As I said in the last of these values you commented, there was an import error for NVDB data at a period when lots of data was imported. The number is "correct", but has a tail of garbage decimals. I have rounded it to 4.8 now. If you see more of these you can safely just round it to nearest decimeter below. I think these are originally measured by laser and are originally centimeter precision but that's overkill.

the 4.73 meters is correct (although on centimeter is a bit overconfident), the tail is an import error from the source data that I've missed. There is likely a few more of these, if you see one more you can safely round it. The data source is high quality from Trafikverket here in Sweden, (and this particular crossing happens to be one I pass myself from time to time) but there was a while a quirk in the import conversion that added lots of dummy decimals.

I'll round the value to 4.7.

Jag antar detta är en skoterled. De ska inte kartläggas som unclassified (som är en mindre bilväg), om den inte går på just en bilväg. Om leden används endast som skoterled har den ingen highway-tagg alls.

Om det inte är en formell led borde det vara snowmobile=yes (annars designated), och över vattnet lägger man på ice_road=yes också.

Se här:
osm.wiki/Sweden/Snowmobile

Skoterleder visas inte på vanliga kartan på www.openstreetmap.org, för att se resultatet får man kika på exempelvis www.skoterleder.org när den hunnit uppdatera.

Jag har nu gjort revert på det som var fel-justerat men behöll delarna som var ok, se
changeset/146078475

Jag har nu gjort revert på det som var fel-justerat men behöll delarna som var ok, se
changeset/146078093

Även detta changeset verkar tillföra fler fel än det löser, troligen för att man ritat efter Bing som ligger fel och är lite gamla. Jag kan göra revert på det, eller så får man justera tillbaka det som blivit fel. Hur vill du att jag gör?

Det verkar som justeringen gjorts efter ojusterad Bing (Bing ligger ofta 10 - 40 meter fel) så det har blivit mer fel snarare än mindre. Trafikverket Road Network är en bra referens på de flesta ställen så flyttar man så vägarna hamnar bredvid det då gör man antagligen fel.

Det finns två sätt att göra, antingen gör man revert på hela changesetet (jag kan göra det åt dig om du vill), eller så justerar man tillbaka det som blivit fel. Hur vill du göra?

Thanks, my mistake, now fixed it.

Tack. Försöker vara noga, men gör miljoner uppdateringar så nåt litet fel kan krypa in ibland.

Redigeringen är en del av det här projektet:
osm.wiki/Import/Catalogue/Sweden_highway_import/Progress

Openstreetmap funkar som så att bara man rör något i geometrin så blir man registrerad som senast ändrare, även om man faktiskt inte ändrade någonting. När man går igenom en kommun (jag gick igenom hela Lycksele kommun februari 2021) så står man som redigare av typ allting i hela kommunen, men är det inte väg så är det antagligen bara att man flyttat existerande geometri hus etc som legat snett, men inte ändrat något annat. I NVDB-genomgången ingår inte postnummer, men som sagt man kan ha rört vid det. Hittar du något fel är det bara ändra direkt, du behöver inte kontakta mig. Eftersom jag jobbat ett några år med att synka upp norrland mot NVDB så har jag några miljoner ändringar på mitt konto, jag har inte koll på alla kan jag säga :-)

Ah kollade historyn, det var bara jag som rört place-noden senast, och tagit bort is_in, namnet var Helsingfors innan. Har hursomhelst korrigerat till Hälsingfors nu. (NVDB har inte place-noder, så ignorera min föregående kommentar. När jag väl gör en genomgång kan jag ibland lägga till eller justera place-noder vid behov.)

Fel data i NVDB alltså, händer ibland. Jag ändrar.

You probably know this already but satellite photos need accurate elevation data to be corrected, so anywhere where the ground isn't flat the correction offset needs changing. Lantmäteriet has accurate high resolution laser data to be able to correct their photos, but Bing probably have a quite low resolution model. ESRI have lower resolution images, but seems to have higher resolution elevation data so I find their imagery to be better corrected than Bing. (It is possible to get Lantmäteriet ortophotos in JOSM but the legality about that is still a bit unclear so I don't do that myself.)

NVDB can indeed have positional errors or small local geometry distortions (quite common where one road connects to another). If you have a Strava account you can enable a Strava heatmap layer in JOSM it's great for checking alignment if unsure. If one only has NVDB and Bing to go on concerning positional information, I'd choose NVDB as reference. If one goes to the data source at Trafikverket I think there is information about positional accuracy embedded in the data, but we don't transfer that into our OSM model for the moment.

There's some more info about NVDB limitations here:
osm.wiki/Import/Catalogue/Sweden_highway_import/Import_guidelines#Typiska_fel_i_NVDB-datat

Regarding the question, it's much better to align to the correct coordinates. When I started out with OSM mapping the accuracy of the satellite photos was much worse than it is today. And it will likely improve in the future. The goal should be to map with as high quality as we can with the sources we have, and if you have a better third party source, use that. It's likely that within a couple of years we will actually have Lantmäteriets ortophotos (not only the historic ones) in the web editor and JOSM as an official imagery layer as it's becoming open data too, and then the positional accuracy will increase dramatically. If the buildings are already correctly placed from the beginning a lot of work is saved. Of course there will always be a risk that someone with lesser access to high quality data sources will come in and "correct" the positions, but in my experience it's quite rare. People mostly map new things, correcting old (like we do in the NVDB import project) is quite rare.

No problem, and sorry if I was sounding a bit grumpy. It's great to get mapping help in northern Sweden, not many are doing work in the more sparsely populated areas. I'll look at reverting when I get back from work this evening, I'm not sure I can do it as a different user (never tried). (The OSM database doesn't really have a revert function as such, what happens is that a new change is applied to restore what was before, and I guess that part can be a bit wonky at times.)

Again this changeset destroys the alignment of roads. You can't just trust Bing to be correctly aligned. ESRI imagery is generally much better aligned in these areas, but you can't trust that 100% either. NVDB is pretty safe in moste places, but it's good to double check with Strava layer. Please revert the change completely, or the the changes that alters the roads, and move the houses into more correct positions.

What has happened here? Road geometry has been moved to incorrect positions. You can't trust Bing imagery to be correctly aligned. NVDB is way more accurate, and if you don't trust that, look at Strava tracks and compare. The whole road network of Arjeplog municipality was refreshed in the NVDB project in 2021: osm.wiki/Import/Catalogue/Sweden_highway_import/Progress

Please revert this whole change, or at least the parts that worsen alignment of roads.

Hello Mateusz, the NVDB project has been ongoing for a couple of years and is well known in the Swedish OSM community and has been discussed according to the OSM process for imports. You can contact me if you have any questions. Note that it is not an traditional automated import, ie it's manual work, all road segments are merged manually, and manually changed as needed. 100 segments takes about 1 hour manual work to go through, in a city with complex pre-existing mapping, 4 hours. One medium-sized municipality takes about 60 - 80 hour manual work to go through. By not having to click out every node the work is about 10 times faster than 100% manual work, and we get more tags like speed limits and such from the high quality government provided open database NVDB. This is necessary to increase the quality of OSM mapping in Sweden as we have many rural areas with very low density of local mappers and thus poor pre-existing quality.

Because of the highly manual process, no special import accounts are used, just our normal accounts.

You can see the current progress here:
osm.wiki/Import/Catalogue/Sweden_highway_import/Progress

ESRI är förvånansvärt snabba ibland, LM ortofoto vinner nog oftast, men har flera gånger varit med om att ESRI har nyare foton. Gissar att det händer oftare ute i skogen där jag tror LM inte uppdaterar lika ofta som i tätbebyggt område.

Den är dock i Härnösands kommun så har inte jobbat igenom hela, vägarna började i Kramfors (som jag jobbar med) så det spall över kommungränsen. NVDB har fortfarande utdaterat data i området, men om nåt år borde det finnas fräscht data

Ja precis, det är en vindkraftspark under uppbyggnad, se ESRI-flygfotona.