Machine Ocean is a project funded by The Research Council of Norway 2020–2024.

The project will be led by physical oceanographer Dr. Cecilie Mauritzen (MET Norway). Mauritzen has extensive experience with leading complex multidisciplinary research projects, including EU, NFR and industry projects. Mauritzen was Lead Author for IPCC’s 4th and 5th Assessment Reports.

The co-lead of the project will be Dr. Jean Rabault (MET Norway). Rabault is a young scientist who has already been a pioneer in the use of Machine Learning and Artificial Neural Networks (ANN) in fluid mechanics, both for image analysis for direct measurements of velocity fields (Rabault et. al., 2017), and for modeling and investigation of the active control of the Navier–Stokes equations (Rabault et. al., 2019a,b). In addition, Rabault has the necessary background in both Fluid Mechanics and geosciences, which places him at the intersection of the domains of competencies needed for this project.

Institutions

The rest of the research team is described below, organized under their home institution.

The Norwegian Meteorological Institute (MET Norway)

Oslo, Norway

MET Norway has the national responsibility for operational ocean and weather forecasting. The institute collects and processes real time earth observations on a massive scale, and is among the world-leading institutions in both remote sensing, weather and climate modeling, and management/processing of large datasets.

The Machine Ocean team includes:

  • Dr. Veronica Berglyd Olsen, who has a background in computational physics and high energy physics, with a focus on numerical modelling and high performance computing.
  • Dr. Martin Lilleeng Sætra, who has extensive experience in scientific computing on massively parallel architectures. In his PhD work he focused on efficient shallow-water simulation and visualization on Graphical Processing Units (GPUs). He is currently working on GPU-accelerated particle filters for predicting drift in the ocean.
  • Dr. Øyvind Sætra, an expert in ocean and wave modelling. He has 20 years’ experience on research and development of operational weather and ocean forecasting systems for ECMWF and MET Norway.
  • Nils Melsom Kristensen, responsible for operational storm surge modelling at MET Norway, and has extensive knowledge and experience in high resolution coastal ocean modelling.
  • Professor Kai Håkon Christensen heads the Division for Ocean and Ice at MET Norway. He specializes in upper ocean dynamics and oceanic transport.

University of Oslo

Oslo, Norway

  • Professor Morten Hjorth-Jensen has a strong track record in applied quantum mechanics and computational physics. He is presently building the Machine Learning group at the Department of Physics.

Woods Hole Oceanographic Institution (WHOI)

Massachusetts, USA

  • Dr. James Edson has been the driving force behind the development of the COARE bulk parameterization (Edson et al., 2013), the most commonly used turbulent flux parameterization in climate modelling today. His research focus includes developing instrumentation and techniques to compute and parameterize atmospheric turbulent fluxes within the marine boundary layer including momentum and energy exchange across coupled boundary layers. He has produced most of the over ocean data to be used during the learning process.
  • Dr. Carol Anne Clayson is the Director of the Center for Air-Sea Interaction and Marine Atmospheric Sciences. Her research focuses on air–sea interaction using satellite remote sensing and modeling, with the aim to make more accurate estimates of surface turbulent fluxes (heat, water vapor, momentum) at high resolution. She has used machine learning to develop satellite based air–sea flux products (see seaflux.org).

Uppsala University

Uppsala, Sweden

  • Professor Anna Rutgersson is focusing on air–sea interaction and the impact of surface gravity waves on the atmosphere. Her group has maintained the micrometeorological site Östergarnsholm for direct measurements of vertical turbulent energy transfer over the ocean since 1995, and will provide critical data for the learning tasks.
  • Dr. Erik Nilsson has long experience on running LES with a moving lower boundary (representing a surface gravity wave) and also in transition zones between different surface roughnesses and temperatures.