Meeting Documents

Examining the Temporal Sensitivity of Mixed Layer Salinity Budget Around the Global Ocean With Unsupervised Machine Learning

Liu, C. and Liang, X. (2024)
Presented at: Ocean Sciences Meeting 2024

Abstract

The mixed layer salinity (MLS) budget and its variability is essential for understanding air-sea freshwater exchange and its implication for the global hydrological cycle. While extensive research on the seasonal and regional patterns of MLS exists, less attention has been given to the temporal sensitivity of the MLS budget on a global scale. In this study, we apply an unsupervised machine learning technique, K-means clustering, to the Estimating the Circulation and Climate of the Ocean (ECCO) estimates, to objectively categorize the MLS budget equation. We identify several distinct dynamical regimes within the MLS budget. Over 70% of the global ocean is significantly influenced by air-sea freshwater flux, which balances oceanic processes like advection and diffusion. In the Southern Ocean, MLS is primarily determined by the interplay of negative contributions from surface forcing and positive contributions from diffusion. Furthermore, we investigate the evolving spatial distribution of these clusters across different timescales and create a map illustrating the local temporal sensitivity of the MLS budget balance. This approach provides a more comprehensive understanding on how to improve the retrieval of freshwater exchange from salinity observations on multiple timescales.
View Document (AGU) »