[Talk-us] GIS projection of OSM files

[Bill Ricker]

On Mon, Jul 18, 2022 at 1:33 PM Mike Thompson <miketho16 at gmail.com> wrote:

> The data in OSM is in the WGS84 datum, it is not projected.
> Some tools such as the JOSM editor, project the data.  The most common is
> to World Pseudo-Mercator (EPSG: 3857).
>

(Agree)


> I would like to know what is the GIS projection of the OSM *files* we can
> export from Open Street Map.
> I believe what you are using it not "wgs84", it is something else.
> Could you please confirm what is the GIS projection of your OSM *files*
> are?
>

Which do you, Tamas, mean by "*files*"?

To OSM field editors, we think of OSM as the *Database* - *un*projected
points on the WGS84 geoid.*
One can indeed export data files of (Lat,Lon,attributes), and they are
*un*projected
WGS84.

To most downstream users, OSM means the *map tileserver* (which is
obviously projected, it's flat).
Nearly all tileservers are *Web Mercator* (we had a chance to move forward
from Mercator and missed it :-/ )
aka  *World Pseudo-Mercator* (EPSG: 3857), to which OSM conforms, as Mike
said.
The tile server and tools that layer tiles and data (e.g. JOSM iD etc) have
to project the data to match, naturally.

* Specifying only WGS84 ellipsoid without further specification of datum &
CRS (G2139? G1674? G1150? G1762? ) is not adequate for fine-scale precise
GIS work. But that fuzziness captures the fuzziness of user-contributed GPS
data, so it's at least consistent. Any OSM points you need in your GIS at
precision greater than WGS84(??) you'll need to either source from
elsewhere (upstream if we got it from a government GIS) or field check with
professional equipment, recording in a fully-specified CRS. Even if someone
says you should use e.g.  EPSG:whatever WGS 84 (something) or the latest
CRS/RF  ITRF2014 (itrf2020 hasn't quite happened yet?) for exported OSM
points, understand that the points were input as WGS84(*any*). We may hope
each point was referenced to the latest realization of WGS at the time of
measurement, but no guarantees; and possibly measured with a retail GNSS
receiver or traced from open-license aero-photo aligned by eye - or
transformed from some local government stateplane by whatever WGS84
transform was handy.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.openstreetmap.org/pipermail/talk-us/attachments/20220718/bbc051a1/attachment-0001.htm>