Date: 18 November 2021
Time: 3:30 pm
Location: Zoom
Presenter: Dr. Anand Gnanadesikan
Title: How can we make marine biogeochemical models more realistic? New directions using machine learning and genomics
Abstract:
Marine biogeochemical models are used for a wide range of purposes including estimating the uptake of anthropogenic carbon dioxide, projecting ocean acidification and deoxygenation and modeling coastal hypoxia. However, it is not always clear whether models get the right answers (or the wrong ones) for the right reasons, as the solutions depend on the representation of both physics and biology. In this talk, I discuss two potential approaches for improving models. The first uses machine learning techniques to isolate apparent relationships between environmental forcing parameters and phytoplankton biomass. The second uses shotgun metagenomic data collected in Chesapeake Bay to find relationships between environmental forcing parameters and gene abundance. In both cases the observed relationships show interesting deviations from modeled relationships, suggesting potential pathways for future improvements.
Bio:
Anand Gnanadesikan fell in love with the oceans as a kid going to Cape Cod, and with oceanography as an undergraduate with Jorge Sarmiento at Princeton. He stayed with Cape Cod for his Ph.D studying Langmuir circulations before, with an apparent lack of imagination, returned to Princeton for 15 years. Feeling left out when his daughter was applying to college, he and his wife decided to apply to colleges as well He ended up at Johns Hopkins, where he has been since 2011. His research centers around using numerical models and data synthesis to understand the interactions of the ocean with the biosphere, atmosphere and cryosphere.
Contact: Jacob Wenegrat
If you are not subscribed to seminar announcements and need a link to this online seminar please contact aosc-helper. You can also subscribe to weekly announcements below.
Department of Atmospheric & Oceanic Science, 16 November 2021. More information.
