netCDF Operators NCO version 5.0.4 are brachiating towards you.
http://nco.sf.net (Homepage, Mailing lists, Help)
http://github.com/nco (Source Code, Issues, Releases, Developers)
What's new?
Version 5.0.4 contains significant changes to the ncclimo splitter
API, minor changes to ncremap's --permute API and to invoking the
internal inverse-distance weight generation. Along with these
features come generally improved WARNINGs, HINTs, reduced
noisiness of diagnostics for frequently used commands, and
mitigation of compression issues triggered by netCDF 4.8.0+ library
changes. Furthermore, numerous minor improvements to mbtempest and
to ERWG support are included.
Work on NCO 5.0.5 has commenced and will improve ncremap support
for MOAB, incorporate new MOAB improvements, and further improve
quantization metadata and algorithms.
Enjoy,
Charlie
NEW FEATURES (full details always in ChangeLog):
A. ncremap removes the previous assumption that input horizontal
dimensions are in lat-lon order. The order can now be lat-lon
or lon-lat. The old behavior would fail when attempting to
regrid variables whose horizontal dimensions were lon-lat to
an unstructured grid.
B. ncremap now prefers to call its inverse distance-weighted
interpolation/extrapolation algorithm "idw" for "inverse
distance-weighted" rather than "dwe" for distance-weighted
extrapolation. However, both forms are acceptable and
back-compatibility was maintained.
C. ncremap adds the --split flag to select splitter mode.
Previously, the splitter was automatically invoked whenever the
input files were provided via stdin, globbing, or positional
command-line arguments, with some exceptions.
That older method became ambiguous and untenable once it was
decided to also allow climos to be generated from files provided
via stdin, globbing, or positional command-line arguments.
Now there are three methods to invoke the splitter:
1) Use the --split flag: this is the most explicit way to invoke
the splitter. 2) Select clm_md=hfs: the high-frequency splitter
mode by definition invokes the splitter, so a more explicit option
than this is not necessary. 3) Set the years-per-file option, e.g.,
--ypf=25: the ypf_max option is only useful to the splitter, and
has thus been used in many scripts. Since this option still causes
the splitter to be invoked, those script will perform as before.
ncclimo --split -c ${caseid} -s 1 -e 2 -i raw -o clm
ls ${caseid1}*h0.000[12]* | ncclimo -f ${fml_nm} -s 1 -e 2 -o out
http://nco.sf.net/nco.html#split
Thanks to Anthony Bartoletti and Jill Zhang (LLNL) for discussions
about this new option.
D. ncremap introduces changes the API for the --permute option:
Previously the argument needed to include the entire option
string to be passed to ncpdq including the -a component, e.g.,
--permute='-a time,new_dim,lat,lon'.
Now ncremap supplies the implicit -a internally so the user does
not need to know the ncpdq syntax.
ncremap --permute=Time,new_dim,nCells --map=map.nc in.nc out.nc
ncremap --permute=time,new_dim,lat,lon --map=map.nc in.nc out.nc
http://nco.sf.net/nco.html#permute
http://nco.sf.net/nco.html#pdq
Thanks to Walter Hannah (LLNL) for prompting this change.
E. ncremap adds the --area_dgn flag to diagnose (rather than copy) the
area of each gridcell in an inferred grid-file. By default, ncremap
simply copies the area variable, if found, into the grid_area
variable of the inferred SCRIP grid-file. When --area_dgn is
invoked, ncremap instead computes the values of grid_area based on
the gridcell boundaries in the input template file.
ncremap --area_dgn -d dst.nc -g grid.nc
http://nco.sf.net/nco.html#area_dgn
Note that --area_dgn has no effect on any mapping weights
subsequently generated from the grid-file because most
weight-generators base their weights on internally computed cell
areas (ERWG does have an option, --user_areas, to override this
behavior).
BUG FIXES:
A. ncwa was failing to set compression parameters.
This problem began with netCDF 4.8.0 which enforces stricter
adherence to the calling constraints rather.
The symptoms were errors when averaging compressed datasets.
The workaround is to convert the dataset to netCDF3, then average,
then recompress. The solution is to upgrade.
A patch has been submitted to netCDF 4.8.2/4.9.0 that will revert
the netCDF library to the previous trouble-free behavior.
Thanks to Thomas Reichler and Michael Rollings for reporting this.
B. netCDF 4.8.0 changed its internal implementation of deflation.
Before netCDF 4.8.0, deflation could be set multiple times.
The properties of final the final call would take effect.
Since netCDF 4.8.0 only the first call takes effect.
This caused subtle changes in NCO compression behavior.
The most visible change was that uncompressing variables silently
stopped working. NCO has been changed to adhere to the new
netCDF 4.8.0+ behavior.
Full release statement at http://nco.sf.net/ANNOUNCE
--
Charlie Zender, Earth System Sci. & Computer Sci.
University of California, Irvine 949-891-2429 )'(
