Rich and CF'ers:
Big topic. But maybe not so big. My FEM files are a subset of the
rest of the world, of course, but they cover most of the ground. I
opened up a question with CF, months ago, about a standard name for the
connection array which defines a triangle mesh. I give it the long name
"Horizontal Triangular Element Incidence List". And since it is an
array of integers which are indices into the other arrays I felt it
needed units which say so and which differentiate number conventions of
0:n-1 or 1:n . So units="index_start_1"
We also need the boundary description with similar units,
long_name="Boundary Segment Node List"
These two arrays are really all that is needed to uniquely describe an
unstructured mesh. Well the Element incidence list needs a dimension of
n,3
for triangles, or n,4 for quadrilaterals etc. n,4 handles mixed meshes
if you fill the triangle's fourth index with -1 or some such flag.
My web page and link to my SourceForge project to build these files:
http://ccmp.chesapeake.org/HYDRONetCDF/HYDRONetCDF.html
Tom Gross
Chesapeake Community Model Program
ccmp.chesapeake.org
Chesapeake Research Consortium
Edgewater, MD 410-798-1283
or on Monday and Tuesdays
tom.gross at noaa.gov
NOAA/NOS/Coast Survey Development Lab.
Silver Spring, MD (301) 713-2809 x139
-----Original Message-----
From: Rich Signell <rsignell at usgs.gov>
Sent: Tuesday, November 09, 2004 5:22 PM
To: cf-metadata at cgd.ucar.edu
Subject: Conventions for Unstructured Grid Data
CF-ers,
I gave a presentation on OpenDAP, CF Conventions, THREDDS & IDV
to a group of ocean modelers today, and they were very enthusiastic.
A fair number of modelers in this group used unstructured grids
(triangle-based) and they were wondering if there were guidelines
for them to follow as well.
Has the CF Group thought about this community?
I did a Google search, and I turned up this statement in one
of the the Earth System Modeling Framework documents that I don't
understand:
(From the Earth Sytem Modeling Framework site
http://www.esmf.ucar.edu/esmf_docs/ESMF_reqdoc/node13.html)
"We assume that the NetCDF conventions for climate and forecast
metadata, ``CF conventions'', will serve as a a basis for ESMF metadata
conventions. NetCDF Climate-Forecast Metadata Conventions [6
<
http://www.esmf.ucar.edu/esmf_docs/ESMF_reqdoc/node16.html#NetCDF_CF_v1
_beta3>]
narrow definitions of NetCDF, an array-oriented data format and a
library for gridded data [4
<
http://www.esmf.ucar.edu/esmf_docs/ESMF_reqdoc/node16.html#NetCDF3_User
sGuide_C>],
to allow a unique and complete description of gridded data used in
geoscience. CF conventions specify standard dimensions, such as date or
time (t), height or depth (/z/), latitude (/y/), and longitude (/x/),
and specify standard units for these dimensions and other quantities.
We expect ESMF metadata conventions to be based on the CF-conventions,
and to cover fields defined on both structured and unstructured grids,
as well as observational data. Unlike the CF conventions which are
tightly associated with NetCDF, the ESMF conventions are supposed to be
format neutral, and cover all of the ESMF data formats. These extensions
will become the ESMF standard, and will be enforced by the ESMF I/O
subsystem. However, the specification of ESMF metadata is optional, and
users desiring not to specify any metadata should be able to do so."
Does this mean that the CF folks are working with the ESMF folks, and
that
unstructured grid data might be handled in a future release of CF?
Thanks,
Rich
--
Richard P. Signell rsignell at usgs.gov
U.S. Geological Survey Phone: (508) 457-2229
384 Woods Hole Road Fax: (508) 457-2310
Woods Hole, MA 02543-1598
Received on Wed Nov 10 2004 - 09:31:02 GMT
