Patrick Duke

Topic

Investigating the Northeast Pacific Ocean Carbon Sink using a Machine Learning Approach

School of Earth and Ocean Sciences

Date & location

• Friday, May 10, 2024
• 9:00 A.M.
• Clearihue Building, Room B007

Examining Committee

Supervisory Committee

• Dr. Roberta Hamme, School of Earth and Ocean Sciences, University of Victoria (Co-Supervisor)
• Dr. Debby Ianson, School of Earth and Ocean Sciences, UVic (Co-Supervisor)
• Dr. Neil Swart, School of Earth and Ocean Sciences, UVic (Member)
• Dr. Peter Landschützer, Research Director, Flanders Marine Institute (Outside Member)

External Examiner

• Dr. Burke Hales, College of Earth, Ocean, and Atmospheric Sciences, Oregon State University

Chair of Oral Examination

• Dr. Margaret-Anne Storey, Department of Computer Science, UVic

Abstract

Improving our understanding of how the ocean absorbs carbon dioxide (CO₂) is critical to climate change mitigation efforts. The global ocean takes up nearly a quarter of anthropogenic CO₂ emissions annually, but the variability of this uptake at regional scales remains poorly understood. In this dissertation I compiled an extensive collection of reported surface ocean air-sea CO₂ exchange values within each of Canada’s three adjacent ocean basins. I go on to summarize current research and identify steps forward to improve our understanding of the marine carbon sink in Canadian national and offshore waters. I then developed advanced techniques for quantifying air-sea CO₂ fluxes in the Northeast Pacific Ocean to improve our understanding of processes driving seasonal, interannual, decadal, and long-term variability, aiding in monitoring, reporting, and verification of future marine carbon dioxide removal, and helping inform carbon and climate policies. Utilizing a neural network approach to interpolate sparse observations, I created monthly gridded seawater partial pressure of CO₂ (pCO₂) data products from January 1998 to December 2019, at 1/12°x1/12° spatial resolution, in the Northeast Pacific Ocean. The two data products, encompassing the open ocean and the coastal ocean, were created using non-linear relationships between pCO₂ observations and a range of predictor variables representing processes affecting pCO₂, at a spatial resolution four times greater than leading global products. Using an ensemble approach, I was able to produce robust pCO₂ estimates, evaluated against independently withheld data, that represent regional variability with better overall performance compared to global products. I conducted a novel sensitivity analysis which identified that the parameters responsible for the neural network’s ability to capture regional pCO₂ variability agrees with mechanistic processes. The regional open ocean and coastal products also reproduced pCO₂ estimates well within the overlapping domain, with differences influenced by scarcity of observations. Using wind speed and atmospheric CO₂, I calculated air-sea CO₂ fluxes. In the open ocean, on sub-decadal to decadal timescales, I found that the upwelling strength of the subpolar Alaskan Gyre, driven by large-scale atmospheric forcing, acts as the primary control on air-sea CO₂ flux variability. In the coastal ocean, I report an anticorrelation between annual air-sea CO₂ flux and its seasonal amplitude with the relationship driven by regional processes. I estimate long-term surface ocean pCO₂ increase at a rate below the atmospheric trend. The slowest rate of increase occurs where there is strong interaction with subsurface waters in the Alaskan Gyre and the West Coast upwelling zone. Basin-wide, my results suggest that the region is a net sink for atmospheric CO₂ with trends indicating increasing oceanic uptake.