An update to the Surface Ocean CO 2 Atlas (SOCAT version 2)

The Surface Ocean CO2 Atlas (SOCAT) is a synthesis of quality-controlled fCO2 (fugacity of carbon dioxide) values for the global surface oceans and coastal seas with annual updates. SOCAT aims to provide data with the highest possible quality for carbon data – consistent quality control (QC) is essential in achieving this primary goal of SOCAT. Currently there are various steps of quality control, and within this task of EuroSea we aimed to develop an operational implementation of QC as a showcase for data within SOCAT from the European Research Infrastructure Integrated Carbon Observing System. The aim within EuroSea is to increase the Technology Readiness Level (TRL) from 5 (Technology validated in relevant environment) to 7 (system prototype demonstration in operational environment) for relevant ICOS data for direct submission to SOCAT. This was achieved by creating automated quality control into the ICOS state-of-art-software QuinCe, a web-based tool for processing and quality control of data from in situ sensors and underway instruments that is used for first and second level quality control for operational ICOS stations. One important aspect of SOCAT is the assessment of data quality, to ensure that all published data is fit for purpose and manual eyes-on QC is currently essential to lower uncertainties. Currently, this assessment consists of evaluating the metadata of each dataset to ensure that the correct Standard Operational Procedures (SOPs) have been followed during data collection, that the system setup is correct, instruments are calibrated and in addition examining data to ensure they are of good quality. SOCAT consists of three steps of QC: 1.) QC while data is being ingested; 2.) Eyes-on QC by regional experts and 3.) QC for the entire dataset defining the uncertainty based upon the submitted metadata and within this task it has been shown that certain parts of this QC process can be automated while other levels bear challenges if a higher level of TRL is aimed for. (EuroSea Deliverable D4.7)

Abstract. The Surface Ocean CO2 Atlas (SOCAT) is a synthesis of quality-controlled fCO2 (fugacity of carbon dioxide) values for the global surface oceans and coastal seas with regular updates. Version 3 of SOCAT has 14.7 million fCO2 values from 3646 data sets covering the years 1957 to 2014. This latest version has an additional 4.6 million fCO2 values relative to version 2 and extends the record from 2011 to 2014. Version 3 also significantly increases the data availability for 2005 to 2013. SOCAT has an average of approximately 1.2 million surface water fCO2 values per year for the years 2006 to 2012. Quality and documentation of the data has improved. A new feature is the data set quality control (QC) flag of E for data from alternative sensors and platforms. The accuracy of surface water fCO2 has been defined for all data set QC flags. Automated range checking has been carried out for all data sets during their upload into SOCAT. The upgrade of the interactive Data Set Viewer (previously known as the Cruise Data Viewer) allows better interrogation of the SOCAT data collection and rapid creation of high-quality figures for scientific presentations. Automated data upload has been launched for version 4 and will enable more frequent SOCAT releases in the future. High-profile scientific applications of SOCAT include quantification of the ocean sink for atmospheric carbon dioxide and its long-term variation, detection of ocean acidification, as well as evaluation of coupled-climate and ocean-only biogeochemical models. Users of SOCAT data products are urged to acknowledge the contribution of data providers, as stated in the SOCAT Fair Data Use Statement. This ESSD (Earth System Science Data) "living data" publication documents the methods and data sets used for the assembly of this new version of the SOCAT data collection and compares these with those used for earlier versions of the data collection (Pfeil et al., 2013; Sabine et al., 2013; Bakker et al., 2014). Individual data set files, included in the synthesis product, can be downloaded here: doi:10.1594/PANGAEA.849770. The gridded products are available here: doi:10.3334/CDIAC/OTG.SOCAT_V3_GRID.

Abstract. A well-documented, publicly available, global data set of surface ocean carbon dioxide (CO2) parameters has been called for by international groups for nearly two decades. The Surface Ocean CO2 Atlas (SOCAT) project was initiated by the international marine carbon science community in 2007 with the aim of providing a comprehensive, publicly available, regularly updated, global data set of marine surface CO2, which had been subject to quality control (QC). Many additional CO2 data, not yet made public via the Carbon Dioxide Information Analysis Center (CDIAC), were retrieved from data originators, public websites and other data centres. All data were put in a uniform format following a strict protocol. Quality control was carried out according to clearly defined criteria. Regional specialists performed the quality control, using state-of-the-art web-based tools, specially developed for accomplishing this global team effort. SOCAT version 1.5 was made public in September 2011 and holds 6.3 million quality controlled surface CO2 data points from the global oceans and coastal seas, spanning four decades (1968–2007). Three types of data products are available: individual cruise files, a merged complete data set and gridded products. With the rapid expansion of marine CO2 data collection and the importance of quantifying net global oceanic CO2 uptake and its changes, sustained data synthesis and data access are priorities.

Observing seawater CO2 concentrations is fundamental to understanding the dynamics of CO2 exchanges at the air-sea interface. Over the last decades, sustained efforts in global data collection have established the Surface Ocean CO2 Atlas (SOCAT) as the leading database for quantifying CO2 fluxes and the ocean carbon budget. SOCAT continuously integrates new data while performing annual quality reassessments, and such adjustments could significantly influence the global flux estimation. This study provides a pioneering analysis of the evolution of data updates across the five most recent SOCAT releases (SOCATv2020 to SOCATv2024) and consequences on flux estimates by mapping in situ CO2 fugacity with reanalysis data of environmental predictor variables. Five neural network ensembles are trained on SOCAT datasets to reproduce global maps of air-sea CO2 fluxes for the period 2000-2023. SOCATv2024 incorporates approximately 20% more data than SOCATv2020, revising the global ocean net CO2 sink estimate to 1.77±0.17 PgC yr-1, approximately 4.5% to 9.6% lower than earlier versions. Global uncertainty (1σ) improves by 0.3% to 7.6% compared to SOCATv2023 and SOCATv2020. However, a decline in data collection since 2017 causes a significant divergence in flux estimates, with annual net sink estimates from SOCATv2024 being 0.2 PgC yr-1 to 1 PgC yr-1 lower than previous estimates. Recent updates primarily fine-tune data coverage, with pronounced impacts on flux estimates in the Southern Ocean, upwelling zones, and subpolar Pacific.

Abstract. Various machine learning methods were attempted in the global mapping of surface ocean partial pressure of CO2 (pCO2) to reduce the uncertainty of the global ocean CO2 sink estimate due to undersampling of pCO2. In previous research, the predictors of pCO2 were usually selected empirically based on theoretic drivers of surface ocean pCO2, and the same combination of predictors was applied in all areas except where there was a lack of coverage. However, the differences between the drivers of surface ocean pCO2 in different regions were not considered. In this work, we combined the stepwise regression algorithm and a feed-forward neural network (FFNN) to select predictors of pCO2 based on the mean absolute error in each of the 11 biogeochemical provinces defined by the self-organizing map (SOM) method. Based on the predictors selected, a monthly global 1∘ × 1∘ surface ocean pCO2 product from January 1992 to August 2019 was constructed. Validation of different combinations of predictors based on the Surface Ocean CO2 Atlas (SOCAT) dataset version 2020 and independent observations from time series stations was carried out. The prediction of pCO2 based on region-specific predictors selected by the stepwise FFNN algorithm was more precise than that based on predictors from previous research. Applying the FFNN size-improving algorithm in each province decreased the mean absolute error (MAE) of the global estimate to 11.32 µatm and the root mean square error (RMSE) to 17.99 µatm. The script file of the stepwise FFNN algorithm and pCO2 product are distributed through the Institute of Oceanology of the Chinese Academy of Sciences Marine Science Data Center (IOCAS, https://doi.org/10.12157/iocas.2021.0022, Zhong, 2021.

The Surface Ocean CO2 Atlas (SOCAT) of CO2 fugacity (fCO2) observations is a key resource supporting annual assessments of CO2 uptake by the ocean and its side effects on the marine ecosystem. SOCAT data are usually released with a lag of up to 1.5 years which hampers timely quantification of recent variations of carbon fluxes between the Earth System components, not only with the ocean. This study uses a statistical ensemble approach to analyze fCO2 with a latency of one month only based on the previous SOCAT release and a series of predictors. Results indicate a modest degradation in a retrospective prediction test for 2021–2022. The generated fCO2 and fluxes for January–August 2023 show a progressive reduction in the Equatorial Pacific source following the La Niña retreat. A breaking‐record decrease in the northeastern Atlantic CO2 sink has been diagnosed on account of the marine heatwave event in June 2023.

Abstract. Since a pH sensor has become available that is principally suitable for use on demanding autonomous measurement platforms, the marine CO2 system can be observed independently and continuously by Biogeochemical Argo floats. This opens the potential to detect variability and long-term changes in interior ocean inorganic carbon storage and quantify the ocean sink for atmospheric CO2. In combination with a second parameter of the marine CO2 system, pH can be a useful tool to derive the surface ocean CO2 partial pressure (pCO2). The large spatiotemporal variability in the marine CO2 system requires sustained observations to decipher trends and study the impacts of short-term events (e.g., eddies, storms, phytoplankton blooms) but also puts a high emphasis on the quality control of float-based pH measurements. In consequence, a consistent and rigorous quality control procedure is being established to correct sensor offsets or drifts as the interpretation of changes depends on accurate data. By applying current standardized routines of the Argo data management to pH measurements from a pH / O2 float pilot array in the subpolar North Atlantic Ocean, we assess the uncertainties and lack of objective criteria associated with the standardized routines, notably the choice of the reference method for the pH correction (CANYON-B, LIR-pH, ESPER-NN, and ESPER-LIR) and the reference depth for this adjustment. For the studied float array, significant differences ranging between ca. 0.003 pH units and ca. 0.04 pH units are observed between the four reference methods which have been proposed to correct float pH data. Through comparison against discrete and underway pH data from other platforms, an assessment of the adjusted float pH data quality is presented. The results point out noticeable discrepancies near the surface of > 0.004 pH units. In the context of converting surface ocean pH measurements into pCO2 data for the purpose of deriving air–sea CO2 fluxes, we conclude that an accuracy requirement of 0.01 pH units (equivalent to a pCO2 accuracy of 10 µatm as a minimum requirement for potential future inclusion in the Surface Ocean CO2 Atlas, SOCAT, database) is not systematically achieved in the upper ocean. While the limited dataset and regional focus of our study do not allow for firm conclusions, the evidence presented still calls for the inclusion of an additional independent pH reference in the surface ocean in the quality control routines. We therefore propose a way forward to enhance the float pH quality control procedure. In our analysis, the current philosophy of pH data correction against climatological reference data at one single depth in the deep ocean appears insufficient to assure adequate data quality in the surface ocean. Ideally, an additional reference point should be taken at or near the surface where the resulting pCO2 data are of the highest importance to monitor the air–sea exchange of CO2 and would have the potential to very significantly augment the impact of the current observation network.

The development of high-quality controlled databases of sea surface partial pressure of CO2 (pCO2) combined with robust machine learning-based mapping methods that fill pCO2 gaps in time and space, enable us to quantify the oceanic air-sea CO2 exchange and its spatiotemporal variability only based on in-situ observations (pCO2-products). However, most existing pCO2-products do not explicitly include the coastal ocean or have a spatial resolution that is too coarse (e.g., 1°) to capture the highly heterogeneous spatiotemporal dynamics of pCO2 in these regions thus limiting our ability to resolve long-term trends and the interannual variability of the coastal air-sea CO2 exchange (FCO2). To address this limitation, we updated the global coastal pCO2-product of Laruelle et al. (2017) using a 2-step machine learning interpolation technique (relying on Self Organizing Maps and a Feed Forward neural Network) combined with the most extensive monthly time series for coastal waters from the Surface Ocean CO2 Atlas (SOCAT), spanning from 1982 to 2020 to reconstruct monthly high spatial resolution (0.25°) continuous coastal pCO2 maps. This updated coastal pCO2-product is then used to reconstruct the temporal evolution of the global coastal FCO2 based on observations. Our results show that since 1982, the extended coastal ocean, covering an area of 77 million km² in this study, has been acting as an atmospheric CO2 sink, removing -0.4 Pg C yr-1 (-0.2 Pg C yr-1 with a narrower coastal domain roughly equivalent to continental shelves) from the atmosphere. Moreover, the intensity of this CO2 sink has been increasing over time at a rate of 0.1 Pg C yr-1 per decade (0.03 Pg C yr-1 decade-1 in the narrower domain). The long-term change in the air-sea CO2 flux is largely driven by the air-sea pCO2 gradient, dominated by the sea surface pCO2, however wind speed and sea-ice coverage play significant roles, regionally. This new coastal pCO2-product provides a valuable constraint for understanding the strengthening of the global coastal ocean CO2 sink, fill the coastal gap in synthesis studies such as the Global Carbon Budget and serves as a benchmark for evaluating emerging results of ocean biogeochemical models.

Abstract. A new feed-forward neural network (FFNN) model is presented to reconstruct surface ocean partial pressure of carbon dioxide (pCO2) over the global ocean. The model consists of two steps: (1) the reconstruction of pCO2 climatology, and (2) the reconstruction of pCO2 anomalies with respect to the climatology. For the first step, a gridded climatology was used as the target, along with sea surface salinity (SSS), sea surface temperature (SST), sea surface height (SSH), chlorophyll a (Chl a), mixed layer depth (MLD), as well as latitude and longitude as predictors. For the second step, data from the Surface Ocean CO2 Atlas (SOCAT) provided the target. The same set of predictors was used during step (2) augmented by their anomalies. During each step, the FFNN model reconstructs the nonlinear relationships between pCO2 and the ocean predictors. It provides monthly surface ocean pCO2 distributions on a 1∘×1∘ grid for the period from 2001 to 2016. Global ocean pCO2 was reconstructed with satisfying accuracy compared with independent observational data from SOCAT. However, errors were larger in regions with poor data coverage (e.g., the Indian Ocean, the Southern Ocean and the subpolar Pacific). The model captured the strong interannual variability of surface ocean pCO2 with reasonable skill over the equatorial Pacific associated with ENSO (the El Niño–Southern Oscillation). Our model was compared to three pCO2 mapping methods that participated in the Surface Ocean pCO2 Mapping intercomparison (SOCOM) initiative. We found a good agreement in seasonal and interannual variability between the models over the global ocean. However, important differences still exist at the regional scale, especially in the Southern Hemisphere and, in particular, in the southern Pacific and the Indian Ocean, as these regions suffer from poor data coverage. Large regional uncertainties in reconstructed surface ocean pCO2 and sea–air CO2 fluxes have a strong influence on global estimates of CO2 fluxes and trends.

The Southern Ocean is a complex system yet is sparsely sampled in both space and time. These factors raise questions about the confidence in present sampling strategies and associated machine learning (ML) reconstructions. Previous studies have not yielded a clear understanding of the origin of uncertainties and biases for the reconstructions of the partial pressure of carbon dioxide (pCO2) at the surface ocean (pCO2ocean). Here, we examine these questions by investigating the sensitivity of pCO2ocean reconstruction uncertainties and biases to a series of semi-idealized observing system simulation experiments (OSSEs) that simulate spatio-temporal sampling scales of surface ocean pCO2 in ways that are comparable to ocean CO2 observing platforms (Ship, Waveglider, Carbon-float, Saildrone). These experiments sampled a high spatial resolution (±10 km) coupled physical and biogeochemical model (NEMO-PISCES) within a sub-domain representative of the Sub-Antarctic and Polar Frontal Zones in the Southern Ocean. The reconstructions were done using a two-member ensemble approach that consisted of two machine learning (ML) methods, (1) the feed-forward neural network and (2) the gradient boosting machines. With the baseline observations being from the simulated ships mimicking observations from the Surface Ocean CO2 Atlas (SOCAT), we applied to each of the scale-sampling simulation scenarios the two-member ensemble method ML2, to reconstruct the full sub-domain pCO2ocean and assess the reconstruction skill through a statistical comparison of reconstructed pCO2ocean and model domain mean. The analysis shows that uncertainties and biases for pCO2ocean reconstructions are very sensitive to both the spatial and temporal scales of pCO2 sampling in the model domain. The four key findings from our investigation are the following: (1) improving ML-based pCO2 reconstructions in the Southern Ocean requires simultaneous high resolution observations of the meridional and the seasonal cycle (< 3 days) of pCO2ocean; (2) Saildrones stand out as the optimal platforms to simultaneously address these requirements; (3) Wavegliders with hourly/daily resolution in pseudo-mooring mode improve on Carbon-floats (10-day period), which suggests that sampling aliases from the low temporal frequency have a greater negative impact on their uncertainties, biases and reconstruction means; and (4) the present summer seasonal sampling biases in SOCAT data in the Southern Ocean may be behind a significant winter bias in the reconstructed seasonal cycle of pCO2ocean.

The Southern Ocean is a complex system yet is sparsely sampled in both space and time. These factors raise questions about the confidence in present sampling strategies and associated machine learning (ML) reconstructions. Previous studies have not yielded a clear understanding of the origin of uncertainties and biases for the reconstructions of the partial pressure of carbon dioxide (pCO2) at the surface ocean (pCO2ocean). Here, we examine these questions by investigating the sensitivity of pCO2ocean reconstruction uncertainties and biases to a series of semi-idealized observing system simulation experiments (OSSEs) that simulate spatio-temporal sampling scales of surface ocean pCO2 in ways that are comparable to ocean CO2 observing platforms (Ship, Waveglider, Carbon-float, Saildrone). These experiments sampled a high spatial resolution (±10 km) coupled physical and biogeochemical model (NEMO-PISCES) within a sub-domain representative of the Sub-Antarctic and Polar Frontal Zones in the Southern Ocean. The reconstructions were done using a two-member ensemble approach that consisted of two machine learning (ML) methods, (1) the feed-forward neural network and (2) the gradient boosting machines. With the baseline observations being from the simulated ships mimicking observations from the Surface Ocean CO2 Atlas (SOCAT), we applied to each of the scale-sampling simulation scenarios the two-member ensemble method ML2, to reconstruct the full sub-domain pCO2ocean and assess the reconstruction skill through a statistical comparison of reconstructed pCO2ocean and model domain mean. The analysis shows that uncertainties and biases for pCO2ocean reconstructions are very sensitive to both the spatial and temporal scales of pCO2 sampling in the model domain. The four key findings from our investigation are the following: (1) improving ML-based pCO2 reconstructions in the Southern Ocean requires simultaneous high resolution observations of the meridional and the seasonal cycle (< 3 days) of pCO2ocean; (2) Saildrones stand out as the optimal platforms to simultaneously address these requirements; (3) Wavegliders with hourly/daily resolution in pseudo-mooring mode improve on Carbon-floats (10-day period), which suggests that sampling aliases from the low temporal frequency have a greater negative impact on their uncertainties, biases and reconstruction means; and (4) the present summer seasonal sampling biases in SOCAT data in the Southern Ocean may be behind a significant winter bias in the reconstructed seasonal cycle of pCO2ocean.

Observational networks monitoring marine carbon variables are established to meet the critical need to estimate ocean CO2 uptake, as well as assessing its consequences on ocean health through changes in carbonate chemistry (ocean acidification). Despite considerable efforts over the past decades, data coverage is still sparse over large ocean regions, prompting the implementation of mapping methods to gap-fill carbon datasets over the globe. Different statistical approaches have been proposed with the aim to generate reconstructions of the complete marine CO2 system at high spatial-temporal resolutions. Following this goal, we first introduce a global reconstruction of surface ocean partial pressure of CO2 (pCO2) at monthly and 0.25-degree resolutions over the period 1985-2021. This high-resolution pCO2 product is derived from ensemble neural network models interpolating monthly gridded observation-based data from Surface Ocean CO2 ATlas (SOCAT). We will assess the ability of the proposed pCO2 ensemble (1) to derive long-term time series of pCO2 and associated 1-sigma uncertainty per 0.25-degree grid cell for each month, (2) to reproduce temporal and horizontal gradients of coastal pCO2 observations in comparison with a coarser spatial resolution, (3) to estimate surface ocean pH and air-sea CO2 fluxes. Furthermore, we will present an extension of the ensemble neural network models, which is referred to as a new module extrapolating pCO2 to several years ahead. The extended ensemble-based approach will ultimately be used to project global ocean CO2 uptake and ocean acidification with low latency.

The Southern Ocean is a complex system yet is sparsely sampled in both space and time. These factors raise questions about the confidence in present sampling strategies and associated machine learning (ML) reconstructions. Previous studies have not yielded a clear understanding of the origin of uncertainties and biases for the reconstructions of the partial pressure of carbon dioxide (pCO2) at the surface ocean (pCO2ocean). Here, we examine these questions by investigating the sensitivity of pCO2ocean reconstruction uncertainties and biases to a series of semi-idealized observing system simulation experiments (OSSEs) that simulate spatio-temporal sampling scales of surface ocean pCO2 in ways that are comparable to ocean CO2 observing platforms (Ship, Waveglider, Carbon-float, Saildrone). These experiments sampled a high spatial resolution (±10 km) coupled physical and biogeochemical model (NEMO-PISCES) within a sub-domain representative of the Sub-Antarctic and Polar Frontal Zones in the Southern Ocean. The reconstructions were done using a two-member ensemble approach that consisted of two machine learning (ML) methods, (1) the feed-forward neural network and (2) the gradient boosting machines. With the baseline observations being from the simulated ships mimicking observations from the Surface Ocean CO2 Atlas (SOCAT), we applied to each of the scale-sampling simulation scenarios the two-member ensemble method ML2, to reconstruct the full sub-domain pCO2ocean and assess the reconstruction skill through a statistical comparison of reconstructed pCO2ocean and model domain mean. The analysis shows that uncertainties and biases for pCO2ocean reconstructions are very sensitive to both the spatial and temporal scales of pCO2 sampling in the model domain. The four key findings from our investigation are the following: (1) improving ML-based pCO2 reconstructions in the Southern Ocean requires simultaneous high resolution observations of the meridional and the seasonal cycle (< 3 days) of pCO2ocean; (2) Saildrones stand out as the optimal platforms to simultaneously address these requirements; (3) Wavegliders with hourly/daily resolution in pseudo-mooring mode improve on Carbon-floats (10-day period), which suggests that sampling aliases from the low temporal frequency have a greater negative impact on their uncertainties, biases and reconstruction means; and (4) the present summer seasonal sampling biases in SOCAT data in the Southern Ocean may be behind a significant winter bias in the reconstructed seasonal cycle of pCO2ocean.

The rate of ocean uptake of anthropogenic CO2 has declined over the past decade, so a critical question for science and policy is whether the ocean will continue to act as a sink. Large areas of the ocean remain without observations for carbonate system variables, and oceanic CO2 observations have declined since 2017. The Mediterranean Sea is one such an area, especially its eastern part, where there is a paucity of carbonate system data, with large areas not sampled or only sampled by ship-based discrete measurements as opposed to high frequency, sensor-equipped time-series fixed stations. The aim of this study was to analyze a multi-year time-series of high-frequency (hourly) partial pressure CO2 (pCO2) and pH measurements in the Eastern Mediterranean, along with low-frequency (monthly) measurements of total dissolved inorganic carbon and total alkalinity. The pCO2 time-series was the first obtained in the Eastern Mediterranean. The study was conducted at a fixed platform of the POSEIDON system (Heraklion Coastal Buoy) located near Crete Island. Temperature was the dominant factor controlling the temporal variability of pCO2 and pH, while the remaining non-thermal variability appeared to be related to evaporation, water mixing, and biological remineralization-production. The air-sea CO2 fluxes indicated a transition from a winter-spring sink period to a summer-autumn source period. The annual air-sea CO2 flux was too low (-0.16 ± 0.02 mol m-2 yr-1) and variable to conclusively characterize the area as a net source or sink of CO2, highlighting the need for additional high frequency observation sites. Algorithms were developed using temperature, chlorophyll and salinity data to estimate pCO2 and total alkalinity, in an effort to provide tools for estimates in poorly observed areas/periods from remotely sensed products. The applicability of the algorithms was tested using Surface Ocean CO2 Atlas (SOCAT) data from the Eastern Mediterranean Sea (1999 to 2020) which showed that the algorithm pCO2 estimates were generally within ±20 μatm of the pCO2 values reported by SOCAT. Finally, the integration and analysis of the data provided directions on how to optimize the observing strategy, by readapting sensor location and using estimation algorithms with remote sensing data.

Observation‐based quantification of ocean carbon dioxide (CO2) uptake relies on synthesis data sets such as the Surface Ocean CO2 ATlas (SOCAT). However, the data collection effort has dramatically declined and the number of annual data sets in SOCATv2023 decreased by ∼35% from 2017 to 2021. This decline has led to a 65% increase (from 0.15 to 0.25 Pg C yr−1) in the standard deviation of seven SOCAT‐based air‐sea CO2 flux estimates. Reducing the availability of the annual data to that in the year 2000 creates substantial bias (50%) in the long‐term flux trend. The annual mean CO2 flux is insensitive to the seasonal skew of the SOCAT data and to the addition of the lower accuracy data set available in SOCAT. Our study highlights the need for sustained data collection and synthesis, to inform the Global Carbon Budget assessment, the UN‐led climate negotiations, and measurement, reporting, and verification of ocean‐based CO2 removal projects.