Bing gets updated, but at a much slower pace and recently with images from Digital Globe, some might be identical to the Maxar ones. The comparison on imgur (you made this ?) is just a spot check. Difference between the different imagery sources range up to 20m, I have plenty of examples for that, best one here (major highway offset and overlap with duplicated power poles etc...) (changeset/104005044#map=19/0.08762/32.48671). From one tile to the next there is a jump of 10.5m N-S ! The blurred overlap consists of over 25m E-W. You can't veify this with Esri or Bing, since the imagery in both is so outdated, I am using this stretch of highway on a daily basis for 3 years. Another problem is stretching of imagery on slopes. Also Maxar does a terrible job. Even if you add a calibration point you need to re-allign like every 100m because you notice singular buildings might differ in length and width up to 2m, which is significant in a country with many small buildings. Bing should not be banned, due to it's high resolution it's often a good reference for details hardly recognisable on other imagery, like power poles. My experience with HOTOSM and replies is different. They might not react immediately through comments or feedback on the HOT website, you know our culture is different. People keep quiet when critisied, especially when it comes from the West, the "bosses" on HOT also. A good example is the comment regarding the imagery source on one of the new cities project last week which you left there. It was not noticed even, we don't have the resources to do that. The PM left a reply for you after I noticed him. Our community is very close, rather small , so we use social media as WhatsApp and Telegram, also when it comes to OSM. I am not affiliated with any of them however I am administrator for local guidelines in a limited workgroup. It works fine, we have weekly meetings and physically meet on a regular basis.

Reverted because of deletions of urban trees, not justified. Most changes recovered like correct use of wikipedia tag. Squared some buildings and resolved overlaps with landuse.

Too much deletions, not justified according your source. No consent from the local community. Please revert , don't delete but correct and use lifecycle tagging for demolished buildings.

In this case true, agreed, much appreciated improvement.

Jean-Marc, another kind request, please stop the deletions. I checked this changeset and although some are justified, as trucks on truck parkings should not be mapped as buildings, you keep on deleting poor quality buildings, especially from HOTOSM projects, some of them still ongoing and waiting for validation. You delete like whole city blocks, that is not OK. Correct misalignments, use lifecycle prefixes on buildings please. Many of the users whose data you deleted are well known and valued contributors in our community, very willing to work on improving the quality of our map. Also avoid making such large changesets, as we lack the resources to assist and it's eating all our data. Please start reverting the deletions you made the past 3 weeks or so, it's too much. Mass deletions without community consent is vandalism.

Hello Jean-Marc. Thank you for mapping a missing wetland in Uganda. Kindly make some improvements so we keep our wetlands data significant and of high quality. The wetland should intially be mapped as a boundary, a protected_area with protect_class=15. The reason is that all of them are government land and protected by law. Only apply natural=wetland to this boundary relation. Our wetlands are not uniform, but a mixture of papayrus reedbeds, mudplains, bogs and marshes with some pools and often ponds along the shores. These can be mapped as separate areas, on top of the multipolygon providing the wetland type. Swamps, being wetland mostly covered in trees are not that common, especially not those covering the whole wetland. So it's better to remove the wetland=swamp form the multipolygon or relation, the rendering will be fine. The added details about the biotopes are defined as inner areas on these relations, until we cover the whole wetland. When the whole wetland is covered the natural=wetland tag can be removed from the boundary relation, leaving us with just the boundary which allows neutral and proper handling of land conflicts. Be carefull at the mapping the boundaries. All wetland=* tagged areas should be completely contained within the boundary. Other natural=* areas , especially those overlapping the boundaries should be not. Those completely in the wetland boundaries should. You also are able to map landuse=* this way. Important is that the boundaries are not connected to any feature, this allos flexibility to change them and map them appropiate according field surveys, most wetlands boundaries cannot be mapped accurately from satellite imagery. If you need any additional information please do ask or refer to WikiProject Uganda guidelines and training (osm.wiki/WikiProject_Uganda/Conventions-Categories still under construction but a good starting point to get information about local guidelines, mapping practices, quality of the map and get in contact with the local community). Greetings, Bert Araali

Alpine_huts do exist n the Ugandan context, but in the mountains, like Mount Elgon National Park and the Rwenzori's. In this context, a suburb of Kampala, it is obvious this cannot be the case. Please explain what you were trying to map so we can help. Also avoid all capitals in the names, small letters correctly capitalised is good practice. Anyway, welcome to OSM and thank you for improving our map.

The tea factory at first glance doesn't seem to be a mistake, on the most recent sattelite imagery it seems demolished, I assume becuase it was build on government land in the wetland. I live only a few kilometers away from it so will go there and check one of the coming days. I got the misalignments covered, your work is not lost although I had to revert to recover the deleted buildings and a subset reverting was not feasible. Please stop the deletions, it is not OK in OSM, nowhere. What you call false mapping is vary rare. You might come along misaligned or even worse non squared buildnigs, all added with good intentions by unexperienced mappers. First ask the originator to change it, assist and help to imporve the quality. Actual buildings no longer there, like the tea factory but as I've seen most of the others you deleted should be tagged with lifecycle tags, like razed: or demolished: . Mass deletions are vandalism, no matter the quality of what was mapped before. Changeset comments are a good way to go, however you might get faster and better response on local forums and social media. With the changeset comments my experience is not that bad, I mostly get replies up to 50%, try to be constructive, not destructive, both in mapping practice as in language works best. If you are interested I can send you invites for the local social media groups where you can get much more constructive discussions.

Our experience is the contrary, Bing in most cases is more accurate then Maxar or Esri. Especially Maxar is troublesome, with large misalignments and semi-transparant overlaps between tiles, f.i. here, gaps between tiles, and tiles captured at different dates at different zoom levels (f.i. around Mount Elgon) ! Keep in mind that Maxar is very inaccurate when it comes to mountainous areas, with offset variations up to 15m in short range like 5km. Referencing the OSM GPS layer is good practice, the Imagery Offset Database even better.

They are in conflict with one basic Good Practice of OSM: One feature one tag. I don't care so much about renderers, sure they can be creative and anticipate on incorrect mapping. For other data consumers this is a huge problem though, they find overlapping identical landuse and are not able to decide what to do, ambiguity and duplicate data doesn't belong in OSM. What you promote as a "standard" is in your local context or areas where you was active so far. It surely isn't true for OSM worldwide, the fine grade practice is evenly favoured and getting more support as the map, worldwide gets more detailed. I do agree we are suffering major quality issues with our mapping in UG, however you need to give a chance for the emerging, mostly younger community, to catch up, respecting our limited resources and limited "volunteer" appetite. It doesn't work that way here but we are working on it. I would applaud support in this process instead of reinforcing armchair mapping by people who don't know our culture, our local social issues or don't take the effort to consult the local community or our local guidelines.

No, I mean a calibrated area stored in the Imagery Offset Database (osm.wiki/Imagery_Offset_Database). The calibration was done with multiple GPS tracks, recorded in the field and Mapillary tracks. I didn't use Strava in this case because we had lots of GPS data. We are investigating the possible use and mapping of Geodetic markers. The system was recently completely reinstated and updated in Uganda (I believe in 2018) however most of the markers are not visible on satellite imagery and the Offset database does not allow point data. So our local workgroup has the updated geodetic markers available from some land surveyors in our group and we need to see how we can integrate the information, probably creating accurate GPS data from the markers to some nearby buildings or better larger areas, but that is a long term process. Currently we stick with the Imagery database since most of the data in the past was added by HOTOSM teams, assuming BING as the golden standard, which it is not. So it will be a long process but we will get there, don't rush, do it once and do it good but KISS (keep it simple and...). Important to know, but disregarded because "unknown" is that ALL satellite imagery is misalligned, especially the alignment in the Maxar imagery is of very low quality. This causes major problems since it is the most recent and up to date. Unfortunetaly the Imagery offset database is not supported in iD, so our advise, until we have the whole country covered, is that those mappers should align with existing items or if no mapping is done in the area yet, check GPS data, if that is missing align according to Bing. Never align the map to match your imagery, that is wrong and leads to large scale damage, in doubt contact our local community through the forum or social media groups. Greetings, Bert Araali

Hello, unfortunately I needed to revert this changeset:
- building deletions based on incorrect source reference. We concluded this because you deleted multiple buildings from a tea factory, which is still shown in Esri imagery and BING, back to 2019, however not on Maxar, dated 2020. So references to Esri as your (only) source didn't allow me to reference justifications of any "false" buildings.
- alignment changes according imagery, local GPS tracks (stored and uploaded to OSM) were not verified , you should align imagery with the map, not the map to your imagery. True the previous mapping showed misalignment with the same GPS track, so for your convenience and future users we added local calibration points in JOSM based on these GPS tracks and Mapillary GPS data.
- deletions without consultation of the previous editors or local community can be regarded as vandalism. Instead please contact them through this medium or any of the forums, we are there and willing to assist. Only delete incorrect items or vandalism, obsolete or demolished items should NOT be deleted but a lifecycle tag added so the history and justification is verifiable for all.

I wouldn't call the lack of landuse tagging in Kampala and UG at large troubling. It's a next step in establishing more details on base maps, the local community with the help of other OSM users and groups is working hard to improve it. Groups include the UN mappers, HOTOSM, youthmappers and a local Workgroup to document the local mapping and tagging practices, make appropriate field checks if in doubt. This with support and grants from the OSMF and other donors. Super-relations are hard if not impossible to maintain on the long term by the average user, so the local community decided to avoid them as we have very bad experience like with f.i. Lake Victoria. The reason why this relation is troublesome is primarily the fact that it is overwriting thousands of already mapped landuses in the area and does not comply with the good mapping guidelines, as I said before, landuse should not be used to map areas as big as an "agglomerate". Mass or edits with a large impact like this one should always be made in consensus with the local community. So did you contact the local community, have consensus ?

Yes, I do mean relation/12571016. You focus on single phrases in a guideline. The wiki is a guideline, not a rule or necessarily reflecting "common use" of OSM. Landuse is used differently across the world, in some countries on a macro scale, in some on a mirco, as you might say on a plot by plot basis if it makes sense. You should follow the local practices and contact the local community before undertaking such large edits. Local guidelines are developed on an ongoing basis for this purpose and the majority of the countries worldwide does not implement landuse on this scale. The examples you refer to are common use outside of Uganda, even outside of Africa, and at least in UG considered as not favourable for our local needs and reality. The wiki, besides the isolated sentence you refer to says also "The extent of the area should mark the boundary of known residential use, not the extent of the whole town or village." What you are mapping is not just the extend of our Capital, Kampala, but even trying to include areas outside the capital district boundaries, into Wakiso district, which neither is managed, belongs too or managed in the same way as the capital. Kampala, meaning the district as it is managed by the city authority KCCA, has so far 588 residential areas mapped, Wakiso district 10912 ! The whole Kampala district could be considered a residential area if I try to follow your reasoning and mapping, we already have a nice administrative boundary relation for that. Your relation tries to overrule this ? Besides this, it is absolutely wrong, locally to align landuse=residential with the boundaries of natural features like wetlands. Encroaching of wetlands is a common detrimental practice in UG. Wetlands are mapped according to the situation on the ground, which in many cases cannot be determined from satellite imagery. All wetlands are community, government owned land. This situation changes rapidly. So we map the encorachers on a neutral basis, adding the buildings and the landuse on an individual basis within the wetlands. So they overlap with the wetlands. Landuse overlaps with natural (or landcover as you like) allowing data consumers to analyse the extend of the mispractices but also retaining a neutral position as a mapper. People are getting killed because of this practice. Another major reason why in UG we don't map landuse on the scale which might be favourable in Western countries.

Hello Jean-Marc, what is the intend of this super-relation ? It conflicts with the local situation on the ground, local mapping practices and I fear it is going to increase the already heated land conflicts in our Capital. Kampla, neither any other Ugandan city is a unified residential zone. We are working for years to map the landuse in the city as neutral as possible, on a detailed, often micro scale. It's like you want to map all of Paris and part of it's suburbs as one huge residential zone. Which is obviously not the case. This relation is going to interfere with all projects and data we offer to various data consumers, overlaps and duplicates already intensively mapped landuse and goes against the practices how the local community maps. So for now, I removed the landuse=residential tag from this relation and will be happy to assist in any philosophy or intend to find appropriate tags for the large job you have done, with the consent of the local community of course.

OK, I was commenting on possible changes in Uganda, as I couldn't find any there. I'm a local and we recently face serious issues with imports.
Service roads I can only agree with the approach, however service roads are interpreted slightly different across the world. In general, a service road can be any road or track not used for general public access. In most cases this means, as you stated that in most cases it branches from a main road or track. This way is in Brasil, so I leave it to the local community there to comment. GPS and Strava are for sure reliable sources for alignment. Also there I made my comment in regard to possible changes in Uganda, where we are since recent making additions of calibration points and offsets in the offsets database. We noticed significant differences between tiles of satellite imagery, especially in mountainous areas and checked those on the ground. But as you say, GPS and Strava are for sure reliable alternative sources. Thank you for replying.

True, the duplicates were created by the multiple imports subsequent to your edit. Most of it was repaired by reverting in this changeset: changeset/101462773#map=13/-0.7592/30.8043. I did include changesets by your team (mainly because of mis-allignment, unattached highways, duplicates) because they fitted in the recovery for this area. This was so large scale it wasn't feasible to trace the history of every problem. So sorry, the duplicates (and unattached highways) where not created by your edit..
Consistent hashes for the Apple Data Team seem to have improved, thank you to the whole team for that.

Dear, please don't do this. You realigned Victoria shores, village boundaries and thousands of features en mass. It's not the mapped features that are wrong, they are all alligned correct with years of field work by the local community. So please revert your changes or ask me to do so. Never do this kind of mass edits without contacting the local community. Regards, Bert Araali

Dear, you are editing to large areas to be managable. You should not edit alignments to better match arial imagery. It's the opposite. The aerial imagery must be alligned to exitsing features, especially when they are traced by local mappers who checked ground truth. Service roads to multiple houses are not be defintion wrong, often they go to larger private properties, even compounds which you could consider as separate villages, it's a wrong interpretation of the highway=service tag from your side. You should always contact the local mapper to decide if it is wrong or not. So please make changesets that only cover smaller geographical areas and stop realligning the map with your imagery. Add what is missing but don't change without consensus with local OSM community.

Hi, thank you for the response. I noticed the Apple team is using the different hashtags more consistently. Good improvement. Thank you.