[Imports] Thailand Land Use Import

[Mateusz Konieczny]

You mention some discussion with a local community - can you link it (all of them)?

Important note: if local community is happy about such data quality then feel free to
 disregard my opinions about data quality, discouraging mappers and usefullnes
of such imports.

If any part of local community is unhappy about data quality: please finish reverts
and removed already imported data.

Feb 13, 2021, 08:53 by nitinatsangsit at gmail.com:

> Hello
>
> I was working on land use classification with a semi-automatic process done by SCP in QGIS with Sentinel imagery. I thought it could be useful so I uploaded it to OSM, but after discuss with community it should be considered as import 
>
Importing external, automatically generated data (without supervision or review) into OSM 
is definitely an import.

> so I have to revert it then and before any upload.
> My work is aim to make land use available in OSM in Thailand. 
>
> Although it is not that much precise, it could be better than leave the area empty. 
>
I strongly disagree. Low quality geometries such as
https://www.openstreetmap.org/edit?editor=id&changeset=97220712#map=18/13.52826/100.40683
are worse than nothing as it discourages real mapping (people are much more likely
to map something rather than cleanup low-quality output of automated image
recognition). See https://i.imgur.com/SYYsWke.png for a copy and representative example
of quality.

Though if local community has consensus that importing something like that is OK
then it is fine to do it.

Also, it appears that you deleted for example https://www.openstreetmap.org/way/390010502/history
https://www.openstreetmap.org/way/387296360/history
- have you mass deleted what was mapped previously as part of your import?
With changeset comment "Add landuse around Bangkok"?

(I am unable to see what was really changed, and whatever deleted mapping was valuable.
Due to size of changeset usual tools failed to work - https://osmcha.org/changesets/95728109/
failed to load.)

> However, this process may lead to some geometry problem, so I will upload only large polygon (more than 10,000 square meters) to avoid small geometry which may look weird.
>
Is quality of large ones better, or is it just less obvious that data quality is low?

>
> To deal with error that may occur, I will fix all self-intersect of geometry, duplicated of pre-mapped things, water features which overlap with highway, etc. to make sure it would not affect any navigation.
>
> The example of my work is around > https://www.openstreetmap.org/changeset/95728073>  (small area and overlap water haven't been fixed yet)
>
Data quality seems to be atrociously low, see for example https://i.imgur.com/R4wbUnI.png

Fixing it would take more effort than mapping it manually.

> I haven't revert it yet because it could be useful to be an example.
>
Please, revert it and present proposed data as upload in format that is not 
harming current OSM data (publish .osm file for viewing in JOSM, geojson file,
images with representative examples - should be possible with QGIS)

>  If it could be accepted, I will create the wiki page, revert it and carefully re-upload it via dedicated user.
>
Obviously, opinion of a local community is the most important.

Have you told them that you are seeking approval for making an import?
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.openstreetmap.org/pipermail/imports/attachments/20210213/d5dec66e/attachment.htm>