Abstract
In this dissertation, a novel optimized mapping technique was developed to estimate the Atlantic Meridional Overturning Circulation (AMOC) time series from Argo and XBT datasets, named AXMOC. This technique enabled the observation of large-scale upper ocean patterns in the South Atlantic basin using sustained in-situ temperature (T) and salinity (S) observations. This led to the first long-term in-situ based AMOC and meridional heat transport (MHT) time series at 22.5°S. The AMOC and MHT time series highlight the significance of eastern and western boundary dynamics in AMOC variability, sea level, and water mass characteristics. The second study applies the AXMOC methodology to 34.5°S, enhancing data coverage to 19 years (2005 - 2023) and validating the method against a machine learning approach. Both methods show robust representations of AMOC, MHT, and freshwater transport (FWT) variabilities. A significant AMOC weakening between 2008 and 2011, driven by geostrophic flow anomalies, was identified, partially aligning with an AMOC observational array (SAMBA). AXMOC data, demonstrating improved variability, could complement SAMBA array data to minimize observational gaps and extend AMOC coverage back to 2005, ensuring consistency despite latitudinal variability in the XBT AX18 transect due to changes in ship routes. Finally, the third study analyzes large-scale water mass patterns in the South Atlantic, comparing upper ocean T and S anomalies across different products. South Atlantic upper ocean S anomalies correlated well with anomalies in a southeastern portion, suggesting the influence of Agulhas leakage and rings on the South Atlantic basin. Based on these findings, a mechanism linking the northwestward propagation of Agulhas leakage to a poleward signal along the upper western boundary system is proposed.