Researcher Spotlight - Meet Erfan Mahmoudi

What motivates you to work in this field?

“I’m motivated by the challenge of improving climate predictions. My background in fluid dynamics helps me understand how the atmosphere works, and my interest in machine learning gives me the tools to model complex processes more effectively.”

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

“Although I am early in my career, trained in fluid dynamics, I have always been interested in understanding the key processes in the atmosphere. More recently, developments in machine learning have motivated me to explore how these methods can be integrated into climate modelling.”

What are your primary research interests?

“My primary research interests focus on atmospheric gravity waves and their parameterization in climate models. I am particularly interested in leveraging machine learning techniques - including dimensionality reduction methods, kernel methods, and deep learning - to improve the representation and prediction of these processes.”

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

“One of the biggest challenges is representing small-scale atmospheric processes in global climate models. Gravity waves are a good example: they transport momentum and energy and can strongly influence large-scale circulation.

In my research, I focus on how these gravity waves are generated. There is no physics-based model for their sources. I develop parameterizations that can be used in ray-tracing models, which are tools that simulate how gravity wave energy travels through the atmosphere, tracking where it moves, where it breaks, and how it affects winds and temperatures at different altitudes, so the impact of gravity waves can be represented and predicted more accurately."

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

“Collaboration is essential in my work because it lies at the interface of machine learning and atmospheric science. Working with atmospheric scientists provides access to high-quality and physically accurate data, while collaboration with statisticians and machine learning experts helps improve the modelling and analysis methods. This combination leads to more reliable models and results that are both scientifically sound and technically robust.”

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

“A recent project I found particularly exciting involved applying Principal Component Analysis (PCA) to the gravity wave wave-action density spectrum. PCA is a statistical method that simplifies complex datasets by identifying the most important patterns within them. Using this approach, I discovered that the spectrum can be represented in a much lower-dimensional space than expected, which is meaningful because it suggests that we can efficiently capture the essential dynamics of gravity waves and use this reduced representation to improve parameterizations in climate models.”

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

“Via email: mahmoudi[at]iau.uni-frankfurt.de”