netCDF Operators NCO version 5.0.7 are unfurled
http://nco.sf.net (Homepage, Mailing lists, Help)
http://github.com/nco (Source Code, Issues, Releases, Developers)
What's new?
Version 5.0.7 has new fixes and features for ncremap and ncclimo
that ensure correct treatment of sub-gridscale fields and missing
values. ncremap now allows and adjusts for missing values in SGS
fields. Previously such missing values could botch SGS normalization.
Users of SGS regridding are encouraged to upgrade. Others might skip
this release.
Work on NCO 5.0.8 has commenced and will improve support for modern
quantization/compression algorithms expected to be in netCDF 4.9.0.
Enjoy,
Charlie
NEW FEATURES (full details always in ChangeLog):
A. ncremap now emits a WARNING when a dataset contains a sub-gridscale
(SGS) indicator variable, such as landfrac (for ELM/CLM), or
timeMonthly_avg_iceAreaCell (for MPASSI), or aice (for CICE), and yet
SGS regridding was not invoked. This will help avert some inadvertent
misuse of the regridder by users unfamiliar with the need for SGS
regridding of datasets where values are only valid over fractions of
gridcells.
ncremap -P elm --map=map.nc in.nc out.nc
http://nco.sf.net/nco.html#sgs
B. ncremap now allows for missing values in fields with valid SGS area.
ncremap previously assumed to be valid wherever sgs_frc > 0.0.
This assumption is valid for ~95% of ELM/CLM fields.
However, it low-biases landunit-specific fields (like TSOI, H2OSOI) in
transition areas (e.g., ablation zone) with valid SGS, invalid landunit.
Now sub-SGS values are extrapolated, incorrect though not crazy-bad.
No more temperatures near 0.0 for landunit-specific fields in ELM/CLM.
Deriving and using landunit area as SGS field is only correct method.
The current method is the best possible with that uses a single SGS
fraction for normalization.
ncremap -P elm --map=map.nc in.nc out.nc
http://nco.sf.net/nco.html#sgs
BUG FIXES:
A. NCO 5.0.6 produced incorrect results when sub-gridscale (SGS)
regridding is invoked by calling ncclimo (or ncremap) with the -P
prc_typ option alone because an ncremap bug prevented that option from
being fully parsed. As a result, the -P prc_typ did not correctly
trigger SGS regridding. The workaround to obtain proper SGS
normalization with NCO 5.0.6 is to explicitly invoke ncclimo (or
ncremap) with the -P prc_typ option AND the --sgs_frc=landfrac
(for ELM/CLM) or --sgs_frc=timeMonthly_avg_iceAreaCell (for MPASSI)
options. To be clear, this workaround only need be applied to
processing of ELM/CLM (not EAM/CAM) and MPASSI (not MPASO) output.
It is recommended to use -P where appropriate, and -P remains
essential in processing MPAS data. As of version 5.0.7, -P is all
that is required, and --sgs_frc is only needed in special cases.
The solution is to upgrade, and let -P fulfill its intended
role of triggering SGS regridding and other dataset-specific tasks
without the user herself having to explicitly turn on all the
switches.
B. ncremap's --add_fll option previously always wrote NC_FILL_DOUBLE
into unmapped gridcells. Now it will prefer to write the defined
_FillValue if one exists.
Full release statement at http://nco.sf.net/ANNOUNCE
--
Charlie Zender, Earth System Sci. & Computer Sci.
University of California, Irvine 949-891-2429 )'(
