March 3, 2025
Pedram Hassanzadeh is an associate professor in the Department of the Geophysical Sciences and Committee on Computational and Applied Mathematics. He is also the Faculty Director of the AI for Climate (AICE) Initiative and Codirector of the Human-centered Weather Forecasting Initiative. He works on improving the fundamental understanding of the multiscale, nonlinear physics of extreme weather events and developing new tools for predicting the variability of these events across time scales, from days to decades. His work combines theory, observational data, physics-based computer models, and, increasingly, data-driven artificial intelligence (AI) models. He worked with NVIDIA to develop the first successful AI weather forecast model, FourCastNet, in 2022.
What overarching question are you trying to answer with your research?
My work is focused on two interconnected overarching questions. The first question is how to predict extreme weather events in a changing climate. Whether we aim to forecast specific events (e.g., 1 or 2 weeks ahead) or quantify changes in their likelihood decades away, we need to predict complex multiscale nonlinear interactions among various physical processes. While our traditional approaches have been slow and even stalling, AI has started to show promise in providing better predictive tools. However, we currently do not understand how/why these AI models work so well. Answering this question is essential to developing more trustworthy and better AI weather and climate models and, maybe even more ambitiously, learning something new about climate physics.
Can you share an example of how interdisciplinary collaboration has enhanced your research and led to unexpected or exciting findings?
I love the interdisciplinary nature of both AI and climate science. Working at their intersection is even more exciting and impactful. I will give you two examples. I worked with an amazing group of AI experts from NVIDIA and Caltech to develop FourCastNet, the first AI weather model that showed comparable accuracy to that of the best operational physics-based models. This work, which was only possible through such an interdisciplinary collaboration, started a second revolution in weather forecasting. Now, we have AI weather models that outperform the best physics-based models at 100,000 times lower computational cost.
Then, right after moving to UChicago, I met Michael Kremer (Economics) and Amir Jina (Harris), who had been doing groundbreaking work on providing weather forecasting to farmers in low- and middle-income countries. However, traditional weather models are computationally expensive and not customized for low- and middle-income countries. Together, we realized that there is so much untapped potential in this AI-driven revolution to democratize weather forecasting worldwide. We now have an interdisciplinary Human-centered Weather Forecasting Initiative that aims to directly translate AI and weather/climate prediction innovations to major socioeconomic benefits. This kind of work is only possible through close interdisciplinary collaborations. I am very excited about how some of the work in my group, which was considered too theoretical and too ambitious until 2–3 years ago, might now be used by millions of people around the world.
What did you want to be when you grew up, or what do you want to do when you retire?
A professor! I really have never considered any other career path. I have not thought of retirement, but I am really enjoying what I am currently doing.
How do you spend your time outside of work?
I am new to Chicago and have been enjoying exploring the city. I love the walkability of Chicago (well, except for a few winter months!). I love the different neighborhoods, museums, restaurants, and things to do around the lake and river. I also enjoy reading biographies, especially those of accomplished scientists.