Bias-corrected upper-ocean data for impact assessments of global warming and assimilation into climate models

We developed monthly time-series datasets of upper-ocean temperature, salinity, and sea-ice concentration for non-warming (NW) and future climate conditions, which are intended to be assimilated into the ocean components of climate models to improve the accuracy of simulations under various climate conditions. Monthly time series of upper-ocean temperature and salinity from an observation-based objective analysis over the past 60 years were decomposed into climatological, externally forced, and internal variability components. The climatological component was defined as the 60-year monthly mean. The externally forced components were extracted using empirical orthogonal function analysis applied to climate model simulations from the Coupled Model Intercomparison Project Phase 6 (CMIP6). Internal variability was estimated as the residual obtained by removing the climatological and externally forced components from the original time series. The NW dataset was constructed by adding the monthly climatology for 1850–1899, which was estimated as the climatology for the past 60 years after removing the externally forced component, to the internal variability component. The NW dataset represents climate conditions with a global-mean sea surface temperature approximately $0.37\ ^\circ\mathrm{C}$ lower than that of the present climate. Future datasets were constructed consistently with the Shared Socioeconomic Pathway scenarios and include scenario-dependent warming trends and sea-ice declines. This framework corrects climatological biases in CMIP6 simulations while preserving externally forced changes. The resulting dataset, when assimilated into the ocean-sea ice components of climate models, would provide a robust basis for quantifying the impacts of anthropogenic warming in climate model simulations.