The netCDF Operators NCO version 4.7.7 are ALIVE!
http://nco.sf.net (Homepage, Mailing lists, Help)
http://github.com/nco (Source Code, Issues, Releases, Developers)
What's new? Version 4.7.7 has modest new features everywhere:
Improved forward and backward compatibility with newer and older
versions of ESMF_RegridWeightGen; Generation and inferral of grids
running north->south (à la ECMWF); exact symmetry for Gaussian grid
interfaces; an ncpdq compression map that converts doubles to floats;
production of "skeleton files" for new grids; equivalent treatment of
MPAS and CESM grid variables in arithmetic; and reversion of the
filename whitelist security.
Work on NCO 5.0.0 has commenced "under the hood". The key leap in that
release will be support for netCDF4 user-defined types. Printing of
netCDF4 user-defined types ENUM and VLEN is ready now (though
unsupported) with the --udt flag. 5.0.0 will contain the finished
version of that, and progress on native weight generation by ncremap.
Enjoy,
Charlie
NEW FEATURES (full details always in ChangeLog):
A. ncpdq has a new packing map: dbl_flt
The map converts to double-precision to single-precision,
so it is technically a precision change, not packing.
It is in ncpdq because it reduces space by ~50%, like packing.
Like packing the precision loss is irreversible.
Unlike packing, no attributes are created, modified, or deleted.
ncpdq -M dbl_flt in.nc out.nc
http://nco.sf.net/nco.html#dbl_flt
http://nco.sf.net/nco.html#ncpdq
B. Filename sanitization via character whitelists, first introduced in
NCO 4.7.3 to increase security against malicious filenames, has
been turned-off, possibly permanently. Points passionately raised
by a user led us to turn-off this feature until lack of it is shown
to lead to a real-world exploit.
https://github.com/nco/nco/issues/104
http://nco.sf.net/nco.html#wht_lst
C. The CCM/CAM/EAM family of atmospheric models does not output a
"bounds" variable or attribute corresponding to its "lev"
coordinate. This prevents NCO from activating some desirable CF
machinery. To workaround this, NCO now outputs the ilev coordinate
(and hyai, hybi) whenever the lev coordinate is also output.
ncks -v T in.nc out.nc
http://nco.sf.net/nco.html#bnd
D. Equivalent treatment of MPAS and CESM grid variables.
NCO does not arithmetically alter grid-related variables.
Who wants to subtract latitude coordinates from eachother?
Previously it only skipped arithmetic on CESM'ish grid variables.
Now NCO has an extensive (pun intended!) list of MPAS and other
grid variables that it will not subtract. For the most part,
ncremap will not regrid these variables either.
http://nco.sf.net/nco.html#cnv_CCSM
http://nco.sf.net/nco.html#cnv_MPAS
http://nco.sf.net/nco.html#ncremap
E. I greatly expanded the field manual to regridding with ncremap.
The documentation is currently available to E3SM users at
https://acme-climate.atlassian.net/wiki/spaces/SIM/pages/754286611/Regridding+E3SM+Data+with+ncremap
and to others via email request (it will be in Users Guide soon).
F. ncremap adds --ignore_degenerate to default ESMF_RegridWeightGen
(ERWG) options when ERWG version >= 7.0. This is done to preserve
backwards compatibility since ERWG 7.1.0r and later require
--ignore_degenerate to successfully regrid some datasets (e.g.,
CICE) that previous ERWG versions handle fine.
ncremap -m map.nc in.nc out.nc
http://nco.sf.net/nco.html#wgt_opt
G. Not only do the British drive on the wrong side of the road, they
prefer North-to-South (n2s) grids over South-to-North (s2n) grids.
NCO now implements a latitude direction keyword lat_drc.
Grid generation implements n2s ordering via --rgr lat_drc=n2s.
Grid diagnosis now recognizes n2s ordering for all rectangular grids.
http://nco.sf.net/nco.html#lat_drc
H. ncremap now generates exactly symmetric interface latitudes for
Gaussian grids. To our knowledge only NCO generates accurate
Gaussian interfaces (via a Newton-Raphson iteration of boundaries
to enclose the area indicated by the Gaussian weight). However,
previously the interface latitudes could accumulate rounding errors
during Newton-Raphson iteration differ between the hemispheres by
~1.0e-14. Now symmetry properties are used to ensure exact symmetry.
http://nco.sf.net/nco.html#gss
I. ncremap now produces a "skeleton file" of grids it generates
when requested with --skl_fl.
ncremap -G latlon=2560,5136#lat_typ=gss#lon_typ=grn_ctr#lat_drc=n2s
-g grd_ecmwf.nc --skl=skl_ecmwf.nc
http://nco.sf.net/nco.html#skl_fl
http://nco.sf.net/nco.html#skeleton
BUG FIXES:
A. ncks -r could incorrectly report that CDF5 was enabled.
This occurred when linked to netCDF 4.5.x-4.6.1.
This has been fixed.
B. ncremap previously broke on weight files that lacked a num_wgts
dimension, even if that dimension was unused (orphaned).
Dependence on num_wgts has been eliminated and fixes this.
This also enables compressed maps to work correctly.
C. Fixed a bug in the convergence test to detect Gaussian grids
Full release statement at http://nco.sf.net/ANNOUNCE
--
Charlie Zender, Earth System Sci. & Computer Sci.
University of California, Irvine 949-891-2429 )'(
