Global mappings of surface ocean CO2 fugacity and air-sea fluxes over the years 2000-2023

This data package includes global datasets of the estimated sea surface CO2 fugacity (fCO2) and air-sea fluxes over the 24–year period (2020–2023) at monthly and 1° resolutions. CO2 fugacity is estimated by using a machine-learning approach mapping the target variable and its proxies (e.g., sea surface temperature SST, salinity SSS, surface height SSH, mixed layer depth MLD, chlorophyll-a CHL, CO2 mole fraction xCO2). We extract global monthly, 1° gridded data of fCO2 since the year 2000 in the five latest Surface Ocean CO2 Atlas versions (SOCATv2020 to SOCATv2024). For each SOCAT version, we run an ensemble of feed-forward neural network models, namely CMEMS-LSCE-FFNN, developed by Chau et al. (2022) to estimate CO2 fugacity. Air-sea fluxes are computed as a function of fCO2, SST, and Wind speed. Each data file was written in netCDF and named under the following format CO2 fugacity: all_fuCO2_meanstd_vYYYY_r100_2000to2023.nc Air-sea fluxes: all_fluxCO2_meanstd_vYYYY0_r100_2000to2023.nc Other variables: e.g., SST_r100_2000to2023.nc where YYYY is the year corresponding to a SOCAT release. For further information, please refer to our article "Impact of Surface Ocean CO2 Atlas (SOCAT) observing network extensions on the quantification of global air-sea CO2 fluxes" published in the Science of The Total Environment (STOTEN) journal (DOI: https://doi.org/10.1016/j.scitotenv.2025.180265)