Unsupervised clustering of Southern Ocean Argo float temperature profiles

• Main Authors: Jones, Dan, Holt, Harry, Meijers, Andrew, Shuckburgh, Emily
• Format: Report
• Language: unknown
• Published: EarthArXiv 2018
• Subjects: Physical Sciences and Mathematics; Oceanography and Atmospheric Sciences and Meteorology; Oceanography; FOS Earth and related environmental sciences; Antarctic; Southern Ocean; The Antarctic; Antarc*
• Online Access: https://dx.doi.org/10.17605/osf.io/m64t7; https://eartharxiv.org/m64t7/

The Southern Ocean features a complex density structure characterised by sharp fronts, deep seasonal mixed layers, water mass formation regions, and salt-stratified waters, carrying the imprints of the Antarctic Circumpolar Current, gyre circulation, overturning, and mixing. Methods of characterising Southern Ocean density structures traditionally rely on somewhat ad-hoc definitions based on physical, chemical, and dynamic properties. As an alternative approach, here we apply an unsupervised classification technique (i.e. Gaussian mixture modelling) to Southern Ocean Argo float profiles. Gaussian mixture modelling, without using any latitude or longitude information, automatically identifies several circumpolar classes, as well as classes that bear the imprint of mode/intermediate water formation and export, large-scale gyre circulation, and the Agulhas Current.