rrobbins at GDB.ORG
Tue Aug 16 20:59:27 CDT 1994
Steve Shattuck raises some important points regarding the conversion of
old place-name data into geographical coordinates. Another issue worth
considering is the problem of representing and conserving qualitative
as well as quantitative locations. If something is collected on the west
bank of a river, that qualitative west-bankedness ought to be preserved
against changes in the river's bed and it ought to be represented in a way
that computational rounding error would never allow the location to be
placed on the other bank no matter how many laps the information took
through some simulation program.
Similar problems arise in genetic mapping data. In some circumstances it
is possible to determine that gene A and gene B are some small distance
apart (say 2 units), that A is to the "left" of B (defined in some
reasonable genetic directional system), and that the absolute location of
the A-B complex in overall genome has an error of, say, plus or minus 20
If the quantitative locations, and errors, were attached to each gene
individually, it would seem to a reasonable person that the order was
unknown. But qualitatively the order is known, with very high certainty.
A challenge to genetic databases is to represent information such that the
most robust qualitative knowledge is not allowed to be eroded by
computation on less well established quantitative measures.
Similar challenges might usefully be recognized in geo-referenced data.
More information about the Taxacom