Yihan Wang 王艺晗

I am a Postdoctoral Researcher at Lawrence Berkeley National Laboratory and an affiliate of the International Computer Science Institute (ICSI), where I work with Dr. Benjamin Erichson in the Deep Learning Group.

My research lies at the intersection of hydrology and scientific machine learning, with an emphasis on developing reliable, physically consistent, and interpretable deep learning methods for hydrologic and Earth system modeling.

My current research focuses on two complementary directions:

• Modern deep learning for hydrology: Exploring advanced architectures, including state-space models and diffusion-based generative models, to improve streamflow simulation and probabilistic forecasting.
• Physics-aware neural networks: Incorporating physical principles, particularly mass conservation, into neural networks to investigate how physical constraints influence predictive accuracy, generalization, and learned representations across diverse hydroclimatic conditions.

Through this work, I aim to bridge advances in artificial intelligence with domain knowledge in hydrology. My broader goal is to develop accurate, robust, and physically meaningful modeling tools that advance our understanding of water systems, improve hydrologic predictions, and support hazard preparedness and informed water-resources management.

You can find my publications on Google Scholar. I’m always happy to connect with others working in hydrology, machine learning, and Earth system science. Feel free to reach out at yhwang@icsi.berkeley.edu.

Education

• Ph.D., University of Oklahoma, 2023–2026
• M.S., University of California, Los Angeles, 2019–2020
• B.E., Hohai University, 2015–2019 (final undergraduate year at UCLA, 2018–2019)