The simulations used in this manuscript were run using the Flexible Ocean and Climate Infrastructure (FOCI) ESM (Matthes et al., 2020). FOCI consists of an atmosphere–ocean–sea ice general circulation model with a surface land component, coupling the Nucleus for European Modelling of the Ocean (NEMO3.6, Madec and the NEMO Team, 2015) for ocean and sea-ice, the Jena Scheme for Biosphere–Atmosphere Coupling in Hamburg (JSBACH; Brovkin et al., 2013) for the land, and ECHAM6.3-HAM2.3-MOZ1.0 (ECHAM6-HAMMOZ; Schultz et al., 2018) for the atmosphere. The ocean is divided into 46 vertical layers with a horizontal resolution of 0.5° and the atmosphere is composed of 95 vertical layers with a horizontal resolution of ∼ 1.8°.

In addition to simulating the land carbon cycle, FOCI incorporates the Model of Oceanic Pelagic Stoichiometry (MOPS) (Kriest and Oschlies, 2015) to enable the representation of marine biogeochemical processes and the ocean carbon cycle. MOPS resolves biological dynamics such as phytoplankton carbon uptake and remineralisation, calcite formation and dissolution (Chien et al., 2022). Alkalinity is a prognostic tracer simulated as a combination of biological sources and sinks, including: nitrate and phosphate supply as sinks, calcium carbonate production and dissolution as sink and source, respectively, and organic matter production and remineralisation as sink and source, respectively (Chien et al., 2022). In FOCI-MOPS, the ocean biogeochemical state, including DIC and alkalinity, is calculated on the NEMO-MOPS ocean grid, while the coupled air–sea CO₂ flux information is transferred through OASIS3-MCT from the ORCA05 ocean grid to the coarser atmospheric T63 grid. The CO₂ flux is then calculated in the atmospheric component of FOCI and subsequently remapped conservatively back to the ocean grid, where it is applied as the surface DIC boundary flux (Matthes et al., 2020). Since the CO₂ flux calculation is performed on the less resolved atmospheric grid, sub-atmospheric-grid oceanic gradients are smoothed before the flux is calculated.

A 1500-year spin-up was carried out with “physics-only” FOCI, followed by a 500-year spin-up including MOPS (Chien et al., 2022). Then, historical simulations were run from 1850 to 2014, when the SSPs begin (Riahi et al., 2017). All spin-ups, control runs and forcings adhered to the CMIP6 protocol (O'Neill et al., 2016; Eyring et al., 2016). To investigate the role of the climate scenario on OAE application, a low- and high-emission SSP have been selected, namely SSP1-2.6 and SSP3-7.0. Simulations were run in emission-driven mode, such that atmospheric carbon is exchanged with the ocean and the terrestrial biosphere.

In the OAE runs, alkalinity is introduced continuously and evenly into the surface layer of the coastal grid cells outlined by the coloured line in Fig. 1a, spanning a vertical depth of roughly 3 m. As the model runs with a horizontal resolution of 0.5°, this corresponds to approximately 55 km in the meridional direction and 45 km zonally in southern Europe, decreasing to about 30 km in northern Europe. OAE is applied from 2025 to 2100, with a linear increase over the first decade of addition (2025–2034), until the equivalent of 1 Gt yr¹ of fast-reacting calcium hydroxide (Ca(OH)₂) is reached. From 2035, this amount, which equals 27 Tmol yr¹ of alkalinity (or 44 mol m² yr¹), is held constant until the end of the century (Fig. 1c). Theoretically, Ca(OH)₂ consumes two moles of atmospheric CO₂, and it produces one mole of calcium (Ca²⁺) and two moles of bicarbonate, therefore increasing alkalinity by a factor of two (Eq. 1) (Chien et al., 2022): 1Ca(OH)2+2CO2↔Ca2++2HCO3-

At the global scale, both alkalinity and DIC in FOCI-MOPS agree well with observations, while still showing a positive bias at the surface and a negative bias below 3000 m of depth. However, model fidelity lowers towards the coasts, which are less resolved with a 0.5° resolution. Though this is a common bias in global models that has been highlighted in Chien et al. (2022), here we validate FOCI-MOPS in the region where alkalinity is added using GLODAPv2 observations (Lauvset et al., 2016; Key et al., 2015) (Fig. 2).

For coastal sea surface temperature, FOCI-MOPS agrees well with GLODAPv2, with slightly lower values near the continental shelf. Towards the North Atlantic, FOCI-MOPS shows a prominent cold bias (Fig. 2c), which was already observed in previous model validation studies (Chien et al., 2022; Matthes et al., 2020). Such a feature is likely due to the lack of mesoscale dynamics, which affects the Gulf Stream path compared to an eddy-resolving ocean model (Matthes et al., 2020; Ojha et al., 2026).

Generally, surface alkalinity has a minor negative bias compared to the GLODAPv2 dataset, which increases off the coasts of Norway and towards the open ocean (Fig. 2f). DIC fields in FOCI-MOPS agree well with observations, while showing a negative bias in the North Atlantic and a positive bias south of the Irish coasts. Importantly, by the Norwegian shelf and in the southern North Sea, the strong negative bias may have an impact on the background carbonate system under OAE application. A bias toward colder, lower-alkalinity conditions also reflects the fresh bias shown in Fig. 2i, which is stronger in the North Atlantic but absent towards the coast.

While FOCI-MOPS shows good agreement with GLODAPv2 observations, we recognise that the use of a lower-resolution model may limit the accuracy of absolute values in our simulations. However, we emphasise that our analysis focuses on differences between simulations (i.e. between an OAE run and a control run, or between SSPs), such that systematic errors are expected to affect experiments similarly and therefore largely cancel.

Our analysis focuses on the comparison between two regions: the “coastline region”, which corresponds to the shelf grid cells highlighted by the dark blue line in Fig. 1b, offers details on regime shifts that take place at the epicenter of the domain; the “European region”, which corresponds to the light blue region in Fig. 1b, measures how the European system as a whole might respond to OAE. As Palmiéri and Yool (2024) showed that half of the OAE-driven carbon uptake can happen far from the injection region, this approach helps maintain some focus beyond the coastline of alkalinity addition without risking to dilute the OAE signal.

Assuming that the most visible control-to-OAE differences are detected towards the end of the century, the annual cycle averaged over the last simulation decade (2090–2099) is considered here. While ocean pCO₂ and the CO₂ flux are calculated only for the surface, alkalinity is represented over all 46 FOCI ocean layers. Thus, the variable was vertically averaged over a seasonally- and scenario-dependent mixed layer depth (MLD) (also a FOCI output variable defined from potential density), which is the region in direct contact with the atmosphere. The alkalinity that is subducted below the MLD is assumed not to induce CO₂ uptake by the ocean on short timescales (He and Tyka, 2023). Importantly, the CO₂ flux outputs shown in the Results section are taken from the coarser atmospheric grid.

In our analysis, we define the “seasonal amplitude” as the difference between the annual maximum and the annual minimum value in a given simulation, and seasons are defined as follows: winter (December, January, February), spring (March, April, May), summer (June, July, August), and autumn (September, October, November). Additionally, we present results for the OAE seasonal carbon sequestration potential, defined as the ratio of the seasonal change in CO₂ flux between a run with OAE and a run without OAE, and the amount of monthly added alkalinity carbon equivalent.

As a continental shelf pump, the European region plays an important role in carbon uptake dynamics by transferring atmospheric CO₂ into interior coastal waters and ultimately to the open ocean in the North Atlantic. As outlined above, the CO₂ flux seasonality is mainly driven by SST, which affects its chemical solubility, and biological processes, whereby carbon is used to sustain photosynthesis. Thus, to provide context for interpreting the results of our OAE runs, we analyse the SSP1-2.6 control seasonal state in the European region and in the coastline region for the CO₂ flux (Fig. 3b, g) and for some of its driving variables: SST (Fig. 3c, h), MLD-integrated NPP (Fig. 3d, i), and MLD (Fig. 3e, j).

For the air-sea CO₂ flux, the European region and the coastline region show two opposing biogeochemical geographies. Up to about 60° N, the seasonal carbon cycle is driven by temperature (Fig. 3c, h), as the largest ocean CO₂ uptake happens in winter, when CO₂ solubility is strongest, while outgassing takes place in summer, when warm temperatures inhibit chemical dissolution. In contrast, between 60 and 65° N, seasonality is driven by biological productivity: carbon escapes to the atmosphere over the cold months, when NPP is limited by sunlight and nutrient availability, whereas, from April to October, increasing productivity consumes CO₂ and ingassing is favoured.

Though this pattern is generally homogeneous throughout the two regions, localised divergences emerge. For example, north of 65°, a deeper mixed layer in the European region (Fig. 3e) seems to neutralise CO₂ outgassing, while in the coastline region, where the MLD remains relatively shallower (Fig. 3j) and SST increases relatively more (Fig. 3h), the air-sea CO₂ flux is driven by chemical solubility and the largest CO₂ uptake takes place in winter. This behaviour may be because a shallow MLD year-round does not replenish the surface ocean with sub-surface nutrient-rich waters. Additionally, temperature seasonality is stronger in the coastline region than in the European region, especially between 50 and 60° N, resulting in a relatively more pronounced CO₂ uptake (outgassing) in winter (summer).

Below, we present results on the OAE-driven seasonal changes in alkalinity, surface ocean pCO₂ and the CO₂ flux, focusing on their temporal evolution over the 2090–2099 decade (Fig. 4) as well as on their spatial differences, both vertically (Fig. 5) and horizontally (Fig. 6). Average values for the main carbonate variables are summarised in Table 1.

In the European region (Fig. 4a, b), surface alkalinity in the control run has lowest values in summer and highest values in winter, with respective extremes of 2198 and 2210 µmol kg¹ in SSP1-2.6, and of 2167 and 2184 µmol kg¹ in SSP3-7.0. Seasonal alkalinity follows MLD seasonality, whereby winter mixing deepens the mixed layer, replenishing the surface with high-alkalinity sub-surface waters. In the OAE runs, alkalinity increases in all seasons. While control values in SSP1-2.6 are higher than in SSP3-7.0, OAE increases surface alkalinity relatively less under lower emissions. In summer, when OAE enhances alkalinity the most in both scenarios, values increase by 117 µmol kg¹ and by 130 µmol kg¹ for the low- and high-emission scenario, respectively.

In the coastline region (Fig. 4a, b), where control alkalinity is similarly highest in winter and lowest in summer, OAE increases surface values up to 2672 µmol kg¹ in SSP1-2.6 and to 2685 µmol kg¹ in SSP3-7.0, both in summer. Furthermore, the amplitude of the seasonal cycle increases greatly compared to the control runs, shifting from about 19 to 109 µmol kg¹ and from 24 to 138 µmol kg¹ in the low- and high-emission scenario, respectively. As increasing emissions favour surface warming, higher surface alkalinity in SSP3-7.0 is the result of enhanced vertical stratification, whereby more of the added alkalinity is retained at the top layer.

Importantly, under both emission scenarios and in both the European and the coastline regions, alkalinity addition reverses its surface seasonal cycle, with summer values exceeding winter values. This reversed pattern is highlighted in Fig. 5, where seasonal alkalinity is shown as a function of depth. Under natural conditions, alkalinity increases with depth and winter mixing helps replenish the upper ocean from deeper layers. However, in the OAE simulations, alkalinity is added at the surface in our simulations, and stronger vertical stratification in summer leads to greater alkalinity retention, favoured by a shallower MLD.

In all OAE runs, while alkalinity mostly accumulates in the first ocean layer (top row in Fig. 5), the rest of the mixed layer (bottom row in Fig. 5) also reflects an alkalinity increase, which is higher in summer than in winter, leading to the seasonal shift that is observed in Fig. 4a, b. Surface alkalinity accumulation is higher in the coastline region, compared to the European region, due to a much higher alkalinity input in the region of addition, and in SSP3-7.0, compared to SSP1-2.6.

Spatially, under both SSPs in the OAE runs, the seasonal amplitude change of MLD-averaged alkalinity reveals strongest variation in the southern North Sea as well as by the UK coastline, where it increases by about 400 µmol kg¹. No clear amplitude change is visible in the remaining coastline region like off the coasts of Iceland, Spain and Norway. In SSP3-7.0 (Fig. 6d), the seasonal amplitude signal extends over a slightly wider area than in SSP1-2.6 (Fig. 6a), and the Norwegian and Icelandic coastlines show a small increase of the alkalinity seasonal amplitude.

Regarding the surface ocean pCO₂, the European region (Fig. 4c, d) is characterised by lowest values of 404 µatm (745 µatm) in May and highest values of 433 µatm (807 µatm) in August under SSP1-2.6 (SSP3-7.0). Minima are registered in spring because of phytoplankton bloom and consequent high NPP (Fig. 3d), whereby carbon is consumed in photosynthetic fixation. In the OAE run, surface ocean pCO₂ decreases for all seasons and spring values remain lowest, dropping to 357 µatm in SSP1-2.6 and to 674 µatm in SSP3-7.0. The seasonal amplitude decreases from 29 to 17 µatm in SSP1-2.6 and from 62 to 45 µatm in SSP3-7.0, with minimum values still detected over spring. While in the European region pCO₂ exhibits the same seasonality in both OAE and baseline simulations, in the coastline region, surface ocean pCO₂ is directly influenced by the OAE-driven reversal of the alkalinity seasonal cycle: highest (lowest) summer (winter) values in the control run become lowest (highest) in the OAE run. In SSP1-2.6 (Fig. 4c), the seasonal amplitude of the ocean pCO₂ decreases from 61 to 50 µatm, while in SSP3-7.0 (Fig. 4d) the amplitude increases from 106 to 117 µatm because OAE drives an even stronger reduction of summer values.

OAE reduces surface ocean pCO₂ in both the European region and in the coastline region, and under both emission scenarios. The largest change occurs in summer, when OAE-driven surface alkalinity has increased the most. This behaviour indicates that, in our simulations, alkalinity addition produces the greatest pCO₂ decline during periods of naturally high CO₂ outgassing, resulting in a seasonal pCO₂ amplitude damping that is relatively more pronounced under low emissions: in the European region, the seasonal amplitude is reduced by 41 % (27 %) in SSP1-2.6 (SSP3-7.0); in the coastline region, the amplitude is reduced (enhanced) by 18 % (10 %) in SSP1-2.6 (SSP3-7.0).

In the SSP1-2.6 scenario (Fig. 6b), a distinct spatial pattern of pCO₂ seasonal amplitude change driven by OAE is identified: at higher latitudes, including off the coasts of Iceland, as well as by the Norwegian and British North-West coastline, the seasonal amplitude slightly increases compared to the control; conversely, at lower latitudes, namely near the continental shelf and by the central-to-southern UK coasts, the seasonal cycle is strongly reduced. In SSP3-7.0 (Fig. 6e), results are amplified in Iceland, in the Southern North Sea, in the UK and Irish coasts. Exceptions are found in the central North Sea and Northern Spain, where strong seasonal amplification is displayed, and off the coasts of Norway, where the ocean pCO₂ seasonal cycle is reduced.

Such a strong latitudinal and scenario divergence in the seasonal amplitude of surface ocean pCO₂ may be due to different factors. North of 60° N, where the air-sea CO₂ flux is driven by its biological component (Fig. 3d), the seasonal cycle could respond differently from the European region and the coastline region, where the CO₂ flux is controlled by SST. At lower latitudes in SSP3-7.0, the increasing seasonal amplitude of surface ocean pCO₂, which is spatially confined to the coastal shelves, is probably driven by an even stronger OAE-induced pCO₂ decline over summer, as it is the case for the coastline region in Fig. 4d.

Regarding the CO₂ flux, seasonality is overall driven by temperature in our regional domain: the system outgasses (takes up) CO₂ in summer (winter), at lowest (highest) chemical solubility, and the largest CO₂ release happens between July and August. This pattern follows MLD seasonality, whereby summer warming shoals the mixed layer, reducing CO₂ uptake, and winter mixing deepens the mixed layer, enhancing the CO₂ flux (Jones et al., 2014). Under SSP1-2.6, the European region ranges from an uptake of 0.102 kg CO₂ m² yr¹ between December and January and an outgassing of 0.009 kg CO₂ m² yr¹ between July and August, while in SSP3-7.0, values are slightly more extreme: 0.142 kg CO₂ m² yr¹ uptake in winter and 0.038 kg CO₂ m² yr¹ outgassing in summer.

In the European region, with OAE deployment the ocean's CO₂ uptake potential is enhanced at all seasons: it ranges from 0.235 (0.327) kg CO₂ m² yr¹ in winter to 0.067 (0.053) kg CO₂ m² yr¹ in summer under SSP1-2.6 (SSP3-7.0). As OAE drives largest CO₂ drawdown during winter, it enhances the seasonal CO₂ flux, with a new amplitude of 0.168 kg CO₂ m² yr¹ in SSP1-2.6 and 0.274 kg CO₂ m² yr¹ in SSP3-7.0, compared to their respective controls (0.111 and 0.18 kg CO₂ m² yr¹). This results in a seasonal amplitude enhancement of 51 % and of 52 % in the low- and high-emission scenario, respectively. In the coastline region, the signal is larger than in the European region: the CO₂ flux seasonal amplitude is 0.33 kg CO₂ m² yr¹, therefore almost tripling the amplitude in the control run, which amounts to 0.114 kg CO₂ m² yr¹. With an amplitude of 0.505 kg CO₂ m² yr¹, the OAE seasonal CO₂ flux in SSP3-7.0 is more than 2.5 times larger than the control (0.2 kg CO₂ m² yr¹).

Importantly, in the OAE runs in both the European region and coastline region, excess surface alkalinity accumulates during summer, prompting a stronger reduction of surface ocean pCO₂ which, due to fast air-sea gas exchange, turns net summer outgassing into a year-round carbon sink. OAE drives the largest excess CO₂ uptake in winter, as the seasonal air-sea CO₂ flux remains in phase with the MLD cycle, regardless of seasonal variations in alkalinity (Nagwekar et al., 2024). In the coastline region, a shallow bathymetry favours year-round mixing and stronger CO₂ uptake than in the European region. As for the role of the emission scenario, SSP3-7.0 drives an even stronger carbon uptake in winter, while maintaining summer values similar to SSP1-2.6, which further increases the CO₂ flux seasonal amplitude. As it will be described later, this behaviour is due to the lowering buffering capacity of the ocean under higher emissions.

Spatially, in SSP1-2.6, OAE amplifies the CO₂ seasonal amplitude along the European coastline, and the signal remains strong until the connection with the open ocean (Fig. 6c). This boundary is set by the 627 m isobath, which separates the region with a strong signal from the open ocean, where changes are almost absent. The only exception is found off the Atlantic coasts of Spain, where the signal persists beyond the selected isobath, which may be resulting from ocean circulation dynamics that keep alkalinity at the surface and drive strong winter CO₂ uptake. The seasonal cycle is further amplified under high emissions (Fig. 6f), especially by the coasts of the southern North Sea, the UK and North-West Iceland.

The effects of OAE on carbonate seasonality are reflected in the seasonal carbon sequestration potential. While both in SSP1-2.6 and SSP3-7.0 the seasonal potential grows over the winter months and decreases over the summer months, spatial differences are present: the European region is characterised by larger variability, with values ranging between 0.2 and 0.44 (0.26, 0.61) in SSP1-2.6 (SSP3-7.0); in the coastline region values are more constant, ranging between 0.06 and 0.12 (0.1 and 0.19) in SSP1-2.6 (SSP3-7.0). Additionally, high emissions drive a larger carbon uptake potential: this is related to the scenario-dependent chemical efficiency of OAE whereby, per unit of added alkalinity, higher atmospheric CO₂ concentrations increase the air-sea disequilibrium and decrease the ocean's buffering capacity (Schwinger et al., 2024; Nagwekar et al., 2024, 2026).

Other than being more variable in the European region than in the coastline region, values are much higher in the former than in the latter, which is explained by the fact that excess carbon uptake is not confined to the place of alkalinity addition. This aspect is highlighted in panels (b) and (d): while most additional air-sea CO₂ flux takes place near the coastal shelves, with little increase over the open ocean, this excess downward flux extends beyond the region of OAE deployment (Palmiéri and Yool, 2024). Furthermore, compared to SSP1-2.6, high emissions strengthen the ocean carbon sink for the same amount of added alkalinity and within the same spatial extent, due to a lower buffer factor in a high-CO₂ world (Nagwekar et al., 2024).