Temperature series and map data

Data for the buoy temperature record from the paper as well as an updated record based on ICOADS version 3 are provided below. We aim to update the version 3.0 data quarterly.

The data are prepared following the method from the paper as follows:

  • The ERSSTv3 data are baselined on the period 1997-2015, requiring data from at least 10 years for each grid cell and month of the year. The remaining cells are set to missing.
  • The buoy record is constructed using the default quality control from the paper, using the OISST 0.5 degree climatology.
  • The buoy record is infilled by kriging using the method of Cowtan and Way (2014).
  • The buoy record is expanded to a 1 degree grid.
  • The buoy record is then matched against the ERSSTv3 baseline, requiring data from at least 5 years for each grid cell and month of the year. The remaining cells are set to missing, as are any cells not also present in ERSSTv4.

The buoy data are converted from the equal area grid to a fine equal angle grid without smoothing: the original grid is visible in the gridded anomaly values.

As with any temperature data, valid comparison of records requires that both records be reduced to common coverage, so comparisons should be performed using the netcdf map series using tools such as CDO or NCO. However coverage of the buoy record is only slightly lower than ERSSTv3 or ERSSTv4, so for convenience time series are provided for the full coverage of the buoy record by calculating an area weighted average on the 1 degree grid using a land mask. These series provide a quick-and-dirty comparison against the ERSST products, however for publication quality results or comparison to other products the map series should be used.


BETA RELEASE - Please report problems to Kevin Cowtan