15 December 2005
Present: Beth Ebert, Rod Potts,
Sandy Dance, David Scurrah
Goals of meeting on XML format for
TIFS / TITAN / B08FDP
1. See to what extent existing WDSS XML files meet our needs (Sandy,
can you
bring a print out or two of WDSS files with cell data?). There currently
seems to be no schema for these files so we will need to create one at
some
point but not tomorrow. Work done by Rod and David and Mey so far with
TITAN
XML will be relevant here.
Can we find out from WDSS developers what state of
development their XML
output is at - is their file format set in stone or can they be
flexible to
suggested changes if we have any.
2. To decide what should be the next steps for us to develop XML for
our own
needs (TIFS, TITAN, B08FDP, other?). Ideally I would like to get a draft
version of XML schema(s) ready sometime next month to circulate among
B08FDP
participants and other nowcast developers (Senesi, NSSL) for comment.
This
will mean defining what the various cell attributes should be and what
they
should be called, etc. Hopefully the WDSS XML files will already contain
much of this information.
Notes and actions (red)
Rod and Mey have put together a draft XML format for TITAN. Click here to see
the schema, click here to
see an example of a TITAN XML file.
Sandy showed some examples of WDSS XML files from the BOM
implementation.
The WDSS schema does not organize data
by cell, as we would intuitively like,
so if we use WDSS we'll want to translate the files into an XML format
that is
more like what we want. WDSS uses 3 XML files for forecasts - "cell",
"meso",
"tds", which are not explicitly linked.
Robert Dahni (BOM Data Management Group) is part of a working group to
consider use of XML in the Bureau. We need to interface with them. Sandy will
talk with Mal Gorman (CCSB).
The Rapidly Developing Thunderstorm (RDT) project that Stephane
Senesi's group
is working on has nice data
definitions that we may wish to borrow or imitate. Beth
to ask Stephane and Paul Joe for a
list of BUFR products used by their systems,
with a view to compiling a list
for use in the Bureau and B08FDP.
David says it makes more sense to use several small schemas that are
more modular,
rather than one very large one. Schemas can use other schemas. David will work on
some primitives.