About Us

International team

Our project comprises ocean, atmosphere, sea-ice scientists, numerical model developers, and machine learning experts. Our collaboration spans 5 jurisdictions and 11 academic institutions.

Innovation

Major innovations of this project include:

1. Bridging the gap between two classes of state-of-the art models: (i) Assimilation systems that are optimized to describe the long-term ocean mean state. (ii) High resolution forward ocean models that simulate changes on fine space and time scales but, for computational reasons, cannot be optimized to represent larger-scale phenomena accurately. We plan to bridge these models via a novel ML+DA framework: Multi-model Machine Learning ocean-atmosphere Emulation-Assimilation Framework (M3Leaf). This framework is trained to emulate the variability of high-resolution forward models and assimilate large-scale mean state and oceanic and atmospheric observations, thus yielding a multi-scale product that is compatible with observations at all scales.
2. Leveraging the coupling between fluxes of O₂, carbon, and heat associated with air-sea exchange, ocean transport, and biological processes. These have been included previously in forward models but not assimilation systems, due to lack of a suitable framework.
3. Representing atmospheric variability within the framework, to take full advantage of the constraints on air-sea fluxes of O₂, CO₂ and heat provided by the measurements of atmospheric O₂ and CO₂ that now span many decades.
4. Filling a major gap in the atmospheric O₂ measurement network by supporting the installation of atmospheric O₂ analyzer systems in the Southern Ocean coordinated out of South Africa, thereby also diversifying international participation.
5. Leveraging recent advances in the parameterization of sub-grid processes informed by ML to improve model fidelity.