[Taxacom] odometer georeferencing

[Doug Yanega]

Fred Schueler wrote:

>Like many who antedate 1995, I've got a huge body of observations
>"georeferenced" by odometer readings along highways.
>
>Kari Gunson - http://www.eco-kare.com/ - has proposed to GIS these by
>listing the geographic co-ordinates of every 10m of the highways they're
>along, so I'll just be able to select reference points from web maps,
>calculate the distance from the reference points, and then assign the
>observation the co-ordinates that match these distances.
>
>I wonder if this method has been used by others, and if there's any
>place where such tables are archived? I haven't tried to search for such
>accounts on line, since I don't know what the method would be called.

As someone who has a full-time technician hired just to do 
georeferencing, I had not heard of any such attempts at mapping 
"reference points" along highways, but it sounds very promising - 
right now, this sort of procedure is done manually with Google 
Earth's path-tracing tool. That being said, one observation MUST be 
kept in mind:

There is a difference between specimen labels that refer to highway 
markers versus labels that refer to odometer readings.

A label based on highway markers *should*, in principle, explicitly 
state "marker #" somewhere, and if it does not, then making the 
assumption that the label refers to a marker can lead to error, given 
that mile/kilometer numbers don't necessarily start at the edge of a 
town (which is where most people will zero their odometer reading). 
Accordingly, there can be up to a several-mile discrepancy between 
the two methods; we have enough cases of known localities with 
various collectors' mileage readings to be quite confident that one 
*cannot* start measuring distance from the center of a town when 
georeferencing. At this point, for cases without highway markers, 
there really is no good substitute for a human being visualizing 
where he/she would zero the odometer, using a Google Earth satellite 
photo. Even the best automated lookup tools (i.e., those with error 
radius, like Biogeomancer) are frequently *woefully* far off simply 
because they use the center of a town as the starting point (e.g., 
just imagine what happens with automated lookup of a specimen from "1 
mi E" of a city that is 5 miles in radius).

Highway markers should also, in theory, be quite stable once 
georeferenced, where the edge of a town is a subjective quantity, and 
subject to *wide* interpretation - especially when dealing with 
legacy material 50 or more years old. At least for US material, our 
experience is that the vast majority of labels use odometer readings, 
and thus are more subject to error. I have little doubt that this is 
correlated with the inconsistency across the US in the presence and 
visibility of highway markers.

It would definitely be interesting if there were variations between 
countries in how highway markers were numbered (e.g., if any of them 
started from the edge of a town/city, and what happens if that edge 
shifts over time).

At any rate, Gunson's proposal would create a very nice and useful 
tool, certainly, but like any tool, one has to be careful to use it 
for the proper job. But Eco-Kare is a commercial enterprise - would 
these data be free, or for sale?

Sincerely,
-- 

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