[CF-metadata] new chemical species

[Philip J. Cameron-smith]

Hi,

There are two key points I want to add:

1) One of the current problems with chemistry output is that there is 
currently NO agreed upon list of species names: many species have various 
common names, and those names are often abbreviated in codes, which can 
lead to confusion and incompatible files (eg ACET could refer to acetone 
or acetaldehyde)

It would be really good if an agreed upon list of names could be agreed 
upon bf CF.

2) The internal structure containing chemicals in models does NOT have to 
be the same as that used in the NetCDF output, and my preference is to 
have them be different.

Jonathan is correct that in all the models I have worked with the species 
are all in the same array with an additional dimension for specie, and 
sometimes an additional dimension for aerosol size.  There are a couple of 
problems with copying this into the netCDF file: first, it means 5 and 6 
dimensional arrays (lat,lon,alt,time,specie,size), which some plotting 
programs don't handle.  The large arrays can also be unwieldy (both memory 
and file-size constraints can be a problem).

Second, and more importantly, chemical mechanisms often change.  If the 
netCDF structure just uses a dimension for species, it is a pain to create 
new input files: I often find myself needing to add or remove species, or 
scale some of the species.  For analysis, I also usually find myself 
wanting to extract a long time history of just one or two species (out of 
a hundred), which is also a real nuisance with a single array for all 
species, and the process is prone to error.  There is also a minor 
nuisance that if a single specie is extracted, the degenerate dimension 
needs to be handled.

Lastly, it is usually sufficient to output a small subset of species 
(saving a lot of file-space), in which case the output array will not 
match the internal array anyway.

The alternative is to have each specie as a different variable in the 
NetCDF file.  The pain here is that the code needs to know the names of 
all the variables, and needs to read them in one by one.

I have used codes that handled this situation both ways.  Personally, I 
prefer to have each specie separate in the netCDF: the ease and accuracy 
of post-processing easily outweighs the disadvantages.

In conclusion, I would like to see:

A) A list of specie names agreed upon by CF,
B) Species stored separately in CF compliant netCDF files.

Best wishes,

      Philip


On Mon, 30 Jul 2007, Jonathan Gregory wrote:

> Dear Stephen
>
> When dealing with 5600 chemical species in a model, I guess that you must
> have arrays in which the chemical species is one of the dimensions, do you?
> In that case, allowing a dimension for chemical species in CF would be
> an obvious generalisation, as with surface cover types.
>
> Cheers
>
> Jonathan
> _______________________________________________
> CF-metadata mailing list
> CF-metadata at cgd.ucar.edu
> http://mailman.cgd.ucar.edu/mailman/listinfo/cf-metadata
>

------------------------------------------------------------------------
Dr Philip Cameron-Smith        Energy & Environment Directorate
pjc at llnl.gov                   Lawrence Livermore National Laboratory
+1 925 4236634                 7000 East Avenue, Livermore, CA94550, USA
------------------------------------------------------------------------