« Abstract Deadlines... | Main | NetCDF operators... »

Version 5.3.4 of the netCDF Operators (NCO) has been released. NCO is an Open Source package that consists of a dozen standalone, command-line programs that take netCDF files as input, then operate (e.g., derive new data, average, print, hyperslab, manipulate metadata) and output the results to screen or files in text, binary, or netCDF formats.

The NCO project is coordinated by Professor Charlie Zender of the Department of Earth System Science, University of California, Irvine. More information about the project, along with binary and source downloads, are available on the SourceForge project page.

From the release message:

Version 5.3.4 contains of bevy of minor features that improve NCO in many different areas. These range from reduced monotonicity WARNINGs from ncrcat, to an easy new method for all operators to remove global attributes, to bulletproofing ncclimo to work as intended with instantaneous input, to new urban landunit diagnostics in ncks. Skip this release if these changes do not interest you.

New Features

1. All operators can now delete user-specified global attributes. The option --gad=att1,att2,...,attN (or long-option equivalent --glb_att_del=att1,att2,...,attN) deletes (rather than adds, like its sister option --gaa) global attributes in an output file. Global attribute deletion requires only the name of the global attribute without any further information, so the argument to the option is simply a comma-separated list of all attributes to delete. No information about the attributes' types, sizes, or values need be given. The option works on attributes of all types, sizes, and values.

   ncks --gad=history_of_appended_files,nco_openmp_thread_number,\
         input_file,map_file,remap_version,remap_hostname,\
         remap_command,remap_script,NCO in.nc out.nc
   

   The global attribution deletion feature helps to avoid the performance penalty incurred by using @command{ncatted} separately to annotate large files. http://nco.sf.net/nco.html#gad
   http://nco.sf.net/nco.html#gaa
2. ncrcat no longer performs monotonicity checks on CF bounds variables (which are never monotonic). This eliminates a lot of noisy monotonicity WARNINGs from many datasets.
   http://nco.sf.net/nco.html#ncrcat
3. ncks --is_var var_nm is a new option that checks whether the single variable var_nm is a "horizontal variable". In this context, horizontal variables means single-level variables that have horizontal dimensions (e.g., lat and lon) yet no vertical dimension and no other dimensions except possibly the temporal dimension (time). ncks prints "Yes" or "No" and then exits. This capability helps ncclimo determine whether a variable is suitable for creating a regional average timeseries from. Due to this context, coordinate variables (including latitude and longitude) are not identified as horiztonal variables. Furthermore, variables identified as horizontal may only have two or three dimensions, and those dimensions must be in this list: lat, lon, ncol, nCells, time, and Time.

   zender@spectral:~$ ncks --is_hrz three_dmn_rec_var ~/nco/data/in.nc
   Yes
   zender@spectral:~$ ncks --is_hrz one ~/nco/data/in.nc
   No
   zender@spectral:~$
   

   The E3SM-centric list of allowed dimension names is only a starting point! Please contact Charlie if you would like this list expanded.
   http://nco.sf.net/nco.html#is_hrz
4. ncclimo can now infer the timesteps per day (tpd) to use for high-frequency instantaneous (in the CF sense) data that lack a CF temporal bounds variable. ncclimo also uses tpd to help it discover how many days-per-file there are. Nevertheless, the metadata in the ncclimo output may be more accurate if the input data contains a temporal bounds variable.
5. The ncks S1D functionality that "gridifies" ELM/CTSM data has new features to aggregate and report Urban landunit characteristics. First, the new S1D option "--rgr lut_out=789" stores the area-weighted mean of the three Urban landunit types (7, 8, and 9) in the output for every variable in a restart file that is defined on columns.

   ncks --s1d --rgr lut_out=789 --hrz=hst.nc rst.nc s1d.nc # Avg Urban

   Second, the frc_column diagnostic variable stores the subtotal of the three Urban landunits areas in (C-based) index 15 of output files with MECs, and in index 5 of output files without MECs. Users can now easily assess the total fraction of a gridcell that is Urban.
   http://nco.sf.net/nco.html#s1d
   http://nco.sf.net/nco.html#frc_column
6. ncks now inserts line-breaks in NC_STRING-valued text when it encounters a C-format carriage return "\n". Now printing text stored as NC_STRING and as NC_CHAR has the same appearance. This makes it easier to read "history" global attributes that are stored as NC_STRING:

   ncap2 -O -4 -h -s 'global@history="Previous history"s' ~/foo.nc
   ncap2 -O -s 'one=1' ~/foo.nc ~/foo.nc
   ncks -M ~/foo.nc # NC_STRING history is now more legible
   

   http://nco.sf.net/nco.html#ncks
   http://nco.sf.net/nco.html#history
7. ncremap has deprecated the use of the "--alg_typ=tempest" regridding algorithm value. This algorithm name simply meant that NCO invoked the TempestRemap (TR) regridder with all default choices, i.e., with no options. However, the default TR algorithm is rarely employed anymore, and many users misinterpreted the meaning of "tempest" . Instead of supplying ncremap with their own options, they received the default, which usually was sub-optimal. We encourage users to migrate to the appropriate and specifically-named TR algorithms such as "fv2fv_flx", "traave", "trfv2", etc.
   http://nco.sf.net/nco.html#alg_typ

Additional details are available in the ChangeLog.