This data package contains a hybrid surface OA data product that is produced based on three recent observational data products: (a) the Surface Ocean CO2 Atlas (SOCAT, version 2022), (b) the Global Ocean Data Analysis Product version 2 (GLODAPv2, version 2022), and (c) the Coastal Ocean Data Analysis Product in North America (CODAP-NA, version 2021), and 14 Earth System Models from the sixth phase of the Coupled Model Intercomparison Project (CMIP6). The trajectories of ten OA indicators, including fugacity of carbon dioxide, pH on Total Scale, total hydrogen ion content, free hydrogen ion content, carbonate ion content, aragonite saturation state, calcite saturation state, Revelle Factor, total dissolved inorganic carbon content, and total alkalinity content are provided under preindustrial conditions, historical conditions, and future Shared Socioeconomic Pathways: SSP1-19, SSP1-26, SSP2-45, SSP3-70, and SSP5-85 from 1750 to 2100 on a global surface ocean grid. These OA trajectories are improved relative to previous OA data products with respect to data quantity, spatial and temporal coverage, diversity of the underlying data and model simulations, and the provided SSPs over the 21st century.

Dataset Description: Gridded monthly data products of surface ocean acidification indicators from 1998 to 2024 and on a 0.25° by 0.25° spatial grid have been developed for eleven U.S. Large Marine Ecosystems (LMEs) using a machine learning algorithm called random forest regression (RFR). The data products are called RFR-LMEs, and were constructed using observations from the Surface Ocean CO2 Atlas — co-located with surface ocean properties from various satellite, reanalysis, and observational products — with an approach that utilized two types of machine learning algorithms: (1) Gaussian mixture models to cluster the data into subregions with similar environmental variability and (2) RFRs that were trained and applied separately in each cluster to interpolate the observational data in space and time. RFR-LMEs also rely on previously published seawater property estimation routines to obtain the full suite of ocean acidification indicators. The products show a domain-wide carbo n dioxide partial pressure increase of 1.6 ± 0.4 μatm yr−1 and pH decrease of 0.0015 ± 0.0004 yr−1. More information on the creation and validation of RFR-LMEs is available in the following publication: Sharp, J.D., Jiang, L., Carter, B.R., Lavin, P.D., Yoo, H., Cross, S.L., 2024. A mapped dataset of surface ocean acidification indicators in large marine ecosystems of the United States. Scientific Data, 11, 715, 10.1038/s41597-024- 03530-7.

This dataset consists of individual data of temperature, salinity, dissolved inorganic carbon (DIC) and total alkalinity (TALK) collected in the Global Ocean and Mediterranean Sea. This is a synthesis of French oceanographic cruises conducted between 1993 and 2022.

This dataset consists of carbonate chemistry discrete profile data collected on two of the Northeast US Coast Shelf LTER cruises, EN617 and EN627 summer 2018 and winter 2019. The dataset includes in situ temperature, salinity, and dissolved oxygen as measured from a CTD, and also includes discrete samples for dissolved inorganic carbon and total alkalinity.

This dataset contains Mapped Observation-Based Oceanic Dissolved Inorganic Carbon Monthly fields from 2004 through 2019. We increase the temporal resolution of the monthly climatology of MOBO-DIC (Keppler et al., 2020a) to resolve fields of DIC from January 2004 through December 2019. MOBO-DIC2004-2019 consists of time-varying, gap-filled mapped fields of DIC on 28 depth levels in the upper 1500 m on a 1°x1° grid, at monthly resolution. The original method for Keppler et al. (2020a) as well as an analysis of the seasonal dynamics of DIC at a global scale can be found in Keppler et al. (2020b). The MOBO-DIC mapping method is an extension and adaptation of the SOM-FFN approach by Landschützer et al. (2013), where the first step is to cluster the ocean into regions of similar physical and biogeochemical properties using self-organizing maps (SOM). In the second step, we run a feed-forward network (FFN) in each SOM-cluster to approximate and apply the statistical relationship between the target data (here: DIC), and better constrained predictor data that are available as mapped global fields. We adapted the SOM-FFN method in several ways compared to the original method by Landschützer et al. (2013), that mapped oceanic surface pCO2. As we map the DIC in the water column, we extend the mapping grid from three dimensions (latitude, longitude, and time), to four (latitude, longitude, time, and depth). As different predictors are available and/or meaningful when mapping DIC in the water column, we also have a different set of predictor data compared to the approach used by Landschützer et al. (2013). To overcome potential biases in the random selection of training and internal validation data, as well as boundary problems associated with the SOM clustering, we use a bootstrapping approach, running the SOM-FFN method 15 times. We use 3 different set-ups for the SOMs and run 5 slightly different FFNs in each of the SOM clusters. We take the mean across this ensemble as our final DIC fields. Due to data availability of the predictors, and different statistical relationships in the upper and deep ocean, we run the method separately for two depth slabs: from the surface to 500m, and from 500m to 1500 m. Thus, there may be small discontinuities at 500 m due to this boundary problem, but they are well within the uncertainties. We calculate the uncertainty based on three components: the prediction uncertainty (the standard deviation across the ensemble, global mean is approx. 7 μmol/kg), the uncertainty associated with the measurements (2.4 μmol/kg), and the uncertainty associated with the representation (16 μmol/kg). We use standard error propagation of these three components to obtain the overall uncertainty of MOBO-DIC2004-2019 (global mean is approximately 18 μmol kg−1). We want to emphasize that the uncertainties in our mapped estimate of DIC are considerably larger than the general uncertainties in direct observations of DIC. Thus, they must be considered in the interpretation of the data. Due to how the mapping method works, MOBO-DIC is most robust when using averages or integrals over large regions. For the full description of the method and its validation, please refer to both the Main Text and the Supporting Information of Keppler et al. (in review.).

This dataset contains mapped observation-based oceanic dissolved inorganic carbon (DIC), monthly climatology from January to December (based on observations between 2004 and 2017), from the Max-Planck-Institute for Meteorology (MOBO-DIC_MPIM). The SOM-FFN approach by Landschützer et al. (2013) was extended and applied to obtain time-varying gap-filled mapped fields of dissolved inorganic carbon (DIC) in the water column. In the SOM-FFN approach, the first step is to cluster the ocean into regions of similar physical and biogeochemical properties using self-organizing maps (SOM). In the second step, a feed-forward network (FFN) is run in each SOM-cluster to approximate and apply the statistical relationship between the target data (here: DIC), and better constrained predictor data that are available as mapped global fields. The SOM-FFN method was adjusted and in several ways compared to the original method by Landschützer et al. (2013), that mapped oceanic surface pCO2. As we map the DIC in the water column, we extended the mapping grid from three dimensions (latitude, longitude, and time), to four (latitude, longitude, time, and depth), and instead of monthly inter-annual fields, we resolved a monthly climatology based on the period from 2004 through 2017. As different predictors are available and/or meaningful when mapping DIC in the water column, we also have a different set of predictor data compared to the approach used by Landschützer et al. (2013). To overcome potential biases in the random selection of training and internal validation data, a bootstrapping approach was used, running the SOM-FFN method ten times. The mean across this ensemble was taken as the final DIC field. We defined the standard deviation across the ensemble as the uncertainty within the method, and name it ensemble spread.

This dataset consists of time-resolved reconstructions of ocean interior acidification from 1800 through 1994, 2004, and 2014. The basis of these reconstructions are observation-based estimates of the accumulation of anthropogenic carbon, combined with climatologies of hydrographic and biogeochemical properties in the ocean interior. Acidification trends are determined for several parameters of the marine CO2 system, namely the saturation state of aragonite (Ωarag), the carbonate ion concentration ([CO32-]), the free proton concentration ([H+]), and pH on the total scale (pHT). The underlying anthropogenic carbon concentration (ΔCant), the computed sensitivities of the four marine CO2 system parameters and their absolute state estimates are provided as well. The datasets contain in addition to the standard estimate also 14 sensitivity cases, which are intended to assess the robustness of our acidification estimates to changes in the estimation procedure of ΔCant as well as the climatological distributions of other hydrographic properties. All estimates are provided on a horizontal grid with 1° x 1° resolution and for 28 depth layers from 0 - 3000m. These data provide strong constraints on ocean interior acidification over the industrial era, unravelling in particular its progression since 1994.