[Tagging] Accuracy of survey

Tom Pfeifer t.pfeifer at computer.org
Tue Dec 23 16:57:37 UTC 2014


I would consider that a non-issue as you said, for those reasons:

- When it comes to GPS traces on objects that don't move (*), the
   beauty of crowdsourcing is on our side. The collection of
   traces over a longer time creates a cloud of traces which
   form a Gaussian bell curve, in density, over the ground truth.

   Thus a junction of two road traced again and again is still
   a good reference point to calibrate aerial imagery.

- We are getting access to increasingly better geo-referenced
   aerial imagery, thus mapping can now use different sources
   and calibrate between them.

The real issue is that in urban areas, lots of object, mostly
houses in absence of GPS traces, have been mapped with offset
imagery and need to be moved a bit.

But this has no implication on tagging.

(*) emphasis on fixed objects, since our friends from OpenSeaMap
have more difficulties creating such repeatable GPS traces since a
ship has no fixed road it would use again and again.

tom

Rainer F├╝genstein wrote on 2014-12-23 17:37:
>
> while we are at it, imagine the following situation:
>
> mapper A, by means of DGPS, MilStd GPS, crystal ball etc., is able to
> achieve an accuracy of, say, a few centimeters and uses it to add new
> nodes (POIs) to OSM.
>
> some time later, mapper B with his/her ancestors mechanical GPS device
> (*), achieving an accuracy of max., say, 15 meters, surveys the same
> area, figures out that (by his/her point of view) POIs added by mapper
> A are 15 meters off and corrects their location.
>
> what is needed here is some tag, saying "don't touch these
> coordinates, they've been surveyed with high(est) accuracy".
>
> I heard this argument from an pipeline expert, noting that marker
> surveyed with consumer GPS are (for their standards) way off their
> real location.
>
> maybe this is a non-issue after all, if consensus is that consumer
> GPS accuracy is sufficient enough.
>
> cu
>
> (*) http://www.kenalder.com/measure/excerpts.htm




More information about the Tagging mailing list