[Talk-us] Labeling forestry service roads/tracks

[Matthew Woehlke]

On 19/07/2020 18.47, tj-osmwiki at lowsnr.net wrote:
> Editing in Boundary County, Idaho in the Panhandle, I've been extending
> the forest landuse area around Bonners Ferry and have come across a
> difficulty in classifying forest roads.
> 
> It seems that many have been automatically imported and have
> highway=residential, which is just plain wrong.

FWIW, this seems to be endemic in TIGER data. I often suspect that 
everything that isn't a primary or secondary gets marked "residential".

> For roads that appear metalled (paved) and/or access mines, quarries,
> communication towers etc. I label highway=service, for roads that are
> unpaved or sometimes seem to almost fade out I label highway=track. For
> roads that appear to be public access (e.g. to go to a lake) but are
> obviously even more minor than tertiary roads I label highway=unclassified.

Sounds about right, at least for the first and last. I'm less certain 
about "highway=track". (Not saying it's *wrong*, just that I don't know, 
vs. the others which sound correct to me.) Well, modulo Mike's comment; 
where I've been using "highway=unclassified" is for things that really 
don't look like service roads (e.g. that connect to other road networks) 
but likewise are clearly not residential. For example, 
https://www.openstreetmap.org/way/20453748.

> TIGER seems to be at best very coarse, at worst fictional.

Yup, that is known to be the case. As I understand it, TIGER was created 
mainly for census-taking, and so as long as someone on the ground could 
look at the map and figure out more or less how to get to the houses on 
a particular road, that is "good enough". Positional accuracy in that 
respect isn't nearly as important as *connectivity* accuracy, which 
partly explains the quality, but even connectivity can be dodgy. (As you 
noted, it's not unusual to be missing entire roads, or to have roads 
that don't really exist, and that's *before* we start worrying about 
changes that have happened since.)

-- 
Matthew