Estimating Atlantic meridional heat transport through Bayesian modelling of altimetry, Argo and GRACE data

The Atlantic Meridional Overturning Circulation (AMOC) plays a pivotal role in the meridional transport of heat, freshwater and major dissolved gases such as carbon, and oxygen, making it a crucial component of earth’s climate system and the biosphere. On millennial timescales, the AMOC is believed to act as a conveyor belt of ocean currents wherein the flow varies coherently across latitudes. Past studies have drawn on this conveyor-belt idea to establish links between the AMOC and the Earth's climate tipping points. However, recent research and observations suggest that, on shorter timescales (days to decades), the AMOC may not operate as coherent flow of water. Understanding AMOC variability and coherence on such timescales and how these might respond to anthropogenic influences is crucial to predicting the climate of the next decades. This is, however, challenging due to the sparseness of the observational data in both time and space. Here, we present a Bayesian Hierarchical modelling framework that combines observations from altimetry, gravimetry, and Argo floats to estimate meridional heat transport across the Atlantic. Our approach considers error structures jointly and accounts for spatiotemporal dependencies between processes (thermosteric, halosteric and ocean mass), providing a coherent way to propagate uncertainty and overcoming the limitations of hydrography-only based analyses. Our estimate of heat transport is in very good agreement (correlation of ~0.8 for 3-month means) with that from RAPID observations at 26°N. A meticulous comparison of mean and variance further underscores the precision of our estimates compared to those derived from heat budgets. Our method can be extended to gain further insights into the dynamics and meridional coherence of AMOC at shorter timescales.