On Tue, Jan 2, 2018 at 11:03 AM, Karl Taylor <taylor13 at llnl.gov> wrote:
> In CMIP we require that both missing_value and _FillValue be defined as
> 1.e20 for requested variables that occupy only a portion of the global
> grid. For bounds variables, I'll suggest that when needed, both _FillValue
> and missing_value also be defined as 1.e20. That way software that relies
> on either one will work.
>
good practice, as the use-cases for _FillValue and missing_value often get
conflated. But it's still a good idea to have one clearly specified as the
preferred one to use for a particular meaning.
> My understanding of UGRID is that it provides for a much fuller
> description of a grid than CF.
>
well, yes, that's why it exists -- but I'm not sure "fuller" is the right
word -- it provides a standardised way to describe unstructured meshes,
which CF simply does not support.
For CMIP, I think we can recommend UGRID be used for non-standard grids,
> but I'm reluctant to require it
>
"standard" isn't the issue - -the issue is whether the grid is structured
or not -- if it is not a structured grid, then there may be a way to
describe it in a CF compliant way , but there won't be ONE way to describe
given type of grid, which, again, is why UGRID conventions were developed.
Why be reluctant? I haven't followed CMIP at all, but it sure looks like
the goal is to have things standardized.
Most grids for most purposes I think can be adequately described with the
> CF-mandated attributes. Does anyone disagree?
>
Well, yes -- again, I haven't followed CMIP, so I don't know what types of
grid in practice you need to support. If you don't need to support
unstructured grids, then, there isn't an issue. But if you do, you really
want to standardize how they are defined, and yes, they CAN be adequately
described by CF (a UGRID file, should also be valid CF). What UGRID
provides not an alternative to CF, but an extension that standardizes how
those grids are described.
By the way, the UGRID standard maps pretty well to how unstructured grid
models generally store their grids and read and write them anyway, so it
shouldn't be a heavy lift to support it.
-CHB
--
Christopher Barker, Ph.D.
Oceanographer
Emergency Response Division
NOAA/NOS/OR&R (206) 526-6959 voice
7600 Sand Point Way NE (206) 526-6329 fax
Seattle, WA 98115 (206) 526-6317 main reception
Chris.Barker at noaa.gov
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mailman.cgd.ucar.edu/pipermail/cf-metadata/attachments/20180102/6f70dede/attachment.html>
Received on Tue Jan 02 2018 - 13:17:53 GMT
