On Wed, 31 Oct 2007, Ben Domenico wrote:
As you are no doubt aware, that issue has to be addressed on multiple fronts.
-- The ISO 19123 definitions are general enough to support such
datasets as coverages.
-- The WCS specifications have been more restrictive to the point
where the grids must be regular in some (perhaps) projected coordinate
-- The CF conventions community has, to this point, focused on the
regular (or at least quasi regular) grids that are the output of
numerical forecast models.
irregular grids still needs some work. If we can complete the process
for the case of "station observations," I believe the others will
follow more quickly by benefit of the hard lessons we are learning
from the first two.
One question I have for you and the other members of the coastal
community is whether those unstructured/irregular grids are an example
of one of the current CDM scientific data types
or whether a new CDM data type is needed.
From that document, the Station (or perhaps Swath?) type seems like the
closest CDM Scientific Data Type to the elements in an unstructured
grid model. However there does not seem to be a clear method to limit the
domain of applicability of a Station's data to some non-simple spatial
region associated with a CDM Station. Unstructured mesh modelers use
nodes and a connectivity matrix to represent the regions over which their
data applies, and are writing these in netCDF outside of the conventions.
Maybe a 'Cell' or 'Element' data type using nodes/vertices and some sort
of connectivity would be a valuable extention.
Dr. David Forrest