Opened 2 weeks ago

Closed 7 days ago

#2850 closed help (fixed)

Meaning question

Reported by: ChrisWells Owned by: um_support
Component: UM Model Keywords:
Cc: Platform: NEXCS
UM Version:

Description

I’ve noticed that the model outputs a different value in annual mean than the average of the monthly values for the same year – i.e. if I download the apy.pp 2014 diagnostic surface temperature, it doesn’t match what I get if I download the 2014 apm.pp and take the average.


Why is this? I guess it might be due to some weighting to account for the assumed 30 day months, but am unsure. I suppose it also means I shouldn’t compare results which I make from different time profile outputs, and not should I calculate values using variables with different time profiles?


Cheers,

Chris

Change History (6)

comment:1 Changed 2 weeks ago by willie

Hi Chris,

Could you let us know the path to these files please. Also which suite are you running?

You can find further information on meaning at https://code.metoffice.gov.uk/doc/um/latest/papers/umdp_C05.pdf.

Willie

comment:2 Changed 2 weeks ago by ChrisWells

Hi Willie,

I did a quick test in here: /home/d00/chwel/expts/cmip6/tests/extrastash/airqual/ncra_check

I downloaded, for suite u-bf926, the apy.pp and apm.pp surface temperatures for 1851. I took the time average of monthly (using ncra) and subtracted this from the annual file, to give temp_annual-monthly_ncra.nc . I thought this would be 0 all over, but it isn't. I also took the global surface average (using nco commands again), and compared the annual global average value to the average of the 12 monthly global average values. These averages were very similar (287.9161 vs 287.9059), but the spatial map has some larger differences.

Do you know what's happening here? I could easily have made a mistake, and I'm unsure how reliable nco commands are considered to be.

Many thanks,
Chris

comment:3 Changed 8 days ago by grenville

Chris

Where have you put the apy.pp and apm.pp data?

Grenville

comment:4 Changed 8 days ago by ChrisWells

Hi Grenville,

the data is all in that directory; I use a bash script (download_template.sh) to download data - it uses a Query file (/home/d00/chwel/Queries/extrastash_ncra_check) to run a moo select query, on the stash codes in the query file - stash 24 in this case. (these are not my original scripts but they follow the methods given in the moose user guide http://gws-access.ceda.ac.uk/public/mohc_shared/moose-user-doc/external_user_guide.html).

Should I try it the long way, by downloading the whole apy and apm files?

Cheers,
Chris

comment:5 Changed 8 days ago by grenville

Chris

I don't know what to look at in /home/d00/chwel/expts/cmip6/tests/extrastash/airqual/ncra_check - there are too many files with names and contents which I can't fathom.

Grenville

comment:6 Changed 7 days ago by ChrisWells

  • Resolution set to fixed
  • Status changed from new to closed

Hi Grenville,

apologies for the confusion. I've taken another look and think I've solved this now.

I realised that even though I ran suite u-bf926 for 2 years, 185001-185201, there is only one file in apy.pp, namely bf926a.py18511201.pp. I didn't understand before, but I guess that the annual files aren't from Jan-Dec inclusive, but in fact from Dec-Nov inclusive, to track the DJF etc standard seasons, so it covers 4 whole seasons. So I downloaded the data from that apy.pp file, and the apm.pp files from 185012-185111 inclusive, and the average of them (my temp_monthly_ncra.nc) is essentially identical to the yearly file (my temp_annual.nc).

Apologies for the hassle; I'll close this ticket now.

Cheers,
Chris

Note: See TracTickets for help on using tickets.