Abstract of associated paper: Many studies on future climate projection point out that, with progressing of global warming, upper-ocean stratification will strengthen over this century and consequently global-averaged ocean primary production will decrease. Observed long-term changes in the stratification to date, however, still show large uncertainties of the change itself and its driver. Focusing on the vertical difference in the emergence of the global warming signals, we used only observational profiles to describe the spatiotemporal characteristic of long-term trend and variability in the upper-ocean stratification. Rapid strengthening of the stratification (defined as the density difference between the surface and 200 m depth) since the 1960s was detected over most of the global ocean. Although the global average increase over 58 years (1960–2017) corresponds to 3.3–6.1% of the mean stratification, these strengthening trends considerably change depending on the regions. In addition to the well-documented explanation of strengthening stratification, namely that the surface intensification of global warming signal, we found that changes in subsurface temperature and salinity stratification associated with changes in atmospheric/ocean circulations and the global water cycle significantly contribute to the long-term change in the stratification and setting its regional difference. In mid- and high-latitude ocean of the northern hemisphere, the long-term trend in density stratification has noteworthy seasonality, which shows faster increase in boreal summer than that in winter. From the detrended time series, interannual variabilities correlated with a particular climate mode are detected in several ocean regions, suggesting that these variabilities are mainly driven by associated sea surface temperature variation.

The two dataset include processed data (ascii format) used to calculate long-term trends of the density stratification and density. Original observational profiles are from World Ocean Databese 2013 (Download: May/29/2018). Six observational platform (CTD, PFL, OSD, MOU, GLD, UOR) are used. After QC procedures, anomalies are annually averaged over 5 (latitude) x 10 (longitude) grid. Please see the main text for the detail of the analytical procedure. RA300.1960-2017.5x10.6DS.ANNUAL.dat -Year, lon, lat, number of grids used for area-averaging, potential density anomaly @ 10m, 50m, 100m, 150m, 200m, 250m, 300m DRA300.1960-2017.5x10.6DS.ANNUAL.dat -Year, lon, lat, number of grids used for area-averaging, anomaly of potential density difference between 10m and 10m, 50m, 100m, 150m, 200m, 250m, 300m Missing value: -0.9999E+02