GLOFs occurring in Iceland (jökulhlaups, in Icelandic) will enable us to better understand their dynamics, which in turn improves the predictability of such events. Improved understanding and accuracy in simulations leads to better abilities in management of natural hazards. Any improvement of code performance in Elmer achieved during ChEESE-2P will aid all the other fields this multi-physics package is applied to (e.g., electrical engine simulation, quantum computer design, climate related ice-sheet simulations).
As an expert in glaciology, what inspired you, and which aspects of your research do you find most exciting?
I like the process to start from a physical problem to a mathematical model and implementation in scientific software, towards optimizing the latter to be run on large computing facilities. It is hard work, bearing long periods of frustration – but the moment when things start to click is indescribably satisfactory.
If you were not a professor or researcher, what career path might you have pursued?
As I enjoy my career as it is, I, frankly, never spent a deep thought on this. Perhaps I would have fixed or ridden bicycles (which I do in my spare time) on a professional basis.
For individuals interested in a career in your field of research, what advice would you offer to guide and inspire them?
Recognise the opportunities life brings and be ready to pick them when they come along, even if it means a 180 degree turn. Initially I was doing something completely different in my basic training in engineering physics, namely neutron interferometry, when I stumbled across a review article on snow avalanche modelling (another aspect of natural hazards in the field of glaciology). As this topic was very close to my hobby of ski-mountaineering (something I really miss after having moved from the Alps to Finland), I decided over night to quit quantum experimentation and chose my master thesis and thereafter my PhD thesis to focus on computational fluid dynamics applied to avalanches, leading to significant contributions to the Austrian avalanche forecast model.
Outside of your scientific work, is there a cultural piece (such as a film, book, or podcast) that has significantly influenced your perspective or approach to your research?
Not one in particular, but a lot of documentaries on ice-sheets and glaciers and also the daily news on climate change is what drives me to improve modelling in that field.
By Aerton Guimarães
From the ChEESE-2P Dissemination Team
