Optimal Fingerprints for Decoupling Atlantic Overturning and Subpolar Gyre

Abstract The Atlantic Meridional Overturning Circulation (AMOC) and the Subpolar Gyre (SPG) are key components of the Earth's climate system, both potentially prone to abrupt shifts under anthropogenic forcing. In the absence of long-term direct observations, several observation-based fingerprints of the AMOC have been proposed to study its response to anthropogenic forcing. However, traditional AMOC fingerprints do not separate the two different Atlantic ocean circulation systems, but rather represent an opaque mixture of both and other effects, causing concern regarding their representativeness of the underlying AMOC dynamics. Here, we assess the performance of the most widely used AMOC fingerprints across CMIP6 models and contrast them with statistically optimal fingerprints that we derive from sea surface temperature and salinity data. We show that traditional AMOC fingerprints perform poorly in reproducing interannual variability and show weak correlations with both AMOC and SPG strength. Our statistically optimal fingerprints, trained on pre-industrial control (piControl) simulations, consistently outperform AMOC fingerprints from the literature in both unseen parts of piControl and historical simulations across all models. Our fingerprints accurately reconstruct AMOC and SPG signals with minimal overlap, offering a clear separation between the two systems. Our results thus give the first reliable, tailored fingerprints of both AMOC and SPG, and highlight the potential of optimized data-driven fingerprints for analyzing these two key Atlantic ocean circulation components, including improved detection of potential AMOC or SPG shifts.