Researcher Spotlight - Meet Dr. Roland Welter

His research focuses on partial differential equations and dynamical systems. He studies the mathematical foundations of climate modelling, especially how approximations and parameterisations affect the reliability of climate predictions. In addition, he investigates how AI and neural networks can be understood more rigorously instead of acting as a “black box.”

What motivates you to work in this field?

“I am motivated by the mathematics behind fluid dynamics. These equations are extremely complex and studying them has led to major advances in dynamical systems, chaos theory, and partial differential equations.

During my PhD, I was motivated by the beauty of discovery. In this climate science role, my motivation has shifted towards producing results that are useful for improving climate predictions. With this position on a climate science, the motivation has shifted a bit more towards obtaining results of practical value for making climate predictions.”

How has your background and career path shaped your approach to climate science?

“When I started my PhD in mathematics, I already knew that I wanted to combine the exploration of theoretical mathematics with the practical skills and methods needed for climate science.”

What are your primary research interests?

“My primary research interests are partial differential equations and dynamical systems I place great value in the certainty that mathematical proofs provide and enjoy the creativity and insight that such proofs require.  The field of climate science employs a great number of approximations in order to simply the extreme complexity down to the point where the computations become tractable.  I greatly enjoy digging in and examining these approximations with a fine-toothed comb to see how valid they really are, and under which conditions.”

What do you believe are the most pressing challenges in climate science today, and how does your work contribute to addressing them?

“The major challenge in climate science is clearly the politicization of science, the wealthy interests who fund denialism and the resulting budget cuts, censorship and inaction over the results of climate science. This is closely connected to education, justice, human rights, and the strength of democratic institutions.  In terms of the purely scientific challenges, as I mentioned before the field of climate science necessarily uses a wide array of approximations and parameterizations.  This can make it difficult to compare different climate models and to understand why model is more accurately capturing certain phenomena. 

Another important purely scientific challenge is the use of AI, which has great potential to improve predictions, but on the other hand operates as a black box.  In this case it is very difficult to know why it is giving good predictions or whether it will still give good predictions in new scenarios such as for a changing climate.”

How does collaboration across different institutions and disciplines enhance the impact of your research?

“Accurately predicting fluid flows on a planetary scale is an unbelievably difficult problem, and one researcher can only feasibly work on at most a few pieces of this problem. I spent a few years during the second phase of the TRR 181 trying to understand how much heat is transferred by a turbulent flow from the bottom of a box to the top. By itself, this information is not that useful, but if contributes to a more accurate climate model then the impact is much larger.”

 In your opinion, what are the biggest opportunities for positive change through climate research and innovation?

“I’ve found that meeting with so many people from all around the world with so much expertise in different areas has been incredibly interesting and enriching. “

Can you share a recent project or discovery that you found particularly exciting or meaningful?

“I’m very proud of the work that I was able to do during the second phase of the TRR 181, in which I was able to dig in deeply to the problem of turbulent heat transfer. Together with my supervisor Pr. Dr. Jens Rademacher, we were able to determine a set of criteria which are necessary and sufficient for energetic consistency in spectral truncations of the Boussinesq-Oberbeck model (a mathematical model commonly used to describe the motion of heat-conducting fluids). I am also excited about my work during phase three, in which I am studying the approximation power of neural networks to perform a so-called balance mode decomposition as well as studying the stability properties of internal gravity waves. However, this work is still in progress.”

What advice would you give to students or young researchers interested in pursuing a career in climate sciences?

“The classic advice to work hard, build connections, and be patient remains very valuable. Although this may sound general, these qualities are often essential for real progress. Today, one possible pitfall is relying too heavily on AI, which I would treat with caution. Hard work is important, but genuine exchange with others is especially valuable, since research often builds on existing knowledge and collaboration. At the same time, networking should be authentic rather than purely strategic. Patience is equally important, because truly new and meaningful results often take a long time to develop.”

What is the best way to connect with you for potential research collaboration?

"The best way to connect with me for a potential research collaboration would be by email, but I’m also on LinkedIn"