CDO

Why keep separate files for seasons/months when cdo has seasonal/monthly operators? Why not mergetime all your files into one large file?

$ cdo mergetime precip_WJB_int.????_???.BN_1958_2013_1km.DD.nc precip_WJB_int.BN_1958_2013_1km.DD.nc

Once you have done this, the trick to getting meteorological seasons is shifttime (https://code.mpimet.mpg.de/projects/cdo/embedded/index.html#x1-2410002.6.4). shifttime back one month before any seasonal calculations(i.e. seastat https://code.mpimet.mpg.de/projects/cdo/embedded/index.html#x1-5070002.8.27). Then shifttime forward one month afterwards to correct the date.

$ cdo -shifttime,1month -seasmax -shifttime,-1month precip_WJB_int.BN_1958_2013_1km.DD.nc precip_WJB_int.BN_1958_2013_1km.DD_seasmax.nc

timstat operators (https://code.mpimet.mpg.de/projects/cdo/embedded/index.html#x1-4460002.8.16) can process your data without manually chopping your files up. I am trying to wean myself from using unneeded intermediate files, when possible.

Hi Brendan,

I appreciate your quick answer.

Why keep separate files for seasons/months when cdo has seasonal/monthly operators? Why not mergetime all your files into one large file?

That's a valid point. The reason I aimed for multiple files has to do with the fact that I am conducting most of my data analysis with Python. I am using CDO for some pre-processing steps like re-structuring the files and clipping my region of interest (sellonlatbox). I have some scripts available that require the file structure to be as I envisioned it when I wrote my earlier post.

I agree with you however, that there might be easier ways to go about this. Another good option for me would be to create 4 files for the whole time series, one for each meteorological season.

If I want to achieve this however, my earlier question still stands. I am unsure how to use CDO in a for-loop to obtain these 4 files.

(...) the trick to getting meteorological seasons is shifttime (https://code.mpimet.mpg.de/projects/cdo/embedded/index.html#x1-2410002.6.4). shifttime back one month before any seasonal calculations(i.e. seastat https://code.mpimet.mpg.de/projects/cdo/embedded/index.html#x1-5070002.8.27). Then shifttime forward one month afterwards to correct the date.

Thank you for this suggestion, as this will be useful if I need to do any other pre-processing steps with CDO that requires this changed season definition.

I am re-reading the selseason docs (https://code.mpimet.mpg.de/projects/cdo/embedded/index.html#x1-1670002.3.4) and although you can specify DJF it does not help with grouping by year; shifttime is still needed.
Okay. Well, it is still easier to start with one big file than to code a mix and match.
Create one big file as shown above:

cdo mergetime precip_WJB_int.????_???.BN_1958_2013_1km.DD.nc precip_WJB_int.BN_1958_2013_1km.DD.nc

Then you need a bash script to loop through seasons and years using the selseason and selyear. I will use bash's associative arrays.

declare -A season_map
cdo_seasons='JFM AMJ JAS OND'
season_map=( [JFM]=DJF [AMJ]=MAM [JAS]=JJA [OND]=SON )
for year in {1958..2013} ; do
    for season in $cdo_seasons ; do
        cdo shifttime,1month -selyear,"$year" -selseason,"$season" -shifttime,-1month precip_WJB_int.BN_1958_2013_1km.DD.nc precip_WJB_int."$year"_${season_map[$season]}.BN_1958_2013_1km.DD.nc
    done
done

Since you know python and not bash, you could also write this in python using cdo's python bindings (https://code.mpimet.mpg.de/projects/cdo/wiki/Cdo%7Brbpy%7D). I have yet to try these. If you do, make sure you read https://github.com/Try2Code/cdo-bindings/blob/master/python/test/test_cdo.py as it shows both the old and new way to chain cdo operators.