georeferencing old specimens

[DAVID C. TAYLOR]

I agree that the asignation of discrete point coordinates to long since
collected specimens with ambiguous geospatial annotation can seem quite
arbitrary.  I think that the solution that will ultimately be decided
upon will involve assigning discrete coordinates in conjunction with a
separate field containing a confidence value.  This confidence value can
be used to evaluate the integrity of the data returned in the other
fields.  In fact one could even add a value for 'Generalized to County
Level Due to Sensitivity' or some such.

The Flora of Texas Consortium has been grappling with this issue in
preparation for issuing standards for the interchange file for all member
herbaria with the ultimate goal of achieving a seamless statewide
database.  See the specifications at:

http://www.csdl.tamu.edu/FLORA/ftc/ftcffld3.htm

Sincerely,

David C. Taylor
Undergraduate, Department of Biological Sciences, Sam Houston State University
GIS Assistant, Texas Regional Institute for Environmental Studies
stddct at unx1.shsu.edu
http://www.shsu.edu/~stddct/


> > At 10:15 AM 12/6/96 -0500, Robin Panza wrote:
>
> > >That's all very well and good for future collecting, but some of us maintain
> > >100-year-old collections.  There is no way to produce GPS reference data for
> > >the 200,000 bird specimens (and probably ca. 1,000,000 specimens in the other
> > >departments) already in our posession.
>
> I've been misunderstood on this subject.  I was not referring to the amount of
> time/effort it will take to provide the lat/long coordinates.  In fact, I have
> already developed a gazetteer of our localities.  I was referring to the
> degree of accuracy of the localities.  Many are "near" some town, with or
> without direction.  Others are from within driving distance of a base camp, in
> all directions.  Still others are road miles from named places, with the roads
> no longer reliably in the same places, or curving such that one cannot be sure
> exactly where that mileage point occurred.  This does not include railroad
> localities on defunct lines, hamlets whose names and/or locations have changed,
> and best-guesses as to where the person really was.  Nor does it include
> large-scale localities like "Paramo de Rosas" or "Shandong, China", or
> "Kissimmee Prairie", or "Rio Quimeche, Nayarit".
>
> Producing pinpoint-accuracy numbers compatible with modern GPS precision is
> grossly misleading.  Even lat/long to minutes is not necessarily very
> accurate in terms of where the animal really was at the time of its demise.
> Such inaccuracy of locality information should *not* be arbitrarily given a
> more precise numerical location that may or may not include the correct spot.
> It implies knowledge that isn't known.