[netcdfgroup] netCDF Operators NCO version 4.9.6 have beamed down to the surface

• To: netcdfgroup@xxxxxxxxxxxxxxxx, NCO Announce Mail List <nco-announce@xxxxxxxxxxxxxxxxxxxxx>
• Subject: [netcdfgroup] netCDF Operators NCO version 4.9.6 have beamed down to the surface
• From: Charlie Zender <zender@xxxxxxx>
• Date: Mon, 30 Nov 2020 17:04:10 -0800

The netCDF Operators NCO version 4.9.6 have arrived.

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

What's new?
Version 4.9.6 is a minor NCO release that contains a few new features
and fixes. Unless the features/fixes listed below affect you, no need
to upgrade this time.

Work on NCO 4.9.7 has commenced and will improve analysis of land
surface datasets packed into sparse-1D formats.

Enjoy,
Charlie

NEW FEATURES (full details always in ChangeLog):

A. ncclimo supports a new option --var_xtr to specify "extra
variables" for the splitter to add to every generated timerseries.
The splitter attempts to keep the generated timeseries as lean as
possible and by default adds only the most useful ancillary variables,
such as grid coordinates and CF-associated variable, to split
timeseries. Previously, there was no way to add arbitrary variables,
e.g., the CO2 volume mixing ratio, to a split timeseries. Now users
can specify an arbitrary number of variables to includes in every
split timeseries, e.g.,

ncclimo --var=T,Q --var_xtr=co2vmr,FSDS ...

In this example the splitter produces two output files because there
are two primary variables (T,Q). Each timeseries also contain co2vmr
and FSDS. Inclusion of other variables, e.g., area, depends on context
and on other options
http://nco.sf.net/nco.html#var_xtr

B. This version is forward-compatible with netCDF 4.8.0, which we know
will contain some backward-incompatibile changes to filter behavior.
Once netCDF 4.8.0 is released, if it becomes the default version of
netCDF on your syste, then you will likely need to upgrade to NCO
4.9.6 as previous versions may be incompatible with netCDF 4.8.0.

C. ncclimo now supports the -P (or --prc_typ) option instead of -m
(or --mdl_nm) to determine the processing type. More importantly,
ncclimo the -P option to ncremap internally, which simplifies some
complex climo/regridding commands. For example, now it is possible to
create regridded climatologies of output that requires sub-gridscale
regridding with a single ncclimo command whereas previouly this
required separate ncclimo and ncremap commands:

ls *elm.h0* | ncclimo -P elm -s 1 -e 500 --map=map.nc
ls *mpaso* | ncclimo -P mpaso -s 1 -e 500 --map=map.nc
ls *mpassi* | ncclimo -P mpassi -s 1 -e 500 --map=map.nc

The ncclimo -m mdl_nm option is still supported, and is required in
some situations, e.g., when the processing type is 'elm' yet the
history filenames use 'clm2' not 'elm'. In such mixed cases both -P
and -m options may be necessary.
http://nco.sf.net/nco.html#ncclimo
http://dust.ess.uci.edu/smn/smn_nco_e3sm_202010.pptx

D. ncclimo and ncremap now support the large memory AMD (cmem) nodes
on Cori when in MPI mode.


BUG FIXES:

A. ncremap fixes a bug that prevented MPI mode from working on Cori.

B. ncremap always retains (rather than potentially altering) the
input values of masks specified in SCRIP gridfiles when it writes
the output map-file.

C. ncremap correctly infers the final vertices of RLL grids stored
in north-to-south order. Previously, the final vertex could be
misdiagnosed for some counter-clockwise orderings of cells.

Full release statement at http://nco.sf.net/ANNOUNCE

--
Charlie Zender, Earth System Sci. & Computer Sci.
University of California, Irvine 949-891-2429 )'(