[OSM-newbies] GPS vs Tracing

[Richard Weait]

On Tue, Nov 30, 2010 at 3:44 PM, Donald Campbell II
<donaciano2000 at gmail.com> wrote:
> Okay so now Potlatch 2 is showing up on the Edit page along with Bing
> background so I've taken a look at my area.
> Since everything here was made from GPS tracks, how do I know which is
> correct, the GPS or the overhead images?  Perhaps I was driving on one side
> of the road and drew it off to the side by a good 10 feet or so, perhaps I
> had low accuracy...  on the other hand perhaps the aerial photos are off by
> a couple feet one way or another.
> How can I tell if I should scoot my tracks to match the overhead or leave
> them alone and assume they're more correct?  Also if the overhead images are
> off a few feet how can I get that fixed?
> Thanks,
> -Don.

Hi Don,

These are common questions, and common challenges for mappers since
the beginning of geography, not just the beginning of OpenStreetMap.

Firstly, every measurement comes with errors.  Every source, GPS,
aerial imagery, lidar, diving rod, they all have inherent errors.  Any
of the post-processing steps of format conversion, aligning to
benchmarks, copy-paste, projection, can introduce further errors.
Even if we check two different sources that agree, they might be
coming from a common, errant source.  We have to know and accept that.
 Frustrating, but true.

Secondly, aerial imagery can never be as current as your own survey
today.  At best, the aerial imagery was taken "yesterday", but likely
the image is even older.  So if you observe a difference between your
survey and aerial sources, it isn't always obvious if the difference
is due to a change since the imagery was acquired, or errors from one
source or another.

So what is a mapper to do?

Firstly, relax.  This is fun stuff.  Sometimes we have to decide,
"I've mapped enough today."  Worrying over every last detail, or every
last centimetre of imagined inaccuracy will extract the fun from your
life.

Secondly, map what you know.  Mapping places you see every day or
every week allows you to use all of the local context that you have at
your disposal.

Thirdly, I like to use different guidelines for adding new objects to
OSM and for editing objects in OSM.  When adding new objects, I feel
free to use the source I best trust.  Ideally this is a combination of
my own survey notes and GPX track, and aerial imagery if it exists.
When editing an existing object, I'm much more likely to trust the
work of the previous editor unless I know I have more-current
information.  Leaving source tags for future mappers is a nice way to
help them.

None of this helps when you have multiple, disagreeing sources with no
clear way to judge which is better.  So we do the best we can.  Split
the difference?  Generalize with a node, rather than a polygon?  Pick
one and describe with a note?  All of these work to one extent or
another.  And the experience of mapping what you know best can help as
you expand your range further from home.