Unsupervised clustering identifies thermohaline staircases in the Canada Basin of the Arctic Ocean

• DOI: 10.1017/eds.2024.13
• Publication Date: 2024-05-06T07:08:48Z
• Authors: Mikhail G. Schee; Erica Rosenblum; Jonathan M. Lilly; Nicolas Grisouard
• Keywords: Temperature salinity diagrams; Stratification (seeds)

Abstract Thermohaline staircases are a widespread stratification feature that impacts the vertical transport of heat and nutrients consistently observed throughout Canada Basin Arctic Ocean. Observations from same time period geographic region form clusters in temperature-salinity ( T – S ) space. Here, for first time, we use an automated clustering algorithm called hierarchical density-based spatial applications with noise to detect connect individual well-mixed staircase layers across profiles ice-tethered profilers. Our application only requires estimate typical layer thickness expected salinity range staircases. We compare this method two previous studies used different approaches reproduce several results, including mean lateral density ratio $ {R}_L difference between neighboring is magnitude larger than variance within layer. find can accurately automatically track coherent space profiles. In evaluating algorithm’s performance, evidence physical features, namely splitting or merging remnant intrusions. Further, dependence on pressure, whereas have reported constant . results demonstrate algorithms effective parsimonious identifying ocean profile data.