Pan-Atlantic decadal climate oscillation linked to ocean circulation

The study investigates whether Atlantic ocean circulation actively drives a pan-Atlantic decadal oscillation (ADO) with a 10–15 year period, beyond the prevailing view that such variability is mainly due to thermodynamic air–sea interactions. Motivated by prominent climate anomalies during 2009/10 (e.g., SST tripole, extreme NAO phases, European and Russian temperature anomalies, active Atlantic hurricane season), and contemporaneous changes in the AMOC and North Atlantic heat transport, the authors hypothesize that decadal-scale ocean circulation variability links extratropical and tropical Atlantic patterns and influences atmospheric teleconnections such as the NAO.

Prior work identified regional decadal SST patterns: the North Atlantic tripole associated with the NAO; tropical Atlantic cross-equatorial gradients tied to trade-wind changes; and a South Atlantic dipole. These have been proposed as facets of a pan-Atlantic decadal oscillation primarily driven by thermodynamic air–sea interactions. Observations and models document AMOC variability and its links to sea level along the U.S. East Coast, multidecadal sea level variability, and decadal shifts in North Atlantic circulation. Studies also highlight roles of high-latitude forcing, thermohaline circulation, and Arctic sea-ice outflow, as well as tropical–extratropical atmospheric connections and NAO variability on decadal timescales. However, the extent to which ocean circulation, versus surface heat fluxes alone, drives the ADO remained uncertain.

- Constructed a sea level-based index of ocean circulation by averaging monthly tide-gauge records from 24 sites along the U.S. northeast coast north of Cape Hatteras (Portsmouth to Boston), deseasonalized, linearly detrended, and sign-inverted so positive values correspond to stronger overturning transport; normalized by its standard deviation. Time span: 1900–2009 (extended to 2019 for AMOC comparison). - Compared the sea-level index with RAPID-MOCHA AMOC observations at 26.5°N (2005–2019); computed seasonal and monthly Pearson correlations, accounting for autocorrelation by adjusted degrees of freedom; assessed uncertainty using multiple overlapping time slices. - Performed spectral analysis (maximum entropy autoregressive method with Daniell smoothing; red-noise AR(1) confidence bounds) on monthly and December–February indices to identify periodicities; focused on the 10–15 year band (dominant peak around 147 months ≈ 12.3 years). - Conducted EOF analysis of Atlantic monthly SST anomalies (1900–2009) to isolate the ADO as EOF2; analyzed the principal component (PC2) spectrum; avoided band-pass/low-pass filtering to prevent spurious low-frequency signals. - Regressed SST, wind stress, sea surface height (from twentieth-century ocean reanalysis), and sea level pressure (from twentieth-century atmospheric reanalysis) onto SST PC2 to characterize spatial patterns and their evolution; assessed statistical significance via two-tailed t-tests with autocorrelation-adjusted dof. - Built an ITCZ index from the latitude of zero meridional wind stress averaged 30°W–40°W (5°S–20°N) and analyzed its coherence with SST PC2 and the sea-level index. - Quantified equatorial Atlantic cold tongue dynamics using the thermocline slope index ΔH, defined as the east–west gradient of the 20°C isotherm depth H averaged over 3°N–3°S: ΔH = H(20–40°W) − H(0–20°W). Assessed correlations with SST PC2 and coherence with the Atlantic Niño index (SST anomalies over 3°N–3°S, 0–20°W). - Analyzed reconstructed turbulent surface heat fluxes over 40–50°N, 20–60°W for coherence with the sea-level index, especially in winter. - Used cross-spectral analysis (maximum entropy, Daniell smoothing) to compute coherence and phase lags between the sea-level index and SST PC2/ITCZ, with 95% confidence levels for coherence.

- The sea level-derived ocean circulation index exhibits a dominant 12.3-year period (spectral peak at 147 months) with variance concentrated in the 10–15 year band; similar periodicity is present in winter (DJF) data. - The index correlates with RAPID-MOCHA AMOC (2005–2019): r = 0.50 (monthly, p = 1.48 × 10^-9); strongest in boreal winter with r = 0.72, explaining about 53–66% variance across time slices (r = 0.77 ± 0.04; p = 0.0023 ± 0.0021), capturing the 2009/10 slowdown event. - EOF2 of Atlantic SST (≈11% variance) defines the ADO with meridional bands from the South Atlantic to Greenland; SST PC2 shows a strong unfiltered 10–15 year peak. Maximum local explained variance occurs in the tropical Atlantic. - Sea surface height anomalies regressed on SST PC2 mirror SST patterns, consistent with upper-ocean heat content variability; tide-gauge sites lie within regions of significant variability. - Cross-spectral analysis shows high coherence (95% confidence) between the sea-level index and both SST PC2 and the ITCZ index in the 10–15 year band, with phase lag ≈ π/3 radians (~2 years), indicating the sea-level (ocean circulation) index leads surface climate variability. - Reconstructed turbulent surface fluxes over the North Atlantic are coherent with the sea-level index in winter and tend to lead the 10–15 year peak, suggesting a role in phase transitions. - Dynamical pathway: High-latitude forcing excites AMOC and gyre changes; coastal Kelvin waves propagate signals equatorward, are transmitted across the equator, and propagate poleward along eastern boundaries, linking North Atlantic to tropics; interior advection delays contribute to the ~2-year lead of the sea-level index over SST PC2. - In the equatorial Atlantic cold tongue, ΔH exhibits interannual to decadal variability with a 10–15-year spectral peak; ΔH correlates with SST PC2 (monthly r = 0.38, p = 2.95 × 10^-14; annual r = 0.45, p = 7.79 × 10^-7) and is significantly coherent with the Atlantic Niño index at 10–15 years with ~π rad phase (larger ΔH associated with colder eastern equatorial SST). - Mature ADO phase features strengthened southeasterly trades, weakened northeasterly trades, a northward-shifted ITCZ, enhanced upwelling and thermocline slope in the eastern equatorial Atlantic, and a negative NAO-like pattern in boreal winter.

The results demonstrate that decadal-scale ocean circulation variability actively drives the pan-Atlantic decadal oscillation, rather than acting passively in response to thermodynamic air–sea interactions. A coherent 10–15 year mode arises from interactions between AMOC, gyre circulations, and boundary-trapped Kelvin waves that transmit high-latitude anomalies to the tropics. The tropical Atlantic, particularly the cold tongue region, provides strong thermocline feedback that amplifies SST anomalies and anchors maximum variance, which in turn forces interhemispheric atmospheric teleconnections and a negative NAO phase over the North Atlantic in winter. The leading role of ocean circulation explains limited variance attribution to surface heat fluxes in studies of individual events (e.g., 2009/10). The identified lead of ocean circulation by about two years relative to surface climate suggests potential predictability for decadal Atlantic climate variability and associated impacts (e.g., hurricanes, European climate, Sahel rainfall).

Using century-long observations and reanalyses, the study identifies a pan-Atlantic 10–15 year oscillation in which ocean circulation variability, captured by a sea level-derived index, leads surface climate variability by about two years. The mechanism links extratropical North Atlantic changes to the tropical Atlantic via AMOC–gyre dynamics and Kelvin wave propagation, with strong thermocline feedback in the cold tongue and atmospheric teleconnections culminating in a negative NAO. These findings resolve a key uncertainty by highlighting the active role of ocean circulation in the ADO and imply enhanced decadal predictability for the Atlantic sector. Future work should examine the seasonality and stationarity of the patterns, impacts of known decadal shifts (e.g., around 1987), refine separation of gyre versus overturning contributions in sea level proxies, and test mechanisms across models with varying resolution and in prediction frameworks.

- The sea level-derived index integrates influences from both gyre and overturning circulations and potentially atmospheric effects; attribution between these components is imperfect. - Strongest AMOC–sea level correlations occur in boreal winter; correlations are weak in other seasons and for annual means. - The analysis relies on proxies and reanalyses (e.g., sea level along the U.S. coast, reanalyzed sea surface height), which carry uncertainties and potential inhomogeneities. - Focus is on the 10–15 year band; interannual and multidecadal variability are present but not the emphasis here. - Questions about the seasonality, stationarity, and impacts of decadal regime shifts (e.g., circa 1987) on the ADO patterns remain for future investigation. - Amplitude differences exist between the sea-level index and AMOC observations (e.g., 2009/10 event).