Disentangled Fingerprints suggest no historical weakening of Atlantic Overturning and Subpolar Gyre

The 30-second take

• What: The authors show common AMOC fingerprints mix AMOC and Subpolar Gyre signals, then train statistically optimal SST- and salinity-based fingerprints on CMIP6 piControl that reconstruct the two systems with less overlap and, on observations, do not confirm the large recent AMOC weakening implied by older indices.
• Abundance angle: today, trustworthy early-warning skill for Atlantic circulation is scarce. Cleaner, separated fingerprints are a step toward more default climate intelligence — knowing what you are actually watching (mid-horizon: measurement and interpretation both matter).
• Who should care: Physical oceanographers, climate-risk teams watching AMOC tipping headlines, and CMIP6 detection-and-attribution groups.

What the paper actually did

The AMOC and the Subpolar Gyre (SPG) are major Atlantic circulation systems that might destabilize under human forcing. Long direct AMOC records are short, so people use observation-based “fingerprints,” mostly from sea-surface patterns. The authors say existing fingerprints do not cleanly split SPG from AMOC; they mix both.

They test widely used AMOC fingerprints across CMIP6 and compare them with statistically optimal fingerprints they derive from sea-surface temperature and salinity, trained on pre-industrial control simulations. Traditional fingerprints correlate only weakly with both AMOC and SPG strength. The optimized ones beat those approaches on held-out piControl segments and on historical experiments, reconstructing AMOC and SPG while minimizing overlap.

They call this the first tailored observational pair that robustly distinguishes the two, useful for detection, attribution, and possible tipping precursors. Applied to SST observations, the optimized fingerprints do not confirm the substantial recent-decade AMOC weakening suggested by traditional fingerprints, but also do not rule out a loss of stability.

What makes this disruptive

The scarce capability is knowing whether the Atlantic overturning is actually slowing — a high-stakes, headline-prone claim. If classic fingerprints are weak and entangled with the gyre, a generation of “AMOC is weakening” stories may have been partly gyre weather.

Statistically optimal, disentangled indices that win in CMIP6 and then soften the observational weakening narrative are disruptive to that literature. The authors are careful: no confirmation of a large recent drop, no all-clear on stability.

This is an attribution-methods paper, not a new observing array.

Why it matters (outside the lab)

Abundance lens: cheap, trustworthy climate intelligence — is a major heat-and-carbon conveyor changing? — is still scarce. Better fingerprints move that intelligence toward a default public layer instead of dueling indices.

Near-term, treat traditional AMOC SST fingerprints as mixed. Medium-term, salinity data, deeper observations, and independent arrays still decide. No invented year of collapse or safety.

Policy-relevant caution: “not confirmed” ≠ “stable.”

Limitations & open questions

Optimal fingerprints are trained on models (piControl) and then applied to observations; model bias can leak. “Do not confirm substantial weakening” is not a proof of a steady AMOC. Stability loss remains explicitly open.

CMIP6 skill on historical experiments is necessary but not sufficient for the real ocean. Abstract does not list the exact SST/SSS regions or coefficients. Preprint ≠ closed case. Abundance is not automatic: a better index does not by itself prevent or prove tipping.