[Taxacom] Impossible geocodes

Richard Pyle deepreef at bishopmuseum.org
Fri Jan 30 13:12:43 CST 2015


I agree with everything Bob says, except that I have no issue with arbitrarily precise DD values.  The obvious answer, I think, is that there should be no inferred precision without an explicit "Uncertainty" value (i.e., representing a radius around the arbitrarily precise point, thereby defining an area).  Although I guess one could argue that there is somewhat more confidence in inferring precision from DMS values, I think that argument is weak.  In other words, we should always assume that DMS values are likewise arbitrarily precise without an explicit "Uncertainty" value.

Aloha,
Rich

> -----Original Message-----
> From: Taxacom [mailto:taxacom-bounces at mailman.nhm.ku.edu] On Behalf
> Of mesibov at southcom.com.au
> Sent: Thursday, January 29, 2015 11:44 PM
> To: alec.mcclay at shaw.ca
> Cc: taxacom at mailman.nhm.ku.edu
> Subject: Re: [Taxacom] Impossible geocodes
> 
> Alex, I agree with you that DD is a more useful lat/lon format, although I
> don't agree with you that DMS is weird, there's a good reason for it. And
> DMS is a little bit better than DD for implying an uncertainty.
> 
> If I write a latitude in the form 43d 16m 23s S, you might guess the latitude is
> known to the nearest second, or roughly +/- 11 m at that latitude. If I write
> 43d 16m S, you might guess that the latitude is only known to the nearest
> minute, or roughly +/- 670 m.
> 
> In DD, the two latitudes to 4 decimal places are
> 
> -43.2731
> -43.2667
> 
> in both cases implying an uncertainty of +/- 4 m. You can't tell from looking at
> these DDs in uniform format (say, to 4 decimal places) that one had a much
> bigger uncertainty than the other, although you might if the number of
> places given reflected the uncertainty, say -43.27 for the second one (roughly
> +/- 400 m at that latitude).
> 
> By an interesting coincidence, the implied uncertainty in DMS-to-the-
> nearest-second is about the same as the horizontal uncertainty in a handheld
> GPS reading (unless conditions are exceptionally good). The DD-to-4-places
> uncertainty is optimistic for handheld GPS, and the DD-to-5-places you often
> see in published papers is +/- half a metre and evidences a religious faith in
> numbers. DMS is definitely handy for GPS-determined locations.
> 
> I would back DD 100% if it was always accompanied by a specified
> uncertainty, like the Darwin Core one: 'The horizontal distance (in
> meters) from the given decimalLatitude and decimalLongitude describing the
> smallest circle containing the whole of the location'. Unfortunately, that's not
> the case. Very few of the museum collection databases I've looked at even
> have an uncertainty field for location data, and if they do it's largely unfilled.
> 
> Until collectors and data compilers and georeferencers routinely add an
> uncertainty to their lat/lons, I prefer to see 43d 16m 23s S and 43d 16m S,
> because I know for sure there's a difference.
> 
> (Sending this from my old email as Taxacom doesn't seem to like my new
> one, robert.mesibov at gmail.com)
> 
> _______________________________________________
> Taxacom Mailing List
> Taxacom at mailman.nhm.ku.edu
> http://mailman.nhm.ku.edu/cgi-bin/mailman/listinfo/taxacom
> The Taxacom Archive back to 1992 may be searched at:
> http://taxacom.markmail.org
> 
> Celebrating 28 years of Taxacom in 2015.




More information about the Taxacom mailing list