Barbara Zemskova, University of Toronto
Title
Bridging the gaps in ocean modeling
Abstract
There is a large separation of scales for oceanic processes: global ocean circulation occurs over time scales of years to decades and length scales of thousands of kilometres, whereas diffusive processes occur within seconds to hours on the order of millimetres. Although small-scale processes are important for distributing momentum, heat, and dissolved materials in the ocean, they are difficult to measure through observations and have to be approximated in large-scale modeling efforts, such as climate models. In this talk, I will explore two of the approaches I have used to bridge the gaps between ocean observations, large-scale models, and small-scale processes. First, as an example of a process-driven approach, I will present results from high-resolution simulations of oceanic flow over small, rough topographic obstacles, which are too small to be captured in climate models. These results show how the generation of waves above these features, their interaction with each other, and their breaking are dependent on the height of the topography, and highlight the non-linear effects that are not captured in current large-scale model approximations. Second, as an example of a data-driven approach, I will present a deep-learning model developed to estimate concentrations of dissolved inorganic carbon in the ocean. This model provides an efficient framework to fill in the observational gaps in ocean carbon measurements, as we used it to analyze trends over the last three decades, and also to continue monitoring future changes in ocean carbon concentration.