[tdwg-tag] DwC change review: geo terms

[John Wieczorek]

Hi TAGers,

I am deep in the review process for the proposed changes to Darwin
Core, trying to do due diligence. Some of the change requests are
challenging to summarize to determine if there is consensus, in spite
of, or because of the discussions. One of the requests on which I’d
like some TAG help before proposing a solution is the request for the
inclusion of the terms from the geo: namespace
(xmlns:geo="http://www.w3.org/2003/01/geo/wgs84_pos#").

Support in tdwg-content for this request comes from multiple
independent sources. There has been a long history of discussion
(http://lists.tdwg.org/pipermail/tdwg-tag/2010-August/000050.html),
beginning in anticipation of the 2010 TDWG BioBlitz. The proposal has
gone through the minimum 30-day public review and discussion on the
forum tdwg-content at lists.tdwg.org:

http://lists.tdwg.org/pipermail/tdwg-content/2011-July/002581.html

There seems to be general support for the additions, however, after
reviewing the discussions and the references. I have the following
observations/concerns:

1) The discussions presented geo:lat and geo:lng as W3C standards.
This is not actually the case. These terms were created by the W3C
Semantic Web Interest Group in 2003. The documentation for these terms
(http://www.w3.org/2003/01/geo/) states:

"This document was created as an informal collaboration within W3C's
Semantic Web Interest Group. This work is not currently on the W3C
recommendation track for standardization, and has not been subject to
the associated review process, quality assurance, etc. If there is
interest amongst the W3C membership in standards work on a
location/mapping RDF vocabulary, this current work may inform any more
formal efforts to follow."

These terms do seem to have widespread usage in the semantic web.
Should we be concerned that they are not part of a standard?

2) geo:lat and geo:lng are not semantically equivalent to the existing
Darwin Core terms decimalLatitude and decimalLongitude, which have
been a part of the Darwin Core since it 2003 (or before, if we ignore
the missing Datum term in earlier versions). The addition of the geo:
terms as a third set of geolocation terms for Darwin Core raised
concerns about confusion. I share this concern. An option would be to
adopt these terms and deprecate dwc:decimalLatitude, dwc:Longitude,
and dwc:geodeticDatum. Data that would have occupied these terms would
go instead to dwc:verbatimLatitude dwc:verbatimLongitude, and
dwc:verbatimSRS. I see a couple of problems with this. First, most of
the time the data in the decimal coordinate fields are not the
verbatim originals, so this would be a misuse of the Darwin Core
terms. Second, this change would make it more difficult for data
consumer’s to use existing georeferences. Here’s how. Right now the
verbatim fields are meant to hold the original coordinate information,
which means they have a wide variety of content - everything from UTMs
to custom-encoded coordinates, in any conceivable format. Meanwhile,
the data in the decimal coordinates fields can be much more readily
transformed into the desired standardized spatial reference system
afforded by the geo: terms, because the values are at least
standardized on decimal degrees and only a datum transformation has to
be done on them.

Do we abandon the dwc: terms decimalLatitude, decimalLongitude, and
geodeticDatum? Do we abandon them now? Do we build the simplest
possible tools necessary for anyone to do the transformations so that
these terms are no longer needed? If so, do we wait until those tools
exist?

3) Additional concern was expressed that the term geo:alt should also
be added. No one has made a formal request for this. However, if the
other geo: terms were adopted, it might be silly not to adopt this one
as well. Doing so would raise a host of issues similar to those raised
for lat and lng.

I don’t have a good solution. The best short-term one, in my opinion,
is to leave Darwin Core as it is, and to recommend that if
applications (or aggregators) want to share “cleansed” point-based
georeferences, that they do so with the geo: tags, the values for
which they derive through transformations to WGS84 of the DwC decimal
coordinates and geodeticDatum.

Options:

1) Accept the proposal, adding geo:lat, geo:lng, and geo:alt to the
list of recommended terms for DwC.

2) Reject the proposal pending further directed research into a
comprehensive solution that considers all geospatial terms in Darwin
Core (including footprintWKT, for example).

3) Reject the proposal for now, reopening the public discussion with
these concerns.

Others?