Fraser King

Fraser King is an incoming Assistant Professor in the Department of Atmospheric and Oceanic Sciences (AOS) at the University of Wisconsin–Madison, starting in Winter 2026. He holds a PhD in Machine Learning and Remote Sensing of Precipitation from the University of Waterloo (2022) and is currently a postdoctoral research associate at NASA Goddard Space Flight Center. His research integrates machine learning with atmospheric physics to advance precipitation and snowfall retrieval, cloud microphysics, and climate modeling. He has held research positions at the University of Michigan and NASA Jet Propulsion Laboratory.

His research interests include:

• Climate and Climate Change
• Radiation and Remote Sensing
• Synoptic Meteorology
• Atmospheric and Cloud Physics
• Large Scale Dynamics
• Machine Learning and Model Interpretability
• Arctic Snowfall Prediction

His recent publications reflect a strong trend in applying deep learning (e.g., U-Net, CNNs) and unsupervised methods (PCA, t-SNE, UMAP) to radar and satellite data for precipitation and snow microphysics. Key themes include radar gap inpainting, melting layer detection, and dimensionality reduction for physical interpretation. His work bridges geoscience and AI, aiming for interpretable models that enhance physical understanding.

Scientific awards and professional service include:

• Finalist for the 2023 Governor General's Gold Medal, University of Waterloo
• Associate Editor, Journal of Atmospheric and Oceanic Technology (AMS)
• Member, AMS Committee on Artificial Intelligence Applications to Environmental Science
• Executive Council Member, AGU Precipitation Technical Committee
• Executive Member, Eastern Snow Conference Research Board

Fraser King has mentored students through research projects and led educational initiatives such as a 12-week course on machine learning for land cover classification. He has secured research experience through internships at Aquanty Inc. and multiple NASA-affiliated institutions. He founded MapsByFraser, a company combining cartography and satellite data, and has collaborated with Google's Quantum AI team. His technical skills span Python, deep learning frameworks, and high-performance computing platforms.

He leads several major research projects:

• Towards Interpretable Physical Models: Using sparse autoencoders and nonlinear dimensionality reduction to interpret geoscience models.
• Microphysical Dimensionality Reduction: Applying PCA, t-SNE, and UMAP to identify physical modes in precipitation data.
• BlindPaint: A U-Net for radar gap inpainting in spaceborne systems.
• DeepPrecip: A deep learning model for surface precipitation retrieval.
• iPhone LiDAR: Using consumer smartphones for snow depth measurement via drones.
• NRCan Machine Learning Land Cover Classifier: Training ML models on Sentinel-2 data.
• Climate Model Calibration: Using ML to correct biases in snow-related climate variables.
• CloudSat Snowfall Validation: Validating high-latitude snowfall estimates.
• Snow Modelling: A Rust-based physical/temperature-index snow model.