Thermohaline Research Articles

Opportunities for Earth Observation to Inform Risk Management for Ocean Tipping Points

AbstractAs climate change continues, the likelihood of passing critical thresholds or tipping points increases. Hence, there is a need to advance the science for detecting such thresholds. In this paper, we assess the needs and opportunities for Earth Observation (EO, here understood to refer to satellite observations) to inform society in responding to the risks associated with ten potential large-scale ocean tipping elements: Atlantic Meridional Overturning Circulation; Atlantic Subpolar Gyre; Beaufort Gyre; Arctic halocline; Kuroshio Large Meander; deoxygenation; phytoplankton; zooplankton; higher level ecosystems (including fisheries); and marine biodiversity. We review current scientific understanding and identify specific EO and related modelling needs for each of these tipping elements. We draw out some generic points that apply across several of the elements. These common points include the importance of maintaining long-term, consistent time series; the need to combine EO data consistently with in situ data types (including subsurface), for example through data assimilation; and the need to reduce or work with current mismatches in resolution (in both directions) between climate models and EO datasets. Our analysis shows that developing EO, modelling and prediction systems together, with understanding of the strengths and limitations of each, provides many promising paths towards monitoring and early warning systems for tipping, and towards the development of the next generation of climate models.

Quantifying risk of a noise-induced AMOC collapse from northern and tropical Atlantic Ocean variability

Abstract The Atlantic Meridional Overturning Circulation (AMOC) exerts a major influence on global climate. There is much debate about whether the current strong AMOC may collapse as a result of anthropogenic forcing and/or internal variability. Increasing the noise in simple salt-advection models can change the apparent AMOC tipping threshold. However, it's not clear if `present-day' variability is strong enough to induce a collapse.
Here, we investigate how internal variability affects the likelihood of AMOC collapse. We examine internal variability of basin-scale salinities and temperatures in four CMIP6 pre-industrial simulations. We fit this to an empirical, process-based AMOC box model, and
find that noise-induced AMOC collapse (defined as a decade in which the mean AMOC strength falls below $5 \ \rm{Sv}$) is unlikely, however, if the AMOC is pushed closer to a bifurcation point due to external climate forcing, noise-induced tipping becomes more likely.
Surprisingly, we find a case where forcing temporarily overshoots a stability threshold but noise decreases the probability of collapse. Accurately modelling internal decadal variability is essential for understanding the increased uncertainty in AMOC projections.

Subpolar North Atlantic Mean State Affects the Response of the Atlantic Meridional Overturning Circulation to the North Atlantic Oscillation in CMIP6 Models

Abstract The Atlantic meridional overturning circulation (AMOC) plays an important role in climate, transporting heat and salt to the subpolar North Atlantic. The AMOC’s variability is sensitive to atmospheric forcing, especially the North Atlantic Oscillation (NAO). Because AMOC observations are short, climate models are a valuable tool to study the AMOC’s variability. Yet, there are known issues with climate models, like uncertainties and systematic biases. To investigate this, preindustrial control experiments from models participating in the phase 6 of Coupled Model Intercomparison Project (CMIP6) are evaluated. There is a large, but correlated, spread in the models’ subpolar gyre mean surface temperature and salinity. By splitting models into groups of either a warm–salty or cold–fresh subpolar gyre, it is shown that warm–salty models have a lower sea ice cover in the Labrador Sea and, hence, enable a larger heat loss during a positive NAO. Stratification in the Labrador Sea is also weaker in warm–salty models, such that the larger NAO-related heat loss can also affect greater depths. As a result, subsurface density anomalies are much stronger in the warm–salty models than in those that tend to be cold and fresh. As these anomalies propagate southward along the western boundary, they establish a zonal density gradient anomaly that promotes a stronger delayed AMOC response to the NAO in the warm–salty models. These findings demonstrate how model mean state errors are linked across variables and affect variability, emphasizing the need for improvement of the subpolar North Atlantic mean states in models.

Worldwide consequences of a mid-Holocene cold event in the Nordic Seas

The present interglacial is a relatively warm and stable period, especially compared to the preceding glacial time. However, the Holocene has seen the emergence of several remarkable cold events, some with worldwide consequences. Leveraging marine records from the Nordic Seas, we provide the first detailed account of a cold event centered around 6.8 ka BP. Utilizing paleoceanographic proxies and advanced modeling, we unveil a distinct subsurface water cooling, associated with a stepwise increase in sea-ice cover in the eastern Fram Strait. Our findings emphasize the role of Greenland Sea deep convection onset and the subsequent westward shift in Atlantic Water flow, enabling sea-ice advection from the Barents Sea. The heightened sea-ice cover weakened Atlantic Water advection, perturbing thermohaline circulation in the eastern Nordic Seas. These perturbations propagated worldwide, affecting North Atlantic deep-water circulation, inducing widespread hemispheric cooling, shifting the Intertropical Convergence Zone southward, and weakening the East Asian monsoon. Incorporating results from the Transient simulations of Climate Evolution of the last 21,000 years (TraCE-21ka) supports and augments proxy-based paleoreconstructions, underscoring sea-ice dynamics and ocean circulation's critical influence. This study highlights the potential for localized cold events within ostensibly warm climatic intervals. It underscores the need to comprehend their mechanisms for precise climate predictions and informed policymaking toward a sustainable future.

Open ocean convection drives enhanced eastern pathway of the Glacial Atlantic Meridional Overturning Circulation

Abundant proxy records suggest a profound reorganization of the Atlantic Meridional Overturning Circulation (AMOC) during the Last Glacial Maximum (LGM, ~21,000 y ago), with the North Atlantic Deep Water (NADW) shoaling significantly relative to the present-day (PD) and forming Glacial North Atlantic Intermediate Water (GNAIW). However, almost all previous observational and modeling studies have focused on the zonal mean two-dimensional AMOC feature, while recent progress in the understanding of modern AMOC reveals a more complicated three-dimensional structure, with NADW penetrating from the subpolar North Atlantic to lower latitude through different pathways. Here, combining 231Pa/230Th reconstructions and model simulations, we uncover a significant change in the three-dimensional structure of the glacial AMOC. Specifically, the mid-latitude eastern pathway (EP), located east of the Mid-Atlantic Ridge and transporting about half of the PD NADW from the subpolar gyre to the subtropical gyre, experienced substantial intensification during the LGM. A greater portion of the GNAIW was transported in the eastern basin during the LGM compared to NADW at the PD, resulting in opposite 231Pa/230Th changes between eastern and western basins during the LGM. Furthermore, in contrast to the wind-steering mechanism of EP at PD, the intensified LGM EP was caused primarily by the rim current forced by the basin-scale open-ocean convection over the subpolar North Atlantic. Our results underscore the importance of accounting for three-dimensional oceanographic changes to achieve more accurate reconstructions of past AMOC.

Centennial‐Scale Variability of the Atlantic Meridional Overturning Circulation in CMIP6 Models Shaped by Arctic–North Atlantic Interactions and Sea Ice Biases

AbstractClimate variability on centennial timescales has often been linked to internal variability of the Atlantic Meridional Overturning Circulation (AMOC). However, due to the scarceness of suitable paleoclimate proxies and long climate model simulations, large uncertainties remain on the magnitude and physical mechanisms driving centennial‐scale AMOC variability. For these reasons, we perform a systematic multi‐model comparison of centennial‐scale AMOC variability in pre‐industrial control simulations of state‐of‐the‐art global climate models. Six out of nine models in this study exhibit a statistically significant mode of centennial‐scale AMOC variability. Our results show that freshwater exchanges between the Arctic Ocean and the North Atlantic provide a plausible driving mechanism in a subset of models, and that AMOC variability can be amplified by ocean–sea ice feedbacks in the Labrador Sea. The amplifying mechanism is linked to sea ice cover biases, which could provide an observational constraint for centennial‐scale AMOC variability.

Arctic sea‐ice export as a mechanism for cold climate events during the last deglaciation

ABSTRACTRecently it was proposed that a sudden drainage of accumulated Arctic sea ice could have caused cold climate events during the last deglaciation. To explore this mechanism, we performed numerical experiments with an atmosphere–ocean–sea ice model. In these experiments, the impact of a large flush of Arctic sea ice was compared to a reference glacial state of the climate. In our results, the sea ice flush produces a major surface freshening of the North Atlantic Ocean and a 17% weakening of the strength of the Atlantic Meridional Overturning Circulation. Together with an increase in surface albedo, this weakening in ocean circulation leads to a cooling over the North Atlantic, extending to the downwind continents. Compared to our reference glacial state, the cooling reaches 5°C and lasts about 80 years. This climate anomaly is similar in magnitude and duration to relatively short cooling events during the last deglaciation, such as the Older Dryas, the Inter‐Allerød cold period and the Preboreal Oscillation. We thus conclude that the sea‐ice flush mechanism is consistent with the occurrence of such cooling events in the North Atlantic region. However, longer cooling events such as the Younger Dryas would require additional mechanisms.

A solution for constraining past marine Polar Amplification

Most climate proxies of sea surface temperatures suffer from severe limitations when applied to cold temperatures that characterize Arctic environments. These limitations prevent us from constraining uncertainties for some of the most sensitive climate tipping points that can trigger rapid and dramatic global climate change such as Arctic/Polar Amplification, the disruption of the Atlantic Meridional Overturning Circulation, sea ice loss, and permafrost melting. Here, we present an approach to reconstructing sea surface temperatures globally using paired Mg/Ca - δ18Oc recorded in tests of the polar to subpolar planktonic foraminifera Neogloboquadrina pachyderma. We show that the fidelity of Mg/Ca-based paleoclimate reconstructions is compromised by variations in seawater carbonate chemistry which can be successfully quantified and isolated from paleotemperature reconstructions using a multiproxy approach. By applying the calibration to the last glacial maximum, we show that marine polar amplification has been underestimated by up to 3.0 ± 1.0 °C in model-based estimates.

Tracking orbital and suborbital climate variability in the westernmost Mediterranean over the past 13,000 years: New insights from paleoperspectives on marine productivity responses

This study presents a comprehensive analysis of a sediment record from the Western Alboran Basin (core GP04PC), utilizing palynological and geochemical tools to investigate marine productivity responses to orbital and suborbital climate variability over the past 13,000 years. High productivity during the Younger Dryas humid phase (∼12.4–11.7 ka) and the Holocene humidity optimum (∼10.5–8.5 ka) was driven by increased local river discharges resulting from rapid mountain glaciers melting and enhanced regional precipitation. During the late Holocene, frequent flood events linked to negative North Atlantic Oscillation (NAO) incursions potentially led to multicentennial-scale productivity increases. The findings indicate that periods characterized by wet regional conditions and increased river run-off, influenced by orbital (e.g., insolation cycles) and suborbital factors (e.g., NAO and Atlantic Meridional Overturning Circulation changes), consistently enhanced marine productivity in the Western Alboran Basin. The study also reveals that the current high productivity and carbon export in the Western Alboran Basin are maintained by active upwelling and downwelling systems driven by a persistent positive NAO phase following the southward migration of the Intertropical Convergence Zone (ITCZ) that occurred around 6.5 ka. Furthermore, geochemical proxies support a strong detrital influence on trace metal concentrations, including barium (Ba), in deep Western Alboran sediments during the Holocene. This limits the use of Ba/Al ratios for accurately reconstructing productivity changes and highlights the importance of dinocyst analysis as a complementary tool for robust marine productivity reconstructions in this region. These observations provide valuable paleoperspectives on marine ecosystem responses to climate variability, contributing to the development of robust long-term productivity models essential for adapting to ongoing environmental changes in the region, and demonstrating the strong influence of North Atlantic climate and ocean dynamics on centennial-scale productivity oscillations in this region.

Late Pleistocene to early Holocene vegetation and environmental changes in the tropical Leizhou Peninsula, South China: New evidence from the n-alkane record

Understanding past long-term vegetation responses to regional or even global climate change and forcing mechanisms is essential to future climate change projections. However, due to the lack of long-term terrestrial sedimentary records, there are few studies focusing on vegetation changes in tropical southern China since the last glacial period, especially from the perspective of peat n-alkane records. Here, we have presented a peat core record from the Xialu peatland in the northern Leizhou Peninsula, and n-alkanes were investigated in conjunction with multiple proxy indicators. Our results showed that the organic matter sources were mainly a mixture of aquatic and terrestrial vegetation, with terrestrial vegetation accounting for most of the bulk organic matter composition. From ∼44.1 to 29 cal kyr BP, the organic matter source was mainly dominated by terrestrial vegetation, which corresponds to warm and humid conditions. From 29 to 14 cal kyr BP, the input of the terrestrial vegetation was reduced, the aquatic vegetation input increased, implying cool and dry conditions. From 14 to 9.3 cal kyr BP, the climate gradually became warmer and wetter, and terrestrial vegetation dominated in this area. Overall, the climatic conditions from the Xialu peatland were generally consistent with other records from adjacent areas. Our results suggest that, during the Last Glacial Maximum (LGM), a substantial drop in regional and global sea levels may have been the main cause of drought in tropical southern China on orbital timescales. Meanwhile, several climatic fluctuations on millennial timescales could have been influenced by the variability of the North Atlantic Meridional Overturning Circulation (AMOC) and migration of the Intertropical Convergence Zone (ITCZ).

A regional NEMO 4.0 configuration of the subpolar North Atlantic

The dynamics of the North Atlantic subpolar gyre (SPG) largely control the Atlantic Meridional Overturning Circulation (AMOC), thus affecting the long-term climate variability of the global ocean. Many key processes in the North Atlantic SPG, such as Gulf Stream detachment, eddy activity, deep convection and overflows, are frequently incorrectly presented or misrepresented in modern ocean general circulation models. Here, we present a new regional eddy-resolving model of the subpolar North Atlantic based on the NEMO4 configuration and analyze its ability to adequately represent the dynamics of the subpolar North Atlantic. The model performance is assessed using in situ data (cross-section hydrography, Argo and sea gliders), remote sensing data (satellite sea surface temperature and salinity) and the ANDRO and ARMOR3D products based on optimal interpolation products. In the next step, we analyze the effects of high-resolution forcing and several recent developments in ocean modeling to improve the model solution. The implementation of the self-diffusive momentum advection scheme with additional viscosity allows for the simulation of more realistic Deep Western Boundary Current and Irminger Rings lifecycle characteristics. The surface current feedback parameterization implemented in the forcing function and the modification of the kinetic energy budget in the upper ocean, especially in intensively eddying areas, simulate the behavior predicted by coupled models. While the model is sensitive to the implementation of cool skin–warm layer parameterization, this parameterization does not significantly improve the solution compared to the observational data. Regional implementation of the local-sigma vertical coordinate system in cascading areas allows overflow waters to spread over the ocean bottom more realistically. The remote effect of the implementation of the local-sigma coordinate is traced in the vertical thermohaline structure at the western slopes of the Irminger Sea and Iceland Basin as well as in the Deep Western Boundary Current in the Labrador Sea. Compared to z-partial step setting, local-sigma allows for correct representation of continuous densification of the water in the overturning part of the AMOC and thus North Atlantic Deep-Water formation. The potential of the new model configuration for analyzing the subpolar North Atlantic, including convective activity, is discussed.

The Bering Strait Throughflow Component of the Global Mass, Heat and Freshwater Transport

AbstractAs the only oceanic connection between the Pacific and Arctic‐Atlantic Oceans, Bering Strait throughflow carries a climatological northward transport of about 1 Sv, contributing to the Atlantic Meridional Overturning Circulation (AMOC). Here, Lagrangian analysis quantifies the global distributions of volume transport, transit‐times, thermohaline properties, diapycnal transformation, heat and freshwater transports associated with Bering Strait throughflow. Virtual Lagrangian parcels, released at Bering Strait, are advected by the velocity of Estimating the Circulation and Climate of the Ocean, backward and forward in time. Backward trajectories reveal that Bering Strait throughflow enters the Pacific basin on the southeast side, as part of fresh Antarctic Intermediate Water, then follows the wind‐driven circulation to Bering Strait. Median transit time from S in Indo‐Pacific to Bering Strait is 175 years. Sixty‐four percent of Bering Strait throughflow enters the North Atlantic through the Labrador Sea. The remaining 36% flows through the Greenland Sea, warmed and salinified by the northward flowing Atlantic waters. Deep water formation of water flowing through Bering Strait occurs predominantly in the Labrador Sea. Subsequently, this water joins the lower branch of AMOC, flowing southward in the deep western boundary current as North Atlantic Deep Water. Median transit time from Bering Strait to S in South Atlantic is 160 years. The net heat transport of Bering Strait throughflow is northward everywhere, and net freshwater transport by Bering Strait throughflow is mostly northward. The freshwater transport is largest in the subpolar region of basin sectors: northward in the Pacific and Arctic and southward in the Atlantic.

A Review of the Oceanography and Antarctic Bottom Water Formation Offshore Cape Darnley, East Antarctica

AbstractAntarctic Bottom Water (AABW) is the densest water mass in the world and drives the lower limb of the global thermohaline circulation. AABW is formed in only four regions around Antarctica and Cape Darnley, East Antarctica, is the most recently discovered formation region. Here, we compile 40 years of oceanographic data for this region to provide the climatological oceanographic conditions, and review the water mass properties and their role in AABW formation. We split the region into three sectors (East, Central and West) and identify the main water masses, current regimes and their influence on the formation of Cape Darnley Bottom Water (CDBW). In the eastern sector, Prydz Bay, the formation of Ice Shelf Water preconditions the water (cold and fresh) that flows into the central sector to E, enhancing sea ice production in Cape Darnley Polynya. This produces a high salinity variant of Dense Shelf Water (DSW) (up to 35.15 g/kg) that we coin Burton Basin DSW. In contrast, the western sector of the Cape Darnley Polynya produces a low salinity variant (up to 34.85 g/kg) we coin Nielsen Basin DSW. The resultant combined CDBW is the warmest (upper temperature bound of 0.05C) AABW formed around Antarctica with an upper bound salinity of 34.845 g/kg. Our findings will contribute to planning future observing systems at Cape Darnley, determining the role that CDBW plays in our global oceanic and climate systems, and modeling past and future climate scenarios.

Kilometer‐Scale Assessment of the Adriatic Dense Water Multi‐Decadal Dynamics

AbstractThe North Adriatic Dense Water (NAddW)—the densest Mediterranean water generated by extreme cooling during wintertime hurricane‐strength winds—drives the thermohaline circulation, ventilates the deep layers, and changes the biogeochemical properties of the Adriatic Sea. However, modeling the dynamical properties of such dense water at the climate scale has been a challenge for decades due to the complex coastal geomorphology of the Adriatic basin not properly reproduced by existing climate models. To overcome these deficiencies, a 31‐year‐long simulation (1987–2017) of the Adriatic Sea and Coast (AdriSC) kilometer‐scale atmosphere‐ocean model is used to analyze the main NAddW dynamical phases (i.e., generation, spreading and accumulation). The study highlights four key results. First, during winter, the NAddW densities are higher in the shallow northern Adriatic shelf than in the deeper Kvarner Bay—where 25%–35% of the overall NAddW are found to be generated—due to a median bottom temperature difference of 2°C between the two generation sites. Second, the NAddW mass transported across most of the Adriatic peaks between February and May, except along the western side of the Otranto Strait. Third, for the accumulation sites, the bottom layer of the Kvarner Bay is found to be renewed annually while the renewal occurs every 1–3 years in the Jabuka Pit and every 5–10 years in the deep Southern Adriatic Pit. Fourth, the NAddW cascading and accumulation is more pronounced during basin‐wide high‐salinity conditions driven by circulation changes in the northern Ionian Sea.

Deep Kinetic Energy Response to the Variability of the Kuroshio Extension

AbstractThe transfer of energy from the upper to deep oceans is a well‐known and challenging subject in physical oceanography. This research investigates the intricate relationship between surface and deep ocean currents. Utilizing nearly 2 years of observations from the Kuroshio Extension System Study (KESS), our study unveils the deep‐ocean kinetic energy response to the Kuroshio Extension (KE) variability. We introduce the Kuroshio Extension Jet Path Index (KEJPI), which identifies two distinct modes of the KE jet on an intra‐seasonal timescale. Our findings reveal a strong correlation (0.73) between KEJPI and the mean kinetic energy of deep‐ocean geostrophic circulation, suggesting that the KE jet's large amplitude meanders have a significant impact on deep‐ocean kinetic energy. Singular Value Decomposition (SVD) analysis further unveils a co‐evolving spatial pattern between upper and lower ocean kinetic energy. We investigate the dynamic vertical coupling (DVC) mechanism by examining the coherent variation among the sea surface height (SSH), 15°C isotherm (Z15), deep pressure anomaly, and abyssal flow. The KE jet migration can induce net divergence and convergence within the water column, which in turn generates deep‐ocean quasi‐geostrophic currents. These currents show a marked increase in kinetic energy, reaching levels three times higher than the background. This DVC‐driven kinetic energy can further cascade into near‐inertial and high‐frequency internal waves, contributing to abyssal mixing. Our study underscores the role of large current system instabilities in transferring energy to the deep ocean and facilitating deep mixing processes.

Long-Term Changes in Deep Ocean Currents between Ulleung Island and Dok Island in the East Sea from 1996 to 2023

This study presents a comprehensive analysis of long-term deep ocean currents between Ulleung Island and Dok Island in the East Sea, based on data observed from the East Sea Current Measurement 1 (EC1) mooring station spanning from 1996 to 2023. The deep current at the Ulleung Interplain Gap, a trough situated between Ulleung Island and Dok Island, originates from the Japan Basin, a region where deep water formation occurs within the East Sea. The southward flow and velocity of this deep water are reflective of both regional and global climate dynamics. Understanding these dynamics is crucial for assessing environmental impact, both locally in Korea and globally. Significant trends and anomalies were identified in the zonal and meridional velocities within the 1,800 and 2,500-meter depth range. Data from 2012 to 2014 required corrections due to unit inconsistencies, which, after adjustment, revealed a deceleration of the southward bottom current, with speeds decreasing from 1.49 cm s<sup>-1</sup> in November 1996 to 1.14 cm s<sup>-1</sup> in April 2023. These changes may be linked to global climate variations and bottom water formation in the Japan Basin, influencing regional circulation patterns. The findings highlight the importance of sustained ocean monitoring and enhanced data management to ensure the accuracy and utility of long-term oceanographic datasets.

Nonstationarity of the Atlantic Meridional Overturning Circulation's Fingerprint on Sea Surface Temperature

AbstractSea surface temperature (SST) has been increasing since industrialization with rising greenhouse gases. However, a warming hole exists in the North Atlantic where SST has cooled by 0.4 K/century during 1900–2017. It has been argued that this cooling is due to a slowdown of the Atlantic Meridional Overturning Circulation (AMOC), and subpolar North Atlantic SST has thus been utilized to estimate AMOC variability. We assess the robustness of subpolar North Atlantic SST as a proxy for AMOC strength under historical forcing, abrupt quadrupling of CO2, and a medium future emissions pathway, finding that AMOC's fingerprint on SST depends upon forcing scenarios. AMOC is important in warming hole development during significant warming periods, although SST may introduce uncertainties for AMOC reconstruction in stabilized regimes due to diverse forcing mechanisms and decadal variability. Our results caution against using SST alone as a proxy for AMOC variability—both on paleoclimatic and contemporary time scales.

Precession‐Driven Variations in the Indonesian Throughflow Thermocline and Its Implications on the Agulhas Leakage

AbstractThe Indonesian Throughflow (ITF) thermocline, as exclusive water source for the “warm water route” of the Atlantic Meridional Overturning Circulation (AMOC), provides waters that exit the Indian Ocean and join the AMOC upper limb via the Agulhas Leakage (AL). Hence, investigating long‐term variations in the ITF thermocline and its implications on the AL is important for understanding dynamics of the AMOC. Here, the thermohaline history of the ITF thermocline was reconstructed for the last ∼410 kyr based on δ18O and Mg/Ca of planktonic foraminifera from International Ocean Discovery Program Site U1483. By integrating the new and published records, we found that precession drives variation of the ITF thermocline through modulating the intensity of the Australian‐Indonesian winter monsoon, El Niño‐Southern Oscillation‐like states and formation of the North Pacific Tropical Water, in turn exerting a transoceanic influence on the amount of the AL and seawater temperature and salinity of the South Atlantic thermocline.

Atlantyk Północny a klimat Europy. Mechanizmy wpływu. Część 1

The work discusses the mechanisms operating within the climate system leading to variability and changes in the atmospheric climate over Europe. The first part discusses the operation of the main factor regulating the variability of atmospheric circulation, which is the North Atlantic thermohaline circulation (NA THC). The variability of NA THC, through its influence on the shaping of the mid-tropospheric circulation and the distribution of geopotential heights, determines the variability of the lower circulation, and thus the variability of climatic elements over Poland and Europe. The analysis showed that the variability of NA THC largely explains the variability of sunshine duration, air temperature and relative humidity, but does not satisfactorily explain the variability of other climatic elements.

Estimation of AMOC transition probabilities using a machine learning based rare-event algorithm

Abstract The Atlantic Meridional Overturning Circulation (AMOC) is an important component of the global climate, known to be a tipping element, as it could collapse under global warming. The main objective of this study is to compute the probability that the AMOC collapses within a specified time window, using a rare-event algorithm called Trajectory-Adaptive Multilevel Splitting (TAMS). However, the efficiency and accuracy of TAMS depend on the choice of the score function. Although the definition of the optimal score function, called “committor function” is known, it is impossible in general to compute it a priori. Here, we combine TAMS with a Next-Generation Reservoir Computing technique that estimates the committor function from the data generated by the rare-event algorithm. We test this technique in a stochastic box model of the AMOC for which two types of transition exist, the so-called F(ast)-transitions and S(low)-transitions. Results for the F-transtions compare favorably with those in the literature where a physically-informed score function was used. We show that coupling a rare-event algorithm with machine learning allows for a correct estimation of transition probabilities, transition times, and even transition paths for a wide range of model parameters. We then extend these results to the more difficult problem of S-transitions in the same model. In both cases of F-transitions and S-transitions, we also show how the Next-Generation Reservoir Computing technique can be interpreted to retrieve an analytical estimate of the committor function.