Indications for a North Atlantic ocean circulation regime shift at the onset of the Little Ice Age

Abstract

A prominent characteristic of the reconstructed Northern Hemisphere temperature signal over the last millennium is the transition from the Medieval Climate Anomaly to the Little Ice Age (LIA). Here we report indications for a non-linear regime shift in the North Atlantic ocean circulation at the onset of the LIA. Specifically, we apply a novel statistical test based on horizontal visibility graphs to two ocean sediment August sea-surface temperature records from the Norwegian Sea and the central subpolar basin and find robust indications of time-irreversibility in both records during the LIA onset. Despite a basin-wide cooling trend, we report an anomalous warming in the central subpolar basin during the LIA that is reproduced in ensemble simulations with the climate model of intermediate complexity CLIMBER-3$\alpha$ as a result of a non-linear regime shift in the subpolar North Atlantic ocean circulation. The identified volcanically triggered non-linear transition in the model simulations provides a plausible explanation for the signatures of time-irreversibility found in the ocean sediment records. Our findings indicate a potential multi-stability of the North Atlantic ocean circulation and its importance for regional climate change on centennial time scales.

Appendices

Appendix 1: Ensemble simulations of the last millennium with CLIMBER-3$\alpha$

In this appendix, we present further details on the ensemble simulations of the last millennium with CLIMBER-3$\alpha$ as well as results of the time-irreversibility test applied to the model output.

1.1 Model description

CLIMBER-3$\alpha$ is a model of intermediate complexity (Montoya et al. 2005). It’s oceanic component is based on the GFDL MOM-3 code (Pacanowski and Griffies 1999), with 24 variably spaced vertical levels, a coarse horizontal resolution of 3.75$^\circ$, a background vertical diffusivity of $\kappa _h = 0.3 \times 10^{-4}$ m$^2$ s$^{-1}$ and an eddy-induced tracer advection with a thickness diffusion coefficient of $\kappa _{gm} = 250$ m$^2$ s$^{-1}$. It contains a coarse resolution statistical-dynamical atmosphere (Petoukhov et al. 2000) and a thermodynamic/dynamic sea-ice component (Fichefet and Maqueda 1997).

Although this model’s coarse resolution and the simplified atmosphere clearly limit its prognostic capabilities at regional scales, CLIMBER-3$\alpha$ has been found to reproduce large-scale characteristics of the global climate system and has been used in a variety of model intercomparison studies for the last millennium and future projections of the stability of the AMOC (Jansen et al. 2007; Eby et al. 2013; Gregory et al. 2005; Stouffer et al. 2006).

1.2 Ensemble simulations of the last millennium

In the simulations over the last millennium presented here, we applied TSI reconstructions by Steinhilber et al. (2009) and volcanic forcing by Crowley (2000) as well as anthropogenic aerosols and greenhouse gas forcing following the PMIP3 recommendations (Schmidt et al. 2011). The combined TSI is shown in Fig. 7a.

Based on a reconstruction of the North Atlantic Oscillation (NAO) by Trouet et al. (2009) as the leading mode of atmospheric variability in the North Atlantic, we stochastically generated an ensemble of 10 independent representations of wind-stress fields for the last millennium (similar to an approach by Sedláček and Mysak (2009), see Schleussner and Feulner (2013) for further details on the method). For illustration purposes, the NAO record by Trouet et al. (2009) as well as one example reconstruction are depicted in Fig. 7a.

It is important to highlight that the response of the North-Atlantic ocean in the ensemble simulations on multi-decadal to centennial time scales is dominated by the coupled sea-ice–ocean mechanism identified (Schleussner and Feulner 2013), although the NAO reconstruction by Trouet et al. (2011) indicates a shift from a persistent positive NAO to a more oscillatory regime during the MCA–LIA transition. While this persistent positive NAO phase during the MCA is not reproduced by AOGCM simulations of the last millennium (Lehner et al. 2012), a less prominent shift in the atmospheric conditions between MCA and LIA would not affect the main findings presented here.

1.3 Bistability in the subpolar gyre circulation in CLIMBER-3$\alpha$

CLIMBER-3$\alpha$ exhibits a regime shift in the subpolar gyre circulation with respect to the convection strength in its centre (Levermann and Born 2007) that can be triggered by a variety of forcings, e.g. applying a very weak freshwater offset of the order of 15 mSv over the Nordic Sea convection side. Mengel et al. (2012) found that the oceanic response to atmospheric variability in CLIMBER-3$\alpha$ performs best in reproducing observed levels close to the threshold of the circulation regime. While this multi-stability can also be a result of the coarse resolution and other shortcomings of the specific model, signatures of multi-stability have also been found in a variety of complex coupled models (Born et al. 2013; Schulz et al. 2007).

While the regime shift itself is a robust finding also without additional freshwater budget adjustment (Schleussner and Feulner 2013), the best match with reconstructed data is achieved for a constant freshwater offset of 5 mSv over the convective region in the Nordic Seas (63.75$^\circ$–78.75$^\circ$N and 11.25$^\circ$W–10$^\circ$E). This adjustment is within the range of observed natural variability since the 1950s (Curry and Mauritzen 2005). The actual timing of this transition shows a considerable ensemble spread, thus indicating the importance of atmospheric conditions, and is also very sensitive to minor changes in the freshwater budget. It is important to highlight the conceptual nature of the model results presented here, since models of intermediate complexity like CLIMBER-3$\alpha$ are not suitable to provide realistic transient dynamics on short time scales, but rather indicate possible mechanisms for transition.

Appendix 2: Time series irreversibility analysis for the CLIMBER-3$\alpha$ simulations

Simulations with CLIMBER-3$\alpha$ reveal a non-linear regime shift in the subpolar North Atlantic at the MCA–LIA transition that should be detectable using the time series irreversibility analysis technique applied here. Figure 8 summarizes the results of the time series irreversibility analysis for the annual Nordic Seas and subpolar basin area-averaged SST signals. It depicts for each time step the median (the $p$-values for 5 out of 10 ensemble members are equal or below this value at this time step) and the 30 % quantile (the $p$-values for 3 out of 10 ensemble members are equal or below this value at this time step). Due to the prescribed atmospheric forcing applied, potential signatures of time-irreversibility in the SST signal in the CLIMBER-3$\alpha$ ensemble simulations may not evolve as they would without it and we can only speculate about the actual effect of this prescription on the highly sensitive analysis methods. This represents a serious limitation and neither the results for individual ensemble members nor the quantile estimates should be directly compared to the results for the individual paleo-record time series presented in Fig. 3. We show the quantile values to give an indication, where the time series reversibility test leads to rejection of the NH in the CLIMBER-3$\alpha$ ensemble simulations bearing the limitations discussed above in mind.

Fig. 8

Time series reversibility test for the CLIMBER-3$\alpha$ ensemble simulations with an annual sampling time. Left (right) panel SST record averaged over the Nordic Seas (subpolar basin). See Fig. 1 for the region and Fig. 7 for the corresponding time series. The upper panels denote median $p$ values ($p_k$: degree-based test, $p_c$: local clustering coefficient-based test) over the model ensemble and the lower panels give the 30 % quantile