Semester of Graduation

Summer 2026

Degree

Master of Oceanography and Coastal Sciences (SOCS)

Department

Oceanography and Coastal Sciences

Document Type

Thesis

Abstract

Monitoring oxygen variability in the deep ocean is critical for understanding ocean circulation, biogeochemical cycling, and long-term deoxygenation, but observations below 2000-m are sparse. In this study, observing system design using the high-resolution CM2.6 ocean model was performed to assess the capability of a possible Deep Argo network to monitor global dissolved oxygen variability in the deep ocean. Over a two-year simulation period, virtual Deep Argo floats were uniformly deployed along Lagrangian trajectories according to CM2.6’s velocity field to sample oxygen concentrations from the surface to 5000-m. The output of these simulations was evaluated by comparing interpolated oxygen fields to the CM2.6 reference state using root-mean-square deviation (RMSD), performing leave-one-out (LOO) cross-validation, and calculating physical oxygen gradient diagnostics.

The results show that increasing the number of virtual floats generally increases oxygen reconstruction skill, but the advantages become progressively lower as deployment densities increase, showing diminishing returns beyond 800-1000 floats. Reconstructed oxygen fields successfully recover the dominant basin-scale structure of deep-ocean oxygen variability across both simulation years, demonstrating the ability of the Deep Argo network to capture large-scale oxygen distributions despite evolving float trajectories and changing sampling geometry.

LOO calculations show that the observational influence of individual floats varies greatly between ocean regions and years. Regions of elevated LOO sensitivity are associated with oxygen gradients depending on location and time. In year 1, when the floats begin in a uniformly spaced distribution, larger observational sensitivity correlates with stronger horizontal oxygen gradients in mid- and high-latitude regions, whereas tropical and subtropical regions have a closer relationship between LOO sensitivity and elevated vertical oxygen gradients. In subsequent years, LOO sensitivity can be associated with bathymetry and other large-scale patterns because of float advection into slower moving, less uniformly distributed configurations. These findings suggest that optimal placement of floats for the purpose of sampling oxygen has a lifetime of at most one year, but the locations where the floats should be seeded in the first year could be predictable.

Overall, this study reveals that effective deep-ocean oxygen monitoring does not necessarily require uniform placement of oxygen sensors across the Deep Argo array. Optimal placement of each Deep Argo float with an oxygen sensor is a process that evolves over time. Strategically optimized sensor placement based on oxygen gradients, variability, and changing float distributions could provide a more efficient and scientifically sound framework for future global deep-ocean monitoring systems.

Date

7-13-2026

Committee Chair

Dr. Kanchan Maiti

LSU Acknowledgement

1

LSU Accessibility Acknowledgment

1

Download

Available for download on Thursday, July 12, 2029

Share

COinS