Hi Ted,
Thanks very much for the info - I'll need to study the attachments!
I think this an additional harvesting method we'd need to add to the
GeoNetwork harvester functions I mentioned in the other email: namely:
search global attributes of any (atomic) dataset for metadata (or very
crudely: if no threddsmetadata elements for dataset in the catalog then
try to discover by opening the dataset, retrieving global attributes,
assembling any metadata from the global attributes (using your
metadata_conventions attribute) and then apply an appropriate xslt to
create the ISO record). Sound reasonable?
Do your fgdc global attributes map through any of the fgdc to ISO
cross-walks that are out there? (ie. can we support this in the
GeoNetwork harvester right now by using implementing such a cross-walk
as an xslt that maps from the metadata in your global attributes
directly to the elements of the ISO record being created by the harvester?).
Cheers and thanks very much,
Simon
Ted Habermann wrote:
Simon et al.,
The idea of harvesting metadata from THREDDS servers is an interesting
one that I thought might deserve its own subject line. We did some
work on this awhile ago and had a few ideas that are instantiated in a
presentation I gave at the Unidata Interoperability Day (attached, use
slide show mode) and in some of the datasets being served on our
THREDDS Server at NGDC (see for example
http://www.ngdc.noaa.gov/thredds/catalog.html?dataset=SST-100km-Agg).
The global attributes for this dataset include a significant number of
fields that were extracted from an FGDC record via an XSLT into NCML
that was used to start the file. Which elements are recorded in the
file is obviously use case dependent, but there are two that we
thought were critical to supporting efforts like yours. The first is
the “Metadata_Conventions” attribute, which is similar to the standard
Conventions attribute in CF. This attribute gives the metadata
conventions that are supported for this dataset. In this case "FGDC,
Unidata Dataset Discovery v1.0". Obviously it is our goal to add ISO
191* to this list, probably 19115-2 in this case.
The other critical tag is the *_metadata_url tag. The idea would be to
have one of these for each supported convention. It provides the URL
for the complete metadata record, in this case:
"http://ngdc.noaa.gov/nmmr/xmls/fgdc.jsp?id=gov.noaa.class:SST100";
(which is a bit out of date, but still redirects to the correct
metadata record).
These two tags provide a clear workflow for harvesting metadata from
THREDDS Servers: determine if the convention you are interested in is
supported and, if so, find the URL for the complete record. We also
thought that this convention could be used to add richer metadata
services to the THREDDS framework.
Of course many questions remain, particularly about metadata
granularity (do you really want a record in GeoNetwork for each
granule?) and integration of access (is pointing to the THREDDS Server
enough?), but we thought this idea was worth pursuing.
Ted
------------------------------------------------------------------------
==== Ted Habermann ===========================
Enterprise Data Systems Group Leader
NOAA, National Geophysical Data Center
V: 303.497.6472 F: 303.497.6513
"If you want to go quickly, go alone.
If you want to go far, go together"
Old Proverb
==== Ted.Habermann@xxxxxxxx
<mailto:Ted.Habermann@xxxxxxxx> ==================
------------------------------------------------------------------------
