IS-ENES-Data / QA-DKRZ

Quality and CF checker of meta-data in climate related data sets (NetCDF files)
4 stars 4 forks source link

Time period and range reported wrongly? #12

Closed uwefladrich closed 5 years ago

uwefladrich commented 5 years ago

While checking cmorised CMIP6 data from EC-Earth, I noticed that inconsistent time ranges are reported in the check_logs/Period/*.range and corresponding *.period files, e.g.

# Time intervals of atomic variables.

Frequency: gr, number of variables: 6
tas_Amon_EC-Earth3-Veg_amip_r1i1p1f1_gr        1979-01-01T00:00:00 - 2017-12-01T00:00:00 <--
tas_day_EC-Earth3-Veg_amip_r1i1p1f1_gr         1979-01-01T00:00:00 - 2017-12-31T00:00:00    

and

--- # Time intervals of atomic variables.

- frequency: gr
  number_of_variables: 6
  - variable: tas_Amon_EC-Earth3-Veg_amip_r1i1p1f1_gr
    begin: 1979-01-01T00:00:00
    end: 2017-12-01T00:00:00
    status: FAIL:E
  - variable: tas_day_EC-Earth3-Veg_amip_r1i1p1f1_gr
    begin: 1979-01-01T00:00:00
    end: 2017-12-31T00:00:00
    status: PASS

It looks like the last time step is missing for all variables, which means that monthly files end one month before the end of the time period while daily files are one day short. Hence, the time periods do not match.

uwefladrich commented 5 years ago

Checking the actual files that should be considered for the time period, I made sure they end at 2018-01-01T00:00:00 and that the time bounds match.

ncdump -v time_bnds [...]/tas_Amon_EC-Earth3-Veg_amip_r1i1p1f1_gr_201701-201712.nc
[...]
 time_bnds =
  [...]
  14214, 14245 ;

and

ncdump -v time_bnds [...]/tas_day_EC-Earth3-Veg_amip_r1i1p1f1_gr_20170101-20171231.nc
[...]
 time_bnds =
  [...]
  14244, 14245 ;

where 1979-01-01 00:00:00 (the reference date) plus 14245 days is 2018-01-01 00:00:00

uwefladrich commented 5 years ago

Could it be related to #10 ?

uwefladrich commented 5 years ago

Seems to be solved with recent Github version of qa-dkrz. Closing.