timmean and the time value stored inside the netCDF file
Added by e ddc about 11 years ago
Hello,
I am using cdo timemean and I think the output file has the wrong time stamp; it should be halfway through the period but I think it defaults to the end of the specified time window.
In the version I tested it was set at the end of the time-averaging period {t1,t2}. E.g.
- average over period t1 --> t2
- NC file sets time to t2
- but proper CF-compliance would suggest setting time to ((t1 + t2) / 2)
more specifically:
cdo timmean -select,name=pr pr_day_MPI-ESM-LR_1pctCO2_r1i1p1_18500101-18591231.nc o.nc
produces a o.nc file with
time_bnds =
0, 3652 ;and
time = 3651.5 ;Cheers,
Eduardo
Data can be download here:
http://bmbf-ipcc-ar5.dkrz.de/thredds/esgcet/1/cmip5.output1.MPI-M.MPI-ESM-LR.1pctCO2.day.atmos.day.r1i1p1.v20111006.html?dataset=cmip5.output1.MPI-M.MPI-ESM-LR.1pctCO2.
Replies (1)
RE: timmean and the time value stored inside the netCDF file - Added by Uwe Schulzweida about 11 years ago
Hi Eduardo,
Yes the default is the time stamp of the last contributing time step of the averaging period. We have already started to change it in the module runSTAT (runmean, ...). The necessary changes for timmean is on the ToDo list, but it could take some month to find time for the implementation.
Thanks for this hint!
Cheers,
Uwe