Extended Reconstructed Sea Surface Temperature, Version 6 (ERSSTv6). Part II: Upgrades on Quality Control and Large-Scale Filter

NOAA’s Extended Reconstructed Sea Surface Temperature (ERSST) is a monthly 2°SST product starting from 1850. Our Part I study indicated that the performance scores of spatial correlation coefficient (SCC) and root-mean-square difference (RMSD) dropped clearly after the mid-1970s in the analysis of ERSST with an artificial neural network (ANN) method. In this Part II study, we demonstrate that ERSST with the ANN method can further be improved progressively in the final ERSSTv6 by the following steps: 1) applying a nearest neighbor check (NNC) quality-control algorithm on ship observations, 2) applying a large-scale (>200 km) filter (LS200) on SST superobservations, and 3) upgrading algorithms in proxy SST from ice concentration. These progressive improvements were assessed against validation and observation datasets. In comparison with ERSST with the ANN method alone, the quality of ERSSTv6 improves in the statistical metrics of SCC and RMSD by 2%–11% and 0.01°–0.24°C, respectively, in the global oceans. In the ice-covered regions, SST bias and RMSD decrease by 0.67° and 0.29°C, respectively. Significance Statement The ERSSTv6 using an artificial neural network (ANN) is further improved by progressively implementing the nearest neighbor check (NNC) in quality control (QC) on ship observations, large-scale filter (LS200) on SST superobservations, and ice-SST proxy algorithm in the ice-covered regions.