My Research

My research is broadly focused on AI for climate and Earth science. I’ve developed deep learning methods that drastically accelerate climate model predictions while quantifying critical uncertainties—bridging the gap between complex Earth systems and actionable insights. While some of my work involves emulation of ice sheet dynamics, the underlying goals are much broader: creating scalable, interpretable, and uncertainty-aware tools for climate modeling and environmental applications.

I’ve published first-author work in journals like JAMES, GMD, and TMLR, and have coauthored interdisciplinary papers including a high-impact Nature review on AI-driven scientific discovery. I maintain open-source packages like ise, which provide reproducible pipelines for building ML models in geosciences. My work is supported by the NSF GRFP and has been presented at AGU, the International Liège Colloquium, and other major venues.

Publications

First-Author

• Van Katwyk & Bergen (2025). HybridFlow: Quantification of Aleatoric and Epistemic Uncertainty with a Single Hybrid Model. TMLR. (to be submitted)

• Van Katwyk et al. (2025). ISEFlow v1.0: A Flow-Based Neural Network Emulator for Improved Sea Level Projections and Uncertainty Quantification. Geoscientific Model Development. https://doi.org/10.5194/egusphere-2025-870

• Van Katwyk et al. (2023). A Variational LSTM Emulator of Sea Level Contribution from the Antarctic Ice Sheet. JAMES. https://doi.org/10.1029/2023MS003899

Coauthored

• Wang et al. (2023). Scientific discovery in the age of artificial intelligence. Nature. https://doi.org/10.1038/s41586-023-06221-2

• Seroussi et al. (2023). Insights into the vulnerability of Antarctic glaciers from the ISMIP6 ice sheet model ensemble and associated uncertainty. The Cryosphere. https://doi.org/10.5194/tc-17-5197-2023

• Barton et al. (2023). The development of laterite weathering profiles as a function of rainfall and time: A geophysical approach. Earth Surface Processes and Landforms. https://doi.org/10.1002/esp.5688

Talks (Invited + Oral Presentations)

• Emulation of sea level rise from the Antarctic and Greenland Ice Sheets using ISEFlow – World Climate Research Programme Emulator Task Team, Feb 2025 (Invited)

• Machine Learning Emulators of Sea Level Contribution from the Antarctic and Greenland Ice Sheets – Machine Learning for Physical Oceanography Seminar, Jan 2025 (Invited)

• AI-ATC: Reinforcement Learning for Optimal ARTCC Sector Traffic Management – NASA Internal Conference, 2024

• Improving Sea Level Projections with AI – Brown University, Research Matters, 2024

• Emulation of ISMIP6 Antarctic Sea Level Contribution and Ensemble Distributions Using Time Series Neural Networks – International Liège Colloquium, May 2023

Posters

• ISEFlow: Emulating Sea Level Rise using a Hybrid Flow-Based Neural Network Architecture – AGU 2024

• Variational LSTM Emulators of Sea Level Contribution from the Antarctic and Greenland Ice Sheets – AGU 2023

• Normalizing Flows: Uncertainty Quantification in Earth Machine Learning – SAGE GAGE, 2021

• Redox and Detrital Geochemical Variation as an Indicator of Organofacies Variability – Heterogeneity, Provenance, and Process in the Mowry Shale – AAPG ACE, 2020

• Gradational Weathering of Molokai, Hawaii: Geophysical Study of Hawaiian Lateritic Weathering Profiles – GSA, 2020