I appreciate your effort. But I think this misses my main point: These imports could be destroying valuable manually added data across Norway. Engerdal is only interesting because I have very good and recent local knowledge, and thus can use this as a sample of the overall quality of these imports. My conclusions are still:
- While this import may be better than the previous import, it is not better than mapping based on surveys and/or orthophotos and DEM.
- The import process appears to indiscriminantly overwrite high quality mapping. Not only the previous low quality import.
- It is appreciated that you try to correct mistakes as they are pointed out. But for large scale imports you cannot expect other people to catch mistakes. The import process must itself make sure existing data is not lost.
What about the (only) two concrete cases I mentioned? Here it was obvious to me that the imported streams were not aligned with neither orthophotos, DEM, nor hand mapped bridges. Did you move those bridges to match the imported data, and thus contradict the mentioned primary sources?
Maybe you could provide some concrete examples where you’ve moved bridges or where bridges and imported streams matched well already. That would make a factual discussion easier.
Surprised we disagree here!
I’m glad it seems we both have a goal of representing whats on the ground as best as possible. It seems somehow we have different opinions of how to best represent whats on the ground, or of how to evaluate this quality.
I base my estimation of the quality on having been to this particular “bakke” taking notes and photos as I walked there, as well as on the most recent orthophotos and terrain and surface lidar data (DTM/DOM) available in JOSM. As an example the very clear glades I observed and mapped appear to be missing in the imported forest. And obviously the tags detailing the type of wood has been deleted entirely.
What do you base your estimation of what matches reality on? Have you been to the place?
This much I understood already. I agree that 1:50k is a course scale. As far as I can see geometries are smoothed for ease of reading when rendered at that particular scale, thus not optimized for representational accuracy. My objection is that the import is NOT used as a starting point for refinement. On the contrary! It is used to replace already refined data!
No I have not. And I don’t think I need to in order to provide my input.
Good data is being lost by doing the imports! It seems fairly pointless to destroy the unique data of OSM, replacing it by a copy of less detailed data that is freely available elsewhere to anyone interested.
But ok, I checked the wikipage of the import. It says it represents consensus from a mailing list. Scrolling back in time on that mailing list, the most recent semirelevant post I found was one from May 2020 complaining that Import ødelegger… I don’t doubt that consensus was reached at some point, but maybe things have changes since then? At least in Denmark ~3 years of mapping changes a lot.
Look, I don’t dispute that the data of this import is better than the last bad import… As I mentioned in my first post, Norway is a big place, and imports may be fine as a starting point for further mapping (as you pointed out).
But please don’t delete other peoples high quality mapping, and replace it with data that is tweaked for rendering at 1:50k that is also freely available somewhere else.
This seems like a fairly silly thing to say… The topic of this thread is “Details removed by import?”. I don’t do imports… So, I’m not discussing my edits. I’m discussing detail-reducing imports in Engerdal (as a sample of all N50 imports). Or did I get the Norwegian possessive pronouns wrong? That would of course be embarrassing…