Inconsistency between TOGA (ERSST v4) output SST and the actual ERSSTv4

[kezhoulumelody]

To whom it may concern,

I am looking at the CAM5 Prescribed SST AMIP ensembles (CESM Climate Variability & Change Working Group), and found something pretty odd. I looked at the monthly mean output of SST from TOGA (ERSST v4) during the period of 1984-2014, and when I compared it to the actual ERSST v4 data (Index of /pub/data/cmb/ersst/v4/netcdf), I found the SST data are not the same. The inconsistency of SST from two data source is shown below:

Both dataset are interpolated into same grids through "cdo remapbil,r360x180", the time plotted here is 1984/01. The differences is calculated as ERSSTv4 minus TOGA.




I assume TOGA is forced with monthly time-varying SST from ERSSTv4, and the SST at least in the tropics should be exactly identical. But I could interpret the TOGA experiment setup wrong or compared it to a wrong ERSST data set.

Thanks,
Melody

 

[kezhoulumelody]

I figured out the issue: I used "cdo selyear,1984/2014" command to select corresponding years but somehow the time window selected by this method is actually 1983/2013. Then I used "cdo seltimestep" instead, and both SST should be pretty much identical (there is very small biases around 0.001 K).

Melody