Posts/NCO tools

Home Publications WorkExp Projects News Posts

Adition and substraction

when variable names are different

  1. rename the variable names in the original file

ncrename -v(variable) original_name,new_name in.nc(file_nc)

ncrename -v gpp,nep can_gpp.nc
ncrename -v ra,nep can_ra.nc
ncrename -v ra,nep can_rh.nc
  1. substract/add using any of the following section’s methods, NEP = GPP - Ra -Rh 
    ncdiff can_gpp.nc can_ra.nc can_gpp-ra.nc
ncdiff can_gpp-ra.nc can_rh.nc can_nep.nc

Note: variable names must be same

for substraction

op_typ key values: ‘sbt’, ‘-’, ‘dff’, ‘diff’, ‘sub’, ‘subtract’, ‘subtraction’

file3=file1-file2
ncdiff 1.nc 2.nc 3.nc
ncbo --op_typ=- 1.nc 2.nc 3.nc

for addition

op_typ key values: ‘add’, ‘+’, ‘addition’

file3=file1+file2
ncadd 1.nc 2.nc 3.nc
ncbo --op_typ=+ 1.nc 2.nc 3.nc

for multiplication

op_typ key values: ‘mlt’, ‘*’, ‘mult’, ‘multiply’, ‘multiplication’

file3=file1*file2
ncmult 1.nc 2.nc 3.nc
ncbo --op_typ=* 1.nc 2.nc 3.nc

for division

op_typ key values: ‘dvd’, ‘/’, ‘divide’, ‘division’

file3=file1/file2
ncdivide 1.nc 2.nc 3.nc
ncbo --op_typ=/ 1.nc 2.nc 3.nc

In case you want to compute the flux density of gpp, i.e. GPP (in gC mon-1) divide area (m2) = gpp (gC m-2 month-1). if GPP is 3-d file and area is 2-d, ncbo commands work well provided the variable name of the area file is renames to gpp.

ncbo --op_typ=/ CESM2_ssp585_r1i1p1f1_gpp_gCm-2_64x128.nc AREA64.nc gpp_flux_density.nc

Averaging an ensemble of realizations:

nces 85_01.nc 85_02.nc 85_03.nc 85_04.nc 85_05.nc 85.nc nces 85_0[1-5].nc 85.nc

Ensemble/Variable averaging:

ncea bnu_gpp_regrid.nc can_gpp_regrid.nc ccsm_gpp_regrid.nc mod_av_gpp_regrid.nc —–

Renaming the attribute(long_name) of a variable “CVEG” to “LEAFC+WOODC”

http://nco.sourceforge.net/nco.html#ncatted
ncatted -a long_name,CVEG,o,c,LEAFC+WOODC cesm1bgc_pftcon_leafc.nc

Adding an attribute

http://dvalts.io/2018/05/14/More-NCO-fun.html
ncatted -a standard_name,col_fire_closs,a,c,"col_fire_closs anomalies" cesm1bgc_pftcon_col_fire_closs_anomalies.nc

Deleting an attribute

ncatted -a stardard_name,col_fire_closs,d,, cesm1bgc_pftcon_col_fire_closs_anomalies.nc

Filling NaN in place of some _FillingValue or vice versa

http://nco.sourceforge.net/nco.html#nan

  • from _FillingValue to NaN
    ncatted -a _FillValue,,o,f,NaN bnu_nep_regrid.nc
  • from Nan to _filling value
    ncatted -a _FillValue,,m,f,1.0e36 in.nc
  • e.g.
    cp gpp_ccsm_hist test.nc #test file has gpp variable
  • setting the fill value to 1e33
    ncatted -a _FillValue,gpp,o,f,1.e33 test.nc
    –info -a:attribute; o: overwrite;f:float \
  • setting the missing value to 1e33
    ncatted -a missing_value,gpp,o,f,1.e33 test.nc

Concatenation

for concatenating the historical and the rcp simulations i used nco concatenation tool called ncrcat
eg.:
ncrcat ccsm_gpp_hist_regrid.nc ccsm_gpp_rcp_regrid.nc ccsm_gpp_regrid.nc
the concatenation result of there two files are stored in the rightmost file

Concatenate nc files over lats

http://forum.marine.copernicus.eu/discussion/205/how-to-mergeconcatenate-cmems-netcdf-files-tips/p1

  • Adding a record dimension only to the first file
    • ncks -O --mk_rec_dmn recorddimension in.nc out.nc
    • mv decomp_trend_ccsm_gpp_000.nc rd_decomp_trend_ccsm_gpp_000.nc
    • ncks -O -h --mk_rec_dmn lat rd_decomp_trend_ccsm_gpp_000.nc rd_decomp_trend_ccsm_gpp_000.nc
  • Conacatenating files
    • ncrcat -h outfrom1.nc file_1.nc file_2.nc file_outfrom1_1_2.nc
      (use [-h] if you dont want to have histroy file)
    • ncrcat -h rd_decomp_trend_ccsm_gpp_000.nc decomp_trend_ccsm_gpp_*.nc decomp_trend_ccsm_gpp.nc
  • Fixing the record dimension
    • ncks -O -h --fix_rec_dmn lat decomp_seasonal_ccsm_nep.nc decomp_seasonal_ccsm_nep.nc
  • Check your file

    • ncdump -h decomp_trend_ccsm_gpp.nc |less

      Masking the values of the original file using the land fraction nc file

ncks -A sftlf.nc gpp.nc
it will copy the variable from sftlf.nc to gpp.nc; so gpp.nc will have variable(s) from sftlf.nc
\where the land frac is 0 or less set the corresponding gpp variable to missing value ncap2 -s "where(sftlf<=0) gpp=gpp.get_miss();" test.nc testfixfill.nc

Extracting Variable(s)

Splitting nc files

ncks -d time,0,729 hgt.2006.nc hgt.2006-01.nc
ncks -d time,730,1459 hgt.2006.nc hgt.2006-02.nc

e.g.

ncks -d time,1560,1919 bnu_gpp_regrid_sel.nc bnu_gpp_regrid_sel_time.nc
ncks -d lat,0,179 bnu_gpp_regrid_sel_time.nc bnu_gpp_regrid_sel_time_l1.nc
ncks -d time,12,1151 b.e10.BRCP85BDRD.pftcon.f09_g16.001.clm2.h0.SMOIST.200501-210012.nc rcp.nc

ncks is smart to understand time slices slicing time period 2001-01-01 to 2013-12-31 e.g.

ncks -d time,2001-01-01,2013-12-31 CESM2_ssp585_r1i1p1f1_gpp_anomalies_gC.nc CESM2_ssp585_r1i1p1f1_gpp_anomalies_gC_2001_2013.nc

Change the type of the variable

ncap2 -s 'var2=double(var1)' in.nc out.nc

Computing the temporal standard deviation

Calculate the temporal standard deviation See the NCO manual on this but, in short, the commands are:

#Calculate the time mean of variable_name.
ncwa -O -v tas -a time CESM2_ssp585_r1i1p1f1_anomalies_tas_2001_2013.nc out.nc
#Calculate the deviations with respect to the mean.
ncbo -O -v tas CESM2_ssp585_r1i1p1f1_anomalies_tas_2001_2013.nc out.nc out.nc
#Sum the square of the deviations, divide by (N-1), and take the square root
ncra -O -y rmssdn out.nc out.nc
mv out.nc CESM2_ssp585_r1i1p1f1_anomalies_tas_2001_2013_std.nc

sum over a time dimension

To sum over a dimension, for example to sum a file of daily precipitation to obtain an annual total

ncra -h -O -y ttl in.nc out.nc

To get rid of the time dimension with the length of 1 month

Best way: ncwa automatically removes the averaged dimensions from the coordinates attribute ncwa -O -v tas -a time CESM2_ssp585_r1i1p1f1_anomalies_tas_2001_2013_std2.nc out.nc