[Taxacom] Inappropriate accuracy of locality data

[Doug Yanega]

Arthur Chapman wrote:

>Somebody mentioned the difference between different DATUMs, and
>especially NAD83 and WGS84.  Indeed - for North America, there is little
>difference between these, however, the further you move away from
>continental North America, the greater the difference.  As far away as
>Hawaii - there may not be a great difference - I've never checked it,
>but NAD was never intended for use so far out into the Pacific. Many
>older maps, and older lats and longs would have been produced prior to
>1983, and in North America probably used NAD27 (North American Datum
>1927), and this can be up to hundreds of meters out (around 480 around
>the Aleutian Islands).  Indeed - on a recent trip to Alaska, I regularly
>found the map used in my GPS showed the road I was on to be consistently
>about 80 meters from my position.  As one reached a town, the maps had
>obviously been updated and I was back on the road.  This is a DATUM
>problem in the maps being used by the GPS company. More and more
>institutions are locking their GPSs to WGS84 or the local equivalent
>(NAD83 in North America; AGD84 in Australia), but I often see people
>using GPSs and have no idea what Datum they are using.  In the Guide to
>Best Practices for Georeferencing we provide a table showing some the
>variations between a number of DATUMs.  We also refer to the MANIS
>Georeferencing Calculator which will calculate the error for any
>position using a large range of different DATUMs.

This is, in a practical sense, rarely likely to be important for any 
vagile terrestrial organisms; something like a single plant, or 
something occurring in a body of water, might indeed require one to 
be precise and careful about one's coordinates, but for the vast 
majority of organisms, if you can come within 100m radius of the true 
collecting point, this is more than sufficient for virtually any 
conceivable purpose. An error of 80m is insignificant, with few 
exceptions. Dealing with legacy data makes the issue even _less_ 
important, because the uncertainty on a typical legacy record (of 
which I have processed well over 100,000 at this point) is between 1 
and 2 km - which utterly swamps out any miniscule* uncertainty 
associated with changing map datums. And when one gets to legacy data 
such as "Chicago, IL" and you're looking at a radius of 25 km or 
more, worrying about which map datum one uses is meaningless. That's 
why the uncertainty parameter is THE essential element without which 
one cannot have a properly georeferenced database. Giving the 
"center" of Chicago to the nearest centimeter is theoretically 
possible, and it may certainly look that way if one has an entry in 
one's database that reads

41 52 38.84 N, 87 39 08.48 W

but it would be a serious mistake to imply or infer that a given 
specimen was actually collected in that particular square centimeter 
- a mistake that can be avoided only by including an uncertainty 
parameter.

*http://www.barnsdle.demon.co.uk/spell/mini.html ;-)

Peace,
-- 

Doug Yanega        Dept. of Entomology         Entomology Research Museum
Univ. of California, Riverside, CA 92521-0314        skype: dyanega
phone: (951) 827-4315 (standard disclaimer: opinions are mine, not UCR's)
              http://cache.ucr.edu/~heraty/yanega.html
   "There are some enterprises in which a careful disorderliness
         is the true method" - Herman Melville, Moby Dick, Chap. 82