The ocean as seen from STS-52 in November 1992.

Actionable ocean insight for Earth's climate.

I am a physical oceanographer using computer science/dynamical systems tools to explore ocean dynamics on decadal to climate scales. Passionate about bringing together different branches of oceanography, my goal is to discover the underlying principles that govern ocean dynamics from small to global scales. My focus is on physics, connecting observational and model efforts, and I also work on ocean acidification and ecology in collaborative efforts. I focus on the global ocean, using scalable methods, with a special interest in the Southern Ocean and the North Atlantic.

I currently focus on inferring sub-surface 3D ocean dynamics using surface fields. This work leverages trustworthy (interpretable and explainable) machine learning, and is designed for use with CMIP6 or similar model data that we've used to investigate the Atlantic and Southern Ocean's role in climate and also the global climate mode El Nino Southern Oscillation. This work continues the development of the SAGE (Systematic AGgregated Eco-province) method, combining statistical tools, interpretable and explainable machine learning and graphs, designed to work with non-linear data ubiquitous in oceanography and beyond.

Research Areas

Determining emergent global patterns in ocean physics and ecology

The observed and simulated ocean often produces vast quantities of high dimensional data. I use oceanographic theory, methods from numerical analysis and machine learning/data science to expose the key underlying patterns that drive physical and biogeochemical variability. The methods address challenges common across data from the geosciences.

Using deep learning for prediction of physical regimes and biological provinces

Knowing what lies beneath the surface, both figuratively and physically can offer great data analysis and collection strategy. My work uses trustworthy machine learning, combining interpretability and explainability, to give in-depth ocean insight into patterns within climate models using only limited surface fields. Beyond models, I also aim to support extreme event prediction and attribution such as hurricanes and El Nino Southern Oscillation, by tying into and supporting observational frameworks.

Revealing the Impact of Global Heating on North Atlantic Circulation Using Transparent Machine Learning

Cover of JAMES

Guest post: How artificial intelligence is fast becoming a key tool for climate science

CarbonBrief

Monitoring of Ocean State for Improved Climate Predictions

NOAA Artificial Intelligence Strategic Plan 2021-2025

Defining the fluid nature of ocean ecosystems

ECCO Media Gallery

How Machine Learning Redraws the Map of Ocean Ecosystems

Eos Buzz News

Selected Events

Upcoming

  • Talk: AGU2021. TBA.
  • Talk: Max Planck Institute, Hamburg. TBA
  • Talk: Kavli Institute for Theoretical Physics. TBA.
  • Talk: NOAA AI, 3rd workshop. TBA.
  • Lecture: Deep Learning in Geophysical Fluid Dynamics (AOS551; Princeton)

2021

  • Talk: International Conference on Machine Learning (ICML). "Revealing the impact of global warming on climate modes using transparent machine learning". ClimateChangeAI workshop spotlight. (view)
  • Tutorial: Society for Industrial and Applied Mathematics (SIAM). Conference on Mathematical and Computational Issues in the Geosciences in Milan, Italy. 60 participants.
  • Talk: GCD seminar (NCAR). "Revealing mechanisms of change in the Atlantic Meridional Overturning under global heating" (view)
  • Keynote: Virtual Summit: Incorporating Data Science and Open Science in Aquatic Research. "No free lunch: robustly revealing mechanisms of ocean circulation change under global heating with transparent ML".
  • Panel: Virtual Summit: Incorporating Data Science and Open Science in Aquatic Research. 624 participants.
  • Keynote: University of Chicago Department of Statistics. Verification, Validation, and Uncertainty Quantification Across Disciplines workshop. "Elucidating Ecological Complexity: Unsupervised Learning determines global marine eco-provinces".
  • Talk: University Corporation for Atmospheric Research. Climate & Global Dynamics Seminar.. "Revealing the impact of global heating on North Atlantic circulation using transparent machine learning".
  • Talk: University of California, Santa Cruz. Climate \& Global Dynamics Seminar. "Revealing the impact of global heating on North Atlantic circulation using transparent machine learning"
  • Talk: GEOMAR Helmholtz Centre for Ocean Research. Ocean Circulation and Climate Dynamics Colloquium. "Revealing the impact of global heating on North Atlantic circulation using transparent machine learning".
  • Poster: ICML 2021.
  • Poster: Knowledge Guided Machine Learning (KGML).
  • Talk: EGU 2021. "Revealing mechanisms of change in the Atlantic Meridional Overturning Circulation under global heating". Highlighted vPICO.

2020

  • Lecture: Oceanhackweek 2020. "Elucidating Ecological Complexity: Unsupervised Learning determines global marine eco-provinces". 20 participants.
  • Panel: Challenges and opportunities of applying AI, ML and DL to problems in the environmental and geosciences. 1200 participants.
  • Panel: NOAA Workshop. Second NOAA Workshop on Leveraging AI in the Environmental Sciences. 60+ participants.
  • Talk: Second NOAA Workshop on Leveraging AI in the Environmental Sciences. "Elucidating Ecological Complexity: Unsupervised Learning determines global marine eco-provinces".
  • Talk: NOAA Senior Management Meeting, Oceanic and Atmospheric Research. "Building geographies of ocean dynamical regimes".
  • Talk: Los Alamos National Laboratory. "Living on the Manifold: A geography of ocean dynamical regimes from eddy to global scale".
  • Talk: University of Washington. Department of Ocean Sciences, "Living on the Manifold: A geography of ocean dynamical regimes from eddy to global scale".
  • Talk: University of British Columbia. "Ocean exploration with machine learning: An Antidote to Chaos?".
  • Poster: AGU 2020."Revealing mechanisms of change in the Atlantic Meridional Overturning Circulation under global heating".
  • Talk: Climate Informatics 2020. "Elucidating Ecological Complexity: Unsupervised Learning determines global marine eco-provinces".
  • Poster: Climate Informatics 2020. "Understanding the ocean dynamics of climate models using deep neural networks.

2019

  • Lecture: Harvard University. Marine Denolle and Brad Lipovsky, graduate level: "Machine Learning in Geoscience". Class size 10.
  • Lecture: Harvard University Data Science Club. "The good, the bad and the ugly of applied unsupervised learning". class size 60.
  • Lecture: Princeton University & GFDL workshop for graduate students. "Machine learning and climate modeling". 3 day course, class size 20-30.
  • Talk: AGU. "The case for machine learning in geoscience".
  • Talk: University of Bergen. "Ocean exploration with machine learning: An Antidote to Chaos?".
  • Talk: Princeton Univeristy. "Ocean exploration with machine learning: An Antidote to Chaos?".
  • Talk: Norway-US bilateral AI workshop. "Elucidating Ecological Complexity".
  • Talk: Norway-US bilateral AI workshop. "Recognising ocean physical regimes".
  • Talk: Woods Hole Oceanographic Institute. "Ocean exploration with machine learning: An Antidote to Chaos?"
  • Talk: University of Tromso. "Ocean exploration with machine learning: An Antidote to Chaos?".
  • Poster: European Geosciences Union (EGU). "Unsupervised Learning Reveals Geography of Global Ocean Dynamical Regions".