2017 was the Warmest Year on Record for the Global Ocean

The IAP provides an observation-based ocean temperature analysis from the sea surface down to a depth of 2000 m, available from the year 1940 (http://159.226.119.60/cheng/). All the available in-situ ocean observations from the National Oceanic and Atmospheric Administration/National Centers for Environmental Information (NOAA/NCEI) (Boyer et al., 2013) are used in the analysis. For the current century, the Argo network provides observations with near-global ocean coverage, significantly improving the availability of ocean observations (Argo, 2000).

The importance of ocean heat content in climate monitoring is thoroughly discussed in (Cheng et al., 2018). A gridded ocean temperature dataset with complete global ocean coverage is a highly valuable resource for understanding climate change and variability. The IAP provides an objective analysis of historical ocean subsurface temperature through several innovative steps. First, an updated set of past observations, newly corrected for biases [e.g., in Expendable Bathythermograph - XBTs], is used. The XBT bias was corrected by the CH14 scheme (Cheng et al., 2014), which is recommended by the XBT community. Second, the covariability between values at different places in the ocean and background information from several climate models that include a comprehensive ocean model is used. Third, the influence of each observation has been extended over a larger area, recognizing the relative homogeneity of the vast open expanses of the southern oceans. Subsequently, the observations can also be used to provide finer-scale detail. And finally, the new analysis has been carefully evaluated using knowledge of recent well-observed ocean states, but subsampled using the sparse distribution of observations in the more distant past to show that the method produces unbiased historical reconstruction. The method works well back to the late 1950s, but prior to then there are too few observations to make reliable ocean state estimates. The methodology is summarized in (Cheng et al., 2017).