The netCDF Operators NCO version 4.9.8 have arrived.
http://nco.sf.net (Homepage, Mailing lists, Help)
http://github.com/nco (Source Code, Issues, Releases, Developers)
What's new?
Version 4.9.8 contains a few cool features.
These include support for unpacking sparse 1D (S1D) data and restart
files used for plant functional types and multiple elevation classes,
splitting monthly resolution timeseries that do not begin/end in
Jan/Dec, seamless climos for the E3SM ice-sheet model (MALI), and
corner-case bugfixes for inferring grid files and handling
sub-gridscale data.
Work on NCO 4.9.9 has commenced and will continue improving support
for analysis of land surface datasets packed into sparse-1D formats,
splitting daily resolution datasets along monthly boundaries,
and begin ncremap support for the MOAB regridding package.
Enjoy,
Charlie
NEW FEATURES (full details always in ChangeLog):
A. Windows builds should now contain the nces.exe and ncrcat.exe
executables. Previously, Windows users had to run nces and ncrcat
commands indirectly, using ncra.exe -Y nces ..., or manually copy
ncra.exe to nces.exe, etc. The executables are identical, and
behave differently depending on their invocation name.
ncrcat ... # These all do the same thing
ncrcat.exe ... # "
ncra.exe -Y ncrcat ... # "
http://nco.sf.net/nco.html#ncrcat
http://nco.sf.net/nco.html#nces
B. ncclimo supports new options --mth_srt and --mth_end in splitter
mode. The arguments to these options specify the (1-based) month in
which the requested timeseries will begin and end, respectively,
and these default to 1 (January) and 12 (December). To extract
14-month timeseries from individual monthly input files one would
use, e.g.,
ncclimo --yr_srt=1 --yr_end=2 --mth_srt=4 --mth_end=5 ...
http://nco.sf.net/nco.html#ncclimo
http://nco.sf.net/nco.html#mth_srt
http://nco.sf.net/nco.html#mth_end
C. ncclimo/ncremap support analysis of MALI land ice model output
when invoked with the new '-P mali' or '--prc_typ=mali' options.
ncclimo -P mali -s 1 -e 2 -o climo mali.hist*.nc
ncremap -P mali --map=map.nc in.nc out.nc
http://nco.sf.net/nco.html#prc_typ
http://nco.sf.net/nco.html#ncclimo
D. ncks now supports unpacking sparse 1D (S1D) ELM/CLM data and
restart files for plant functional types (PFTs) and multiple
elevation classes (MECs).
ncks --s1d -v cols1d_topoglc --hrz=elm_mali_ig_history.nc \
elm_mali_restart.nc out.nc
ncks --s1d -v GPP,pfts1d_wtgcell elm_crop_history.nc out.nc
After such S1D datasets have been "unpacked", they may easily
be regridded with, e.g., ncremap.
This feature is under continued development and user-feedback
would be appreciated. Documentation does not yet exist.
http://nco.sf.net/nco.html#ncks
E. ncks now prints more informative messages when users attempt
to vertically interpolate variables (such as EAM/CAM US, VS) in
hybrid coordinate data files that contain multiple grids (e.g.,
the normal and staggered grids) though only one surface pressure
variable.
http://nco.sf.net/nco.html#ncremap
Thanks to Walter Hannah for noticing this.
F. As of version 4.9.8 the default ncremap behavior is to omit the
staggered grid from output files on an cap grid.
The new flag --stg_grd turns-on outputting the staggered
grid, and thus recovers the previous default behavior.
BUG FIXES:
A. ncap2's array() function was rewritten to use a starting value
plus an index times an increment, rather than adding increments, in
order to reduce rounding errors. Thanks to Henry Butowsky.
B. The regridder in ncremap no longer attempts to access the
(non-existent) tally array in SGS-weighted fields without missing
values. The previous behavior would cause a crash. Fortunately
this combination does not often occur in practice. Upgrading to
4.9.8 is the only solution, there is no workaround.
C. The capability to infer MPAS grids introduced in 4.9.8 introduced
a bug that requires that horizontal coordinates have a units
attribute. Many datafiles strongly disagree with this presumption!
The requirement has been removed. The solution is to upgrade to
4.9.8, there is no workaround.
D. The ncclimo --xcl_var flag has been fixed. Previously the
flag had no effect. Now a variable extraction list may be turned
into a variable exclusion by using --xcl_var --vars=v1,v2,...vn.
The workaround is to use the option combination
--nco_opt='-x' -v v1,v2,...vn. The solution is to upgrade.
E. At some unknown point, the behavior of ncclimo in MPI mode on Cori
seems to have changed. Slurm on Cori does not (never did? no longer
does?) allow multiple srun commands to execute concurrently on the
same node by default. Thanks to Noel Keen to guiding me to the
options (--gres=craynetwork:0 --mem=20000) required to fix this.
This Slurm issue does not appear to affect any other DOE machines.
The workaround is to avoid MPI mode on Cori, the solution is to
upgrade.
Full release statement at http://nco.sf.net/ANNOUNCE
--
Charlie Zender, Earth System Sci. & Computer Sci.
University of California, Irvine 949-891-2429 )'(
