Search

In recent years, the growing number of available climate models and future scenarios has led to emergent constraints becoming a popular tool to constrain uncertain future projections. However, when emergent constraints are applied over large areas, it is unclear (i) if the well-performing models simulate the correct dynamics within the considered area, (ii) which key dynamical features the emerging constraint is stemming from, and (iii) if the observational uncertainty is low enough to allow for a considerable reduction in the projection uncertainties. We therefore propose to regionally optimize emergent relationships with the twofold goal to (a) identify key model dynamics associated with the emergent constraint and model inconsistencies around them and (b) provide key areas where a narrow observational uncertainty is crucial for constraining future projections. Here, we consider two previously established emergent constraints of the future carbon uptake in the North Atlantic (Goris et al., 2018). For the regional optimization, we use a genetic algorithm and pre-define a suite of shapes and size ranges for the desired regions. Independent of pre-defined shape and size range, the genetic algorithm persistently identifies the Gulf Stream region centred around 30∘ N as optimal as well as the region associated with broad interior southward volume transport centred around 26∘ N. Close to and within our optimal regions, observational data of volume transport are available from the RAPID array with relative low observational uncertainty. Yet, our regionally optimized emergent constraints show that additional measures of specific biogeochemical variables along the array will fundamentally improve our estimates of the future carbon uptake in the North Atlantic. Moreover, our regionally optimized emergent constraints demonstrate that models that perform well for the upper-ocean volume transport and related key biogeochemical properties do not necessarily reproduce the interior-ocean volume transport well, leading to inconsistent gradients of key biogeochemical properties. This hampers the applicability of emergent constraints over large areas and highlights the need to additionally evaluate spatial model features.

Riverine transport of nutrients and carbon from inland waters to the coastal and finally the open ocean alters marine primary production (PP) and carbon (C) uptake regionally and globally. So far, this process has not been fully represented and evaluated in the state-of-the-art Earth system models. Here we assess changes in marine PP and C uptake projected under the Representative Concentration Pathway 4.5 climate scenario using the Norwegian Earth system model, with four riverine transport configurations for nutrients (nitrogen, phosphorus, silicon, and iron), carbon, and total alkalinity: deactivated, fixed at a recent-past level, coupled to simulated freshwater runoff, and following four plausible future scenarios. The inclusion of riverine nutrients and carbon at the 1970 level improves the simulated contemporary spatial distribution of annual mean PP and air–sea CO2 fluxes relative to observations, especially on the continental margins (5.4 % reduction in root mean square error (RMSE) for PP) and in the North Atlantic region (7.4 % reduction in RMSE for C uptake). While the riverine nutrients and C input is kept constant, its impact on projected PP and C uptake is expressed differently in the future period from the historical period. Riverine nutrient inputs lessen nutrient limitation under future warmer conditions as stratification increases and thus lessen the projected decline in PP by up to 0.66 ± 0.02 Pg C yr−1 (29.5 %) globally, when comparing the 1950–1999 with the 2050–2099 period. The riverine impact on projected C uptake depends on the balance between the net effect of riverine-nutrient-induced C uptake and riverine-C-induced CO2 outgassing. In the two idealized riverine configurations the riverine inputs result in a weak net C sink of 0.03–0.04 ± 0.01 Pg C yr−1, while in the more plausible riverine configurations the riverine inputs cause a net C source of 0.11 ± 0.03 Pg C yr−1. It implies that the effect of increased riverine C may be larger than the effect of nutrient inputs in the future on the projections of ocean C uptake, while in the historical period increased nutrient inputs are considered the largest driver. The results are subject to model limitations related to resolution and process representations that potentially cause underestimation of impacts. High-resolution global or regional models with an adequate representation of physical and biogeochemical shelf processes should be used to assess the impact of future riverine scenarios more accurately.

The Southern Ocean is a major region of ocean carbon uptake, but its future changes remain uncertain under climate warming. Here we show the projected shift in the Southern Ocean CO2 sink using a suite of Earth System Models, revealing changes in the mechanism, position and seasonality of the carbon uptake. Dominant CO2 uptake shifts from the Subtropical to the Antarctic region under the high-emission scenario by the end of the 21st century. The warming-driven sea-ice melt, increased ocean stratification, mixed layer shoaling, and a weaker vertical carbon gradient will together reduce the winter outgassing in the future, which will trigger the switch from mixing-driven outgassing to solubility-driven uptake in the Antarctic region during the winter season. The future Southern Ocean carbon sink will be poleward-shifted, operating in a hybrid mode between biologically-driven summertime and solubility-driven wintertime uptake with further amplification of biological uptake by the increasing Revelle Factor.

The inter-annual variability of the air-sea CO2 flux, is non-negligible, can modulate the global warming signal, yet it is poorly represented in Earth Systems Models (ESMs). ESMs are highly sophisticated and computationally demanding, which makes it challenging to perform dedicated experiments to investigate the key drivers of the CO2 flux variability across different spatial and temporal scales. Machine leaning methods can objectively and systematically explore large datasets, ensuring physically meaningful results. Here, we show that a Kernel Ridge Regression can reconstruct the present and future CO2 flux variability in five ESMs. Surface concentration of dissolved inorganic carbon (DIC) and alkalinity emerge as the critical drivers, but the former is projected to play a lesser role due to decreasing vertical gradient. Our results demonstrate a new approach to efficiently interpret the massive datasets produced by ESMs and at the same time offer guidance into future model development and monitoring strategies to constrain the CO2 flux.

The dynamic and thermohaline characteristics of the Atlantic Ocean linked to the Atlantic Meridional Overturning Circulation (AMOC) give it a specific role in the accumulation of heat and CO2, either of natural or anthropogenic origin (Cant), from the surface layer to the deep waters, significantly mitigating the impacts of anthropogenic climate change. Here, we evaluate the annual mean, long-term trends, seasonal cycle and interannual variability of net sea-air CO2 fluxes (FCO2) between 1985 and 2018 based on observation products (pCO2-products) and global ocean biogeochemical models (GOBMs) for the Atlantic from 30ºS to the Nordic Seas (~79ºN) and the Mediterranean. The mean contemporary FCO2 (sum of anthropogenic and natural components) is estimated to be 0.362 ± 0.067 and 0.47 ± 0.15 Pg C yr-1 using pCO2-products and GOBMs, respectively. The GOBMs show consistent growth trends in CO2 uptake with rates similar to the atmospheric CO2 growth, however trends obtained from CO2-products show a sharp increase from the pre-2000 period to the post-2000 period. There is overall agreement between pCO2-products and GOBMs results for mean values, seasonal cycle and interannual variability in all biomes, except for the North Atlantic subpolar biome, where pCO2-products show lower mean values, larger trends, and a different seasonal cycle than GOBMs. The GOBMs and pCO2-products show very concordant values in equatorial and subtropical regions, where CO2 variability is strongly determined by temperature. For the period 1994-2007, GOBMs show concordant values in annual Cant storage rate with carbonate marine system observations (Gruber et al., 2019) with values of 0.506 ± 0.106 Pg C yr-1 vs 0.673 ± 0.066 Pg C yr-1, respectively. The Cant storage rate agreement between GOBMs and observations are also registered in the different biomes, although in both permanently stratified subtropical in North and South Atlantic biomes, the storage rates in GOBMs show a larger spread with their mean values 30 and 40% lower than those estimated from observations. In general, the Atlantic accumulates more Cant than that inferred from the cumulative FCO2 changes, partly due to a significant lateral Cant transport from the Southern Ocean (about 30%).

The El Niño–Southern Oscillation (ENSO) widely modulates the global carbon cycle. More specifically, it alters the net uptake of carbon in the tropical ocean. Over the tropical Pacific, less carbon is released during El Niño, while the opposite is the case for La Niña. Here, the skill of Earth system models (ESMs) from the latest Coupled Model Intercomparison Project (CMIP6) to simulate the observed tropical Pacific CO2 flux variability in response to ENSO is assessed. The temporal amplitude and spatial extent of CO2 flux anomalies vary considerably among models, while the surface temperature signals of El Niño and La Niña phases are generally well represented. Under historical conditions followed by the high-warming Shared Socio-economic Pathway (SSP5-8.5) scenarios, about half the ESMs simulate a reversal in ENSO–CO2 flux relationship. This gradual shift, which occurs as early as the first half of the 21st century, is associated with a high CO2-induced increase in the Revelle factor that leads to stronger sensitivity of partial pressure of CO2 (pCO2) to changes in surface temperature between ENSO phases. At the same time, uptake of anthropogenic CO2 substantially increases upper-ocean dissolved inorganic carbon (DIC) concentrations (reducing its vertical gradient in the thermocline) and weakens the ENSO-modulated surface DIC variability. The response of the ENSO–CO2 flux relationship to future climate change is sensitive to the contemporary mean state of the carbonate ion concentration in the tropics. We present an emergent constraint between the simulated contemporary carbonate concentration with the projected cumulated CO2 fluxes. Models that simulate shifts in the ENSO–CO2 flux relationship simulate positive bias in surface carbonate concentrations.

Understanding and limiting the spread of ocean carbon sink projections are crucial to effectively guide the development of climate mitigation policies, determine accurate future carbon budget and subsequently climate change. The North Atlantic and the Southern Ocean are two of the most intense sink regions for anthropogenic CO2 emissions, while the tropical Pacific is the largest outgassing system. Nevertheless, CMIP6 models simulate growing inter-model spread in future carbon fluxes in these regions. In this study, we apply an emergent constraint approach to reduce the projections uncertainties under the high-CO2 future scenario. The efficiency of surface-to-deep transport of anthropogenic carbon is commonly identified as the key mechanisms driving the systematic inter-model spread in high latitude regions. For the North Atlantic region, we further use a genetic algorithm to optimize our identified emergent constraint relationship by isolating the region where contemporary model bias strongly correlates with the projection spread. The interior biogeochemical state is key for constraining the future tropical Pacific CO2 flux. Our study consolidates the importance of improving representations of anthropogenic carbon ventilation mechanisms in models and sustaining carbon and watermass monitoring network in these regions to improve the fidelity of future model projections.

Bromoform (CHBr3) is one of the most important precursors of atmospheric reactive bromine with an atmospheric lifetime of ~20 days. Natural production, being the main source of oceanic CHBr3, is high at the coasts and in open ocean upwelling regions due to production by macroalgae and phytoplankton. Although highly relevant for the future halogen burden and ozone layer in the stratosphere, the global bromoform production in the ocean and their emissions are still poorly constrained in observations and are mostly neglected in Earth System Model (ESM) climate projections. Here, we show first model results of fully coupled ocean-atmosphere bromoform interactions in the Norwegian ESM (NorESM) with the ocean model BLOM and the ocean biogeochemistry component iHAMOCC for the CMIP6 historical period from 1850 to 2014. Our results are validated using oceanic and atmospheric measurements listed in the HalOcAt (Halocarbons in the Ocean and Atmosphere) data base and show an overall good agreement with those observations in open ocean regions. The NorESM open ocean emissions of CHBr3 are higher than previously published emission estimates from bottom-up approaches. Moreover, the emissions are mainly positive (sea-to-air fluxes) driven by the oceanic production, sea surface temperature and wind speed, dependent on season and location. However, during low-productive winter seasons, model results also show local negative fluxes (air-to-sea fluxes) in high latitudes, suggesting some oceanic regions to be a sink of atmospheric bromoform. Driving factors will be shown for different case studies, e.g. the tropical West Pacific, which is a hot spot for oceanic bromine delivery to the stratosphere.

We present a framework that links in situ observations from the biogeochemical-Argo (BGC-Argo) array to biogeochemical models. The framework allows a minimized technical effort to construct a Lagrangian type 1D modelling experiment along BGC-Argo tracks. We utilize the Argo data in two ways; (1) drive the model physics, (2) evaluate the model biogeochemistry. BGC-Argo physics data is used to nudge the model physics closer to observations to reduce the errors in biogeochemistry stemming from physics errors. This allows us to target model biogeochemistry and by using the Argo biogeochemical dataset, we identify potential sources of model errors, introduce changes to model formulation, and validate model configurations. We present experiments for the Nordic Seas and showcase how we identify potential BGC-Argo buoys to model, prepare forcing, design experiments and approach model improvement and validation. We used ECOSMO II(CHL) model as the biogechemical component and focused on chlorophyll a. The experiments revealed that ECOSMO II(CHL) required improvements during low-light conditions, as the comparison to BGC-Argo reveals that ECOSMO II(CHL) simulates a late spring bloom and does not represent the deep chlorophyll maximum formation in summer periods. We modified the productivity and chlorophyll a relationship and statistically documented decreased bias and error in the revised model using BGC-Argo data. Our results reveal that nudging the model T and S closer to BGC-Argo data reduces errors in biogeochemistry, and we suggest a relaxation time-period of 1–10 days. The BGC-Argo data coverage is ever growing and the framework is a valuable asset for improving models in 1D-model efficiently and transfer the configurations to 3D-model with a wide range of focus from operational, regional/global and climate scale.

There is an urgent need to improve capacity to predict marine heatwaves given their substantial negative impacts on marine ecosystems. Here we present the added value of a regional climate simulation, performed with the regional Coupled-Ocean–Atmosphere-Wave-Sediment Transport model COAWST, centered over the Caribbean – one of the first of its kind on a climatological scale. We show its added value with regards to temporal distribution of marine heatwaves, compared with state-of-the-art global models. In this region, global models tend to simulate too few heatwaves that last too long compared to the observation-based dataset of CoralTemp. The regional climate model agrees more favourably with the CoralTemp dataset, particularly in winter. While examining potential mechanisms behind the differences we find that the more realistic representation of marine heatwaves in the regional model arises from the sea surface temperatures ability to increase/decrease more quickly in the regional model than in the global model. The reason for this is two fold. Firstly, the regional model has a shallower mixed layer than the global model which results in a lower heat capacity that allows its sea surface temperatures to warm and cool more quickly. The second reason is found during days when marine heatwaves are increasing in intensity. During these days, reduced wind speeds leads to less latent heat release and a faster warming surface, more so in the regional model than in the global models.

The interannual variability of the air-sea CO2 flux is non-negligible, modulates the global warming signal, and yet it is poorly represented in Earth System Models (ESMs). ESMs are highly sophisticated and computationally demanding, making it challenging to perform dedicated experiments to investigate the key drivers of the CO2 flux variability across spatial and temporal scales. Machine learning methods can objectively and systematically explore large datasets, ensuring physically meaningful results. Here, we show that a kernel ridge regression can reconstruct the present and future CO2 flux variability in five ESMs. Surface concentration of dissolved inorganic carbon (DIC) and alkalinity emerge as the critical drivers, but the former is projected to play a lesser role in the future due to decreasing vertical gradient. Our results demonstrate a new approach to efficiently interpret the massive datasets produced by ESMs, and offer guidance into future model development to better constrain the CO2 flux.

The double ITCZ (Intertropical Convergence Zone) is a ubiquitous bias in global coupled ocean-atmosphere models. It manifests as a spurious band of precipitation south of the equator, specifically in the tropical Pacific. The ITCZ position is extremely sensitive to changes in the magnitude and distribution of the Sea Surface Temperature (SST) in the tropical band, due to its strong coupling with convective precipitation. We use the Coupled-Ocean-Atmosphere-Wave-Sediment Transport (COAWST) regional modeling system to investigate the impact of different parametrizations of the oceanic vertical mixing on the water column dynamic, SST and subsequently the distribution of convective precipitation in the eastern tropical Pacific. While the role of ocean models has often been overlooked as a possible cause for this double ITCZ bias, we demonstrate that excessive stratification south of the equator can lead to a complex chain of air-sea interaction that generates the double ITCZ., The 28th IUGG General Assembly (IUGG2023) (Berlin 2023)

Downscaling physical forcing from global climate models is both time consuming and labor demanding and can delay or limit the physical forcing available for regional marine ecosystem modelers. Earlier studies have shown that downscaled physics is necessary for capturing the dynamics of primary production and lower trophic levels; however, it is not clear how higher trophic levels respond to the coarse resolution physics of global models. Here, we apply the Nordic and Barents Seas Atlantis ecosystem model (NoBa) to study the consequences of using physical forcing from global climate models versus using that from regional models. The study is therefore (i) a comparison between a regional model and its driving global model to investigate the extent to which a global climate model can be used for regional ecosystem predictions and (ii) a study of the impact of future climate change in the Nordic and Barents Seas. We found that few higher trophic level species were affected by using forcing from a global versus a regional model, and there was a general agreement in future biomass trends and distribution patterns. However, the slight difference in temperature between the models dramatically impacted Northeast Arctic cod (Gadus morhua), which highlights how species projection uncertainty could arise from poor physical representation of the physical forcing, in addition to uncertainty in the ecosystem model parameterization. publishedVersion

Ocean alkalinity is critical to the uptake of atmospheric carbon in surface waters and provides buffering capacity towards the associated acidification. However, unlike dissolved inorganic carbon (DIC), alkalinity is not directly impacted by anthropogenic carbon emissions. Within the context of projections of future ocean carbon uptake and potential ecosystem impacts, especially through Coupled Model Intercomparison Projects (CMIPs), the representation of alkalinity and the main driver of its distribution in the ocean interior, the calcium carbonate cycle, have often been overlooked. Here we track the changes from CMIP5 to CMIP6 with respect to the Earth system model (ESM) representation of alkalinity and the carbonate pump which depletes the surface ocean in alkalinity through biological production of calcium carbonate and releases it at depth through export and dissolution. We report an improvement in the representation of alkalinity in CMIP6 ESMs relative to those in CMIP5, with CMIP6 ESMs simulating lower surface alkalinity concentrations, an increased meridional surface gradient and an enhanced global vertical gradient. This improvement can be explained in part by an increase in calcium carbonate (CaCO3) production for some ESMs, which redistributes alkalinity at the surface and strengthens its vertical gradient in the water column. We were able to constrain a particulate inorganic carbon (PIC) export estimate of 44–55 Tmol yr−1 at 100 m for the ESMs to match the observed vertical gradient of alkalinity. Reviewing the representation of the CaCO3 cycle across CMIP5/6, we find a substantial range of parameterizations. While all biogeochemical models currently represent pelagic calcification, they do so implicitly, and they do not represent benthic calcification. In addition, most models simulate marine calcite but not aragonite. In CMIP6, certain model groups have increased the complexity of simulated CaCO3 production, sinking, dissolution and sedimentation. However, this is insufficient to explain the overall improvement in the alkalinity representation, which is therefore likely a result of marine biogeochemistry model tuning or ad hoc parameterizations. Although modellers aim to balance the global alkalinity budget in ESMs in order to limit drift in ocean carbon uptake under pre-industrial conditions, varying assumptions related to the closure of the budget and/or the alkalinity initialization procedure have the potential to influence projections of future carbon uptake. For instance, in many models, carbonate production, dissolution and burial are independent of the seawater saturation state, and when considered, the range of sensitivities is substantial. As such, the future impact of ocean acidification on the carbonate pump, and in turn ocean carbon uptake, is potentially underestimated in current ESMs and is insufficiently constrained.

We combine consistently dated benthic carbon isotopic records distributed over the entire Atlantic Ocean with numerical simulations performed by a glacial configuration of the Norwegian Earth System Model with active ocean biogeochemistry in order to interpret the observed Cibicides δ13C changes at the stadial–interstadial transition corresponding to the end of Heinrich Stadial 4 (HS4) in terms of ocean circulation and remineralization changes. We show that the marked increase in Cibicides δ13C observed at the end of HS4 between ∼2000 and 4200 m in the Atlantic can be explained by changes in nutrient concentrations as simulated by the model in response to the halting of freshwater input in the high-latitude glacial North Atlantic. Our model results show that this Cibicides δ13C signal is associated with changes in the ratio of southern-sourced (SSW) versus northern-sourced (NSW) water masses at the core sites, whereby SSW is replaced by NSW as a consequence of the resumption of deep-water formation in the northern North Atlantic and Nordic Seas after the freshwater input is halted. Our results further suggest that the contribution of ocean circulation changes to this signal increases from ∼40 % at 2000 m to ∼80 % at 4000 m. Below ∼4200 m, the model shows little ocean circulation change but an increase in remineralization across the transition marking the end of HS4. The simulated lower remineralization during stadials compared to during interstadials is particularly pronounced in deep subantarctic sites, in agreement with the decrease in the export production of carbon to the deep Southern Ocean during stadials found in previous studies.

Optimization problems can be found in many areas of science and technology. Often, not only the global optimum, but also a (larger) number of near-optima are of interest. This gives rise to so-called multimodal optimization problems. In most of the cases, the number and quality of the optima is unknown and assumptions on the objective functions cannot be made. In this paper, we focus on continuous, unconstrained optimization in moderately high dimensional continuous spaces (

Ocean alkalinity is critical to the uptake of atmospheric carbon in surface waters and provides buffering capacity towards associated acidification. However, unlike dissolved inorganic carbon (DIC), alkalinity is not directly impacted by anthropogenic carbon emissions. Within the context of projections of future ocean carbon uptake and potential ecosystem impacts, especially through Coupled Model Intercomparison Projects (CMIPs), the representation of alkalinity and the main driver of its distribution in the ocean interior, the calcium carbonate cycle, have often been overlooked. Here we track the changes from CMIP5 to CMIP6 with respect to the Earth system model (ESM) representation of alkalinity and the carbonate pump which depletes the surface ocean in alkalinity through biological production of calcium carbonate, and releases it at depth through export and dissolution. We report a significant improvement in the representation of alkalinity in CMIP6 ESMs relative to those in CMIP5. This improvement can be explained in part by an increase in calcium carbonate (CaCO3) production for some ESMs, which redistributes alkalinity at the surface and strengthens its vertical gradient in the water column. We were able to constrain a PIC export estimate of 51–70 Tmol yr-1 at 100 m for the ESMs to match the observed vertical gradient of alkalinity. Biases in the vertical profile of DIC have also significantly decreased, especially with the enhancement of the carbonate pump, but the representation of the saturation horizons has slightly worsened in contrast. Reviewing the representation of the CaCO3 cycle across CMIP5/6, we find a substantial range of parameterizations. While all biogeochemical models currently represent pelagic calcification, they do so implicitly, and they do not represent benthic calcification. In addition, most models simulate marine calcite but not aragonite. In CMIP6 certain model groups have increased the complexity of simulated CaCO3 production, sinking, dissolution and sedimentation. However, this is insufficient to explain the overall improvement in the alkalinity representation, which is therefore likely a result of improved marine biogeochemistry model tuning or ad hoc parameterizations. We find differences in the way ocean alkalinity is initialized that lead to offsets of up to 1 % in the global alkalinity inventory of certain models. These initialization biases should be addressed in future CMIPs by adopting accurate unit conversions. Although modelers aim to balance the global alkalinity budget in ESMs in order to limit drift in ocean carbon uptake under preindustrial conditions, varying assumptions in the closure of the budget have the potential to influence projections of future carbon uptake. For instance, in many models, carbonate production, dissolution and burial are independent of the seawater saturation state, and when considered, the range of sensitivities is substantial. As such, the future impact of ocean acidification on the carbonate pump, and in turn ocean carbon uptake, is potentially underestimated in current ESMs and insufficiently constrained.

As one of the major carbon sinks in the global ocean, the North Atlantic is a key player in mediating the ongoing global warming. However, projections of the North Atlantic carbon sink in a high-CO2 future are highly uncertain due to greatly varying model results. A previous study analysed an ensemble of 11 CMIP5-models and identified two indicators of contemporary model behavior that are highly correlated with a model´s projected future carbon-uptake in the North Atlantic: (i) the high latitude winter pCO2sea-anomaly, which is tightly linked to winter mixing and nutrient supply and (ii) the fraction of the anthropogenically altered carbon-inventory stored below 1000 m depth, indicating the efficiency of dissolved inorganic carbon transport into and within the deep ocean. Both relationships build so-called emergent constraints, where observed contemporary indicators can be used to improve future North Atlantic carbon sink estimates. In this study, we apply a genetic algorithm to optimize these emergent relationships by constraining the spatial extent of the indicators, i.e. to identify key regions that maximise the cross-correlations between the indicators and the future carbon uptake. We pre-define the shape of the desired regions as (i) rectangles and ellipses of different sizes for the first surface-2D indicator and (ii) cuboids and ellipsoids of different volumes for the second water column-3D indicator. Independent on shape and size, the genetic algorithm persistently identifies the Gulf Stream region as optimal for the first indicator as well as the pathway of the broad interior southward volume transport for the second indicator. This is further confirmed with high correlations between the North Atlantic future carbon uptake and volume transport values extracted for the central latitudes and depths of these optimal regions. Though the importance of volume transport for the carbon uptake is well known, our results go beyond traditional knowledge and identify which depth-ranges and latitudes of this volume transport are consistently of importance across the multi-model ensemble. Our study shows that regional optimisations of emergent constraint can isolate key drivers responsible for multi-model spread and furthermore provide information on where observations are most crucial to constrain future projections. Moreover, a comparison of the model performance in the identified key regions and the large-scale North Atlantic indicates that models whose mean values are in good agreement with observations within one key area do not necessarily perform well when looking at another key area. This hampers the applicability of emergent constraints and highlights the need to additionally evaluate spatial model features.

The Atlantic Niño is a major tropical interannual climate variability mode of the sea surface temperature (SST) occurring during boreal summer that shares many similarities with the tropical Pacific El Niño. Although the tropical Atlantic is an important source of CO2 to the atmosphere, the impact of Atlantic Niño on the sea-air CO2 exchange is not well understood. We reveal that the Atlantic Niño enhances (weakens) CO2 outgassing in the central (western) tropical Atlantic. In the western basin, freshwater-induced changes in surface salinity, which considerably modulate the surface ocean CO2 partial pressure (pCO2), is the primary driver for the observed CO2 flux variations. Contrastingly, pCO2 anomaly in the central basin is dominated by the SST-driven solubility change. Here, we reveal that this multi-variable mechanism for pCO2 anomaly differs remarkably from the Pacific where the response is predominantly controlled by upwelling-induced dissolved inorganic carbon anomalies. The contrasting behavior is characterized by the high CO2 buffering capacity in the Atlantic, where the subsurface water mass contains higher alkalinity than in the Pacific. The different mechanisms driving the internannual variability in sea-air CO2 exchange in the tropical Atlantic and Pacific suggests that they may respond differently to climate change.

The El Niño Southern Oscillation (ENSO) widely modulates the global carbon cycle, in particular, by altering the net uptake of carbon in the tropical ocean. Indeed, over the tropics less carbon is released by oceans during El Niño while it is the opposite for La Niña. Here, the skill of Earth System Models (ESM) from the latest Coupled Model Intercomparison Project (CMIP6) to simulate the observed tropical Pacific CO2 flux variability in response to ENSO is assessed. The temporal amplitude and spatial extent of CO2 flux anomalies vary considerably among models, while the surface temperature signals of El Niño and La Niña phases are generally well represented. Under historical conditions followed by the high warming Shared Socio-economic Pathway (SSP5-8.5) scenarios, about half the ESMs simulate a reversal in ENSO-CO2 flux relationship. This gradual shift, which occurs as early as the first half of the 21st century, is associated with a high CO2-induced increase in Revelle factor that leads to stronger sensitivity of partial pressure of CO2 (pCO2) to changes in surface temperature between ENSO phases. At the same time, uptake of anthropogenic CO2 substantially increases upper ocean dissolved inorganic carbon (DIC) concentrations, reducing its vertical gradient in the thermocline, and weakening the ENSO-modulated surface DIC variability. The response of ENSO-CO2 flux relationship to future climate change is sensitive to the contemporary mean state of the carbonate ion concentration in the tropics. Models that simulate shift in ENSO-CO2 flux relationship simulate positive bias in surface carbonate concentration.

As one of the major carbon sinks in the global ocean, the North Atlantic is a key player in mediating the ongoing global warming. However, projections of the North Atlantic carbon sink in a high-CO2 future vary greatly among models, with some showing that a slowdown in carbon uptake has already begun and others predicting that this slowdown will not occur until nearly 2100. For an ensemble of 11 CMIP5-models, we identify two indicators of contemporary model behavior that are highly correlated with a model´s projected future carbon-uptake in the North Atlantic. The first indicator is the high latitude winter pCO2sea-anomaly, which is tightly linked to winter mixing and nutrient supply, but also to deep convection. The second indicator is the fraction of the anthropogenic carbon-inventory stored below 1000-m depth, indicating the efficiency of dissolved inorganic carbon transport into the deep ocean. We further use a genetic algorithm to identify sub-regions of different shapes and sizes that optimise the correlations between our indicators and the future carbon uptake in the North Atlantic. Independent of size and shape, the genetic algorithm persistently identifies the gulf stream region as optimal for the first indicator as well as the pathway of the deep western boundary current for the second indicator. When extracting the simulated contemporary AMOC-strengths for the central latitudes and depths of these optimal regions, we also find high correlations between AMOC-values and the North Atlantic future carbon uptake. Our regional optimisation shows that modelled discrepancies in the future North Atlantic carbon uptake originate in different transport efficiencies of dissolved inorganic carbon from the surface to the deep ocean. We find a strong and highly important link between a model’s performance for gulf stream and deep western boundary currents and a model’s ability to accurately project the future carbon uptake in the North Atlantic.