« EarthCube 2021 Annua... | Main | New: Downloadable... »

17 May 2021

Version 4.9.9 of the netCDF Operators (NCO) has been released. NCO is an Open Source package that consists of a dozen standalone, command-line programs that take netCDF files as input, then operate (e.g., derive new data, average, print, hyperslab, manipulate metadata) and output the results to screen or files in text, binary, or netCDF formats.

The NCO project is coordinated by Professor Charlie Zender of the Department of Earth System Science, University of California, Irvine. More information about the project, along with binary and source downloads, are available on the SourceForge project page.

From the release message:

Version 4.9.9 contains more fixes than features, and can probably be skipped by those who do not use ncremap or ncclimo. ncclimo now supports the mth_srt/end options for high resolution timeseries that do not begin/end in Jan/Dec (previously it only supported those arguments for monthly resolution). ncremap has crucial fixes for weight generation and map-checking for masked grids, fixes for curvilinear grid inferral, and automatic rounding when regridding integer-valued fields (like masks).

New Features

ncclimo now supports options --mth_srt and --mth_end in high-frequency splitter mode. Previously it only supported those arguments for monthly resolution timeseries. The arguments specify the (1-based) month in which the requested high-frequency timeseries should begin and end, respectively. These default to 1 (January) and 12 (December). To extract 14-month timeseries from high frequency (e.g., daily-resolution) input files one would use, e.g.,
```
ncclimo --clm_md=hfs --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
ncclimo is now smarter about inferring both the temporal resolution of high frequency timeseries, and the number of timesteps per file. As a result, explicit invocation of the tpd (timesteps-per-day) and dpf (days-per-file) option should no longer be necessary. Explicitly specifying these options will, of course, prevent ncclimo from trying to infer them.
```
ncclimo --clm_md=hfs           ... # This should work fine
ncclimo --clm_md=hfs --tpd=8   ... # No longer necessary
ncclimo --clm_md=hfs --dpf=10  ... # No longer necessary
```
http://nco.sf.net/nco.html#ncclimo
http://nco.sf.net/nco.html#tpd
http://nco.sf.net/nco.html#dpf
ncremap now automatically rounds integer-valued fields, such as masks. Previously when ncremap regridded an integer-valued field, it used implicit type-conversion rules to convert the regridded destination value to an integer value for output. However, users generally expect a regridded value of, say, 1.9999999, to be converted to the integer 2 not 1. Hence, ncremap now automatically applies the rint() function to internal floating point values before converting them to integers for output. Regrid integer valued fields such as masks now generally behave as desired.
ncremap now always invokes TempestRemap's GenerateOverlapMesh with its --allow_no_overlap option. This option permits TR to generate meshes from not-completely-overlapped regional grids. This is often useful in post-processing regional data, for example.
http://nco.sf.net/nco.html#tr
http://nco.sf.net/nco.html#allow_no_overlap
Conda builds should now be more robust, due to extensive testing. Thanks to Xylar Asay-Davis for helping us understand and alter the nco-feedstock for Conda. As a result of this, OpenMP should work on all MacOS Conda installations, and thus make map-file generation much faster for high-resolution maps.
This version works-around a regression in netCDF 4.7.4 (fixed in 4.8.0) that prevents NCO from manipulating HDF4 files, and a regression in netCDF 4.8.0 (subsequently fixed in 4.8.1) that prevents NCO from building its main test file from CDL, and thus prevents NCO from building without manual intervention.