[CF-metadata] new TEOS-10 standard names

From: Jonathan Gregory <j.m.gregory>
Date: Tue, 26 Jul 2011 16:45:27 +0100

Dear all

I understand the need to be clear, with new standard names, which observational
quantity is being collected in future. I do not agree, however, that we should
make the plain "salinity" name an alias for something more precise. This is
partly because that might change the meaning of existing data, possibly
incorrectly as Trevor points out. Partly it is also because I think it is
quite possible that models, perhaps idealised, may be used in which it would
not be meaningful to be more precise than just "salinity".

Trevor argues that existing ocean models use practical salinity because (a)
they are initialised with observations of that and (b) they assume so in their
equation of state. I don't think (a) is necessarily so. In some cases, they
might not be initialised with observations, for instance in idealised
investigations of spin-up. Even when initialised from obs,
they will almost certainly drift to a less realistic state. I don't know enough
about it to be sure about (b). Unless we could be certain this is always the
case, I think plain "salinity" should be retained for possible use in models.
However, we could certainly recommend that models should use one of the new
more precise terms if definitely appropriate. This recommendation could be
included in the standard_name definition of plain "salinity".

I understand the existing standard name sea_water_temperature to mean in-situ
temperature, as it does for air temperature. This could be stated in the
definition.

The purpose of standard names themselves is not to prescribe or recommend what
quantities people should store in netCDF files. It is to allow them to describe
with sufficient precision the quantities they have chosen to store, in order
to make it possible to decide which quantities from different datasets should
be regarded as comparable.

Standard names are all in lower case, regardless of what case is used in
ordinary writing. This is for simplicity in matching strings. Case-sensitive
matching would inevitably trip people up and cause a nuisance when they got
it wrong.

Best wishes

Jonathan
Received on Tue Jul 26 2011 - 09:45:27 BST