« NSF Unidata Staff at... | Main | NSF Unidata Update:... »

30 December 2024

Version 5.3.0 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 5.3.0 significantly improves the unpacking of ELM/CLM/CTSM history and restart datasets, and seems complete in this regard. ncrcat has more friendly behavior when appending, and ncclimo now outputs more metrics (such as valid area per gridcell) when in regional statistics mode. This release many be skipped if these changes do not interest you.

New Features

Once again, ncks leveled-up how it unpacks ELM/CLM history/restart files. New features include automatic writing of fractional landunit and fractional column variables frc_landunit and frc_column, respectivly. frc_landunit is equivalent to the variable PCT_LANDUNIT sometimes placed in history files, but is expressed as a decimal fraction (not a percent) and is derived from restart files. The landunit dimension of frc_landunit is of size ten: Index zero contains the sum (should be 1.0) of standard landunit types/indexes 1-9 soil, crop, ..., urban medium density). frc_column is a bespoke version of the same, only for column (not landunit) types. For datasets with Multiple Elevation Classes (MECs): index = 0 is soil column, index = 1 is MEC == 1, indexes 2..10 are remaining MEC columns, index 11 is sub-total of MEC columns, index = 12 is deep lake column, index = 13 is wetland column, and index = 14 is grand total of natural columns. For non-MEC datasets, index = 0 is soil column, index = 1 is glaciated column, index = 2 is deep lake column, index = 3 is wetland column, and index = 4 is grand total of natural columns.
```
ncks --s1d --hrz=history.nc restart.nc s1d_r05.nc # Restart file
ncks -m -v frc.? s1d_r05.nc
netcdf s1d_r05 {
    dimensions:
      column = 15 ;
      landunit = 10 ;
      lat = 360 ;
      lon = 720 ;

    variables:
      float frc_column(column,lat,lon) ;
      float frc_landunit(landunit,lat,lon) ;
}
```
http://nco.sf.net/nco.html#s1d
http://nco.sf.net/nco.html#unpack_sparse
ncks can now grid snow-related variables into a top-down (ocean-like) vertical grid that many think is more intuitive. By default ELM/CLM/CTSM store the negative of the number of active snow layers in the variable SNLSNO. Restart files for these models store the active snow layer butted-up against the lowest layers in the level dimension (so that they are continguous with soil layers to simplify hydrologic calculations). This makes snow variables in restart files hard to visualize. By default S1D now uses SNLSNO, if present, to unpack snow variables in a first layer downwards order, increasing with depth. Inactive layers at the bottom (i.e., where they reside physically). The resulting snow variables appear like ocean state variables over uneven bathymetry, with missing value underneath. We call this "snow-ocean" ordering to contrast it with the on-disk storage order of snow variables.
http://nco.sf.net/nco.html#s1d
http://nco.sf.net/nco.html#snw_ocn
http://nco.sf.net/nco.html#snw_rdr
http://nco.sf.net/nco.html#no_snw_ocn

ncks S1D now accepts the "--rgr lut_out=$lut_out" option to specify which column type(s) to place in each gridcell in the output file. The argument lut_out is the standard landunit type of the column, with (thus far) two additional values that will output area-weighted averages of multiple landunit types:

lut_out Output will be value of column(s) in this Landunit
1       Vegetated or bare soil
2       Crop
3       Landice (plain, no MEC)
4       Landice multiple elevation classes
5       Deep lake
6       Wetland
7       Urban tall building district
8       Urban high density
9       Urban medium density
10      Area-weighted average of all landunit types except MEC glaciers
13      Area-weighted average of soil+(non-MEC) glacier

For example,

ncks --s1d --rgr lut_out=1 --hrz=hst.nc rst.nc s1d.nc # Output Soil LUT ncks --s1d --rgr
lut_out=13 --hrz=hst.nc rst.nc s1d.nc # Avg Soil+Glacier

http://nco.sf.net/nco.html#s1d
http://nco.sf.net/nco.html#lut_out

ncremap always warns when it regrids land model files that contain a "landfrac" variable and SGS renormalization is not requested. This is because ELM/CLM/CTSM use "landfrac" for sub-gridscale fraction, and ignoring that causes errors along coastlines. EAMxx also archives that same "landfrac" but every EAMxx variable is valid throughout every gridcell so renormalization would be superfluous at best. Now ncremap no longer emits the "landfrac" WARNING for EAMxx files.
http://nco.sf.net/nco.html#eamxx
As of NCO version 5.3.0 (December, 2025) ncclimo automatically outputs additional metrics with global statistics. The output files containing the global timeseries also contain the variable valid_area_per_gridcell. This field is equivalent to the product of the area variable and the sgs_frc variable (if any). Thus for ELM/CLM/CTSM, this field equals area times landfrac, while for EAM/CAM this variable simply equals area. The output files also contain the area and sgs_frc variables separately. The presence of these variables in output allows downstream processors (e.g., zppy) to generate additional masks and weights for rescaling the statistics. For example, these fields can be used to rescale global sums into any units desired.
```
ncclimo -P elm --split --rgn_stt=sum -c v3.LR.piControl -s 460 -e 461 \
--drc_in=drc_in --drc_out=drc_out
```
http://nco.sf.net/nco.html#glb_stt
http://nco.sf.net/nco.html#rgn_stt