[netcdfgroup] netCDF Operators NCO version 5.3.3 split the uprights

• To: NCO Announce Mail List <nco-announce@xxxxxxxxxxxxxxxxxxxxx>, netCDF Mail List <netcdfgroup@xxxxxxxxxxxxxxxx>
• Subject: [netcdfgroup] netCDF Operators NCO version 5.3.3 split the uprights
• From: Charlie Zender <zender@xxxxxxx>
• Date: Fri, 28 Mar 2025 18:18:06 -0700

http://nco.sf.net (Homepage, Mailing lists, Help)
http://github.com/nco/nco (Source Code, Issues, Releases)

What's new?
Version 5.3.3 adds IEEE floating-point exception testing to ncks.
All operators include corner-case quantization fixes for strange
numbers (-0.0, NaN) that are also now in the netCDF-C Library.
ncclimo now only tries to create regional timeseries statistics
on variables that are single-level horizontal lat/lon/ncol fields.
ncremap supports interpolation of pure-pressure to hybrid
sigma-pressure coordinates with distinct source and destination
surface pressure fields. And ncrename adds some words of caution.
Skip this release if these changes do not interest you.

Did you realize that the first public release of ncks was May 2, 1995?
https://www.unidata.ucar.edu/mailing_lists/archives/netcdfgroup/1995/msg00069.html
Happy 30th birthday ncks!!!

Enjoy,
Charlie

NEW FEATURES (full details always in ChangeLog):

A. The new ncks --fpe option reports the status of all IEEE Floating
Point Exceptions (FPEs) after execution. This can be useful to verify
whether input datasets contain strange numbers that could be
arithmetically processed, and to ensure the integrity of NCO
arithmetic.

% ncks --fpe -O -7 -L 1 -C -v ppc_zro_ngt_nan_flt --qnt_alg=gbr \
       --qnt default=3 --cmp='shf|zst' in.nc out.nc
ncks: Successfully set FE_INVALID|FE_DIVBYZERO|FE_OVERFLOW exception flags
ncks: FE_DIVBYZERO is not raised
ncks: FE_INEXACT is raised
ncks: FE_INVALID is not raised
ncks: FE_OVERFLOW is not raised
ncks: FE_UNDERFLOW is not raised

http://nco.sf.net/nco.html#fpe
http://nco.sf.net/nco.html#ncks

B. All operators can now quantize -0.0 (a legal number) without
triggering any IEEE Floating Point Exceptions (FPEs). Other weird
numbers, like NaN, already avoided FPEs. Moreover, the necessary
changes to avoid FPEs have been merged upstream into the quantization
algorithms in libnetcdf version 4.10-development. Thanks to Jeff
Whitaker (NOAA) for bringing this to our attention, to M. Garcia
for the original bug report, and to Ward Fisher for merging.
http://nco.sf.net/nco.html#qnt

C. When no explicit variable extraction list is specified in
timeseries mode, ncclimo automatically creates timeseries for all
input variables with at least two dimension. However, when regional
statistics are also requested with --rgn_stt, these statistics are
only meaningful for spatial timeseries, i.e., single level fields with
latitude, longitude, and time. Under these circumstances ncclimo now
only creates regional statistics for single level spatial timeseries.
Previously it would try (and fail) to generate regional statistics
for multi-level or otherwise inappropriate rank >=2 variables.

ls caseid*.nc | ncclimo --split --rgn_avg -s 2013 -e 2014 -o foo

http://nco.sf.net/nco.html#ncclimo

D. Previous versions of NCO could only interpolate ECMWF/IFS
data to hybrid sigma-pressure coordinates when the input
surface pressure (lnsp) was for the output as well. Now NCO
can interpolate to a distinct output surface pressure that
is storeed in the vertical template file (a so-called "fat"
template that includes the surface pressure, rather than a
"skinny" template that does not).

ncremap --ps_nm=lnsp --vrt_fl=vrt.nc in.nc out.nc

http://nco.sf.net/nco.html#vrt
http://nco.sf.net/nco.html#ncremap

E. ncks --chk_map now prints the minimum and maximum angular
and horizontal resolution of the source and destination grids.
The angular resolution is estimated as the square-root of the area
(in sr) of the smallest/largest gridcells, and is expressed in
degrees. The spatial resolution is estimated
This is the angle subtended by a spherical cap with the same area
as the gridcell.

% ncks --chk_map map_ne30pg2_to_cmip6_180x360_ncoaave.20231201.nc
...

...

...

This feature can be useful in assessing the uniformity of grids, and
http://nco.sf.net/nco.html#chk_map
http://nco.sf.net/nco.html#ncks

F. ncrename now WARNs users attempting to rename coordinates in
netCDF4 files. The WARNING, which prints only when debugging level
> 0, the input file is in netCDF4 format, and the user tries to
rename a variable or dimension, is self-explanatory:

zender@spectral:~$ ncrename -D 1 -O -v lat,latitude in.nc out.nc

ncrename HINT: Do not expect this issue to be fixed in the near future :)

zender@spectral:~$ ncrename -O -v lat,latitude in.nc out.nc
zender@spectral:~$ ncrename -D 1 -O -a lat@units,meaning in.nc out.nc

zender@spectral:~$ ncrename -D 1 -O -v lat,latitude in3.nc out.nc
zender@spectral:~$

The second and third and fourth examples show no WARNING output
because the debugging level is the default (0) in the second, and
because the user is not renaming a variable or dimension in the
third, and because the input file is netCDF3-format in the fourth.
The decision was made to turn-on this WARNING only in a non-default
debug level to avoid annoying users with existing workflows that
expect no text output from ncrename. This compromise is intended
to prevent widespread panic and mass hysteria. Thanks to Toby Searle
(UKMO) for prompting this change.

BUG FIXES:

A. Previous versions of NCO could omit storing the attributes
for the frc_column and frc_landunit variables created by the S1D
feature. This is fixed. There is no workaround. The solution is
to upgrade
http://nco.sf.net/nco.html#s1d

Full release statement at http://nco.sf.net/ANNOUNCE
--
Charlie Zender, Dept. of Earth System Science
University of California, Irvine 949-891-2429