Around 5 months ago, I worked on a small area of Dubai to improve the coverage in the area, changing a significant number of building outlines and adding tens of thousands of missing nodes, including missing pools, trees, fences, parks, paths, and increasing the detail of many models significantly through both satellite imagery and local knowledge from living there for over 10 years.
I recently went to check out of any changes were made recently to my data, and found that in quite a large area, which encompasses a solid 80% of my work, Every single node of type tree has been deleted. Some others nearby were kept untouched, but I’m talking thousands of nodes and entire months of work was wiped.
How can I know what person or bot caused this to report it and how can I revert this data, since it is unacceptable that all these hours of work get wiped. I spent way too much time and effort to make sure that all the data was precise and exact for it to be wiped clean randomly.
Any help is appreciated, I’ll gladly supply with more details if necessary.
Your changeset history is here. Can you identify something in one of these that you added that is no longer there?
If not can you identify an area where you added something and a time in the past when it would have been there (for example, just after you added it)? If you can, please zoom in to that area and provide the OpenStreetMap URL here.
Of course, here is the URL: (It’s quite a large area)
as an update, I have found the user that has deleted all the tree coverage data, and have sent them a message through OSM’s Direct Message section, waiting for a response from them
Of course. I compiled as many as I can, I believe there are a lot more that include the deletion of this data: #157995894 #158015306 #158039164 #158046677 #158047227 #158088126 #158086125 #158068446 #158065924 #158063816 #158053734 #158052637
This is just for a small area, the actual damage is much wider, there could be hundreds of different changes which all delete similar data in and around this area.
You can comment on other users’ changesets in order to discuss them. I’ve commented here suggesting that they might want to join the discussion here.
The reasons to remove the trees are that these trees:
Tend to evolve. The info barely will be maintained. Even the author personally will forget about it soon.
Useless if they all are just average, and not notable. One can name it as, for example, a managed forest or a park, it is much faster. No need to spend 100-200 hours.
Create noise and lags for navigation, editing, adding w/o providing any support. There were more trees than all the other elements in summary. Btw, they are partially located on the private territory, I doubt if any editor had an access to it.
Have nothing to do with the quality. Currently, in that area, there are/were missing walls, houses, buildings, paths, house numbers, crossings; there are/were walls marked as paths, false crossings, rows of houses marked as a single house, street names doubling. But there were thousands of trees.
I propose not to flood the map with this kind of trees info. It does not improve the quality, it reduces the quality. Especially if there are issues that are not fixed since 2017.
Welcome to the forums and thank you for joining this discussion.
These are all valid points but none allow for deleting valid data gathered by someone else without first talking to that person about it.
Everybody decides for themselves how to contribute. And each data consumer can decide on what data to include in their product. It is not up to you to make these decisions for these entities.
Please communicate first before you delete other people’s work.
I agree with you on the part of trees in areas between houses, which would of course change over time as home owners change and landscaping efforts edit houses. However, that can be said for every tree on earth, and so technically no trees should be mapped anywhere since any tree could be removed.
In that case, I would agree that those would not be necessary and looking back I would likely remove those that I placed in backyards and the likes. However, there are still thousands of trees which will continue to exist in those locations for the foreseeable future due to the landscaping requirements and standards in that area, which actually forbids any changes to the layouts of those trees, making them effectively permanent thanks to property managers being solely responsible for these trees. As such, that data should 100% be shown and will remain accurate for decades to come, which is why I believe that it should not be deleted.
So to compromise, I would meet in between by instead only representing trees that would definitely remain unchanged for the forceable future, which would also reduce lag on platforms that implement 3D assets for trees, like Mapbox’s Standard Demo.
The issue is as well that you just deleted so many trees that now means there exists some random lines where on one side trees are prominent and on the other side its just empty.
I will gladly make adjustments to improve performance, but ideally it there is a way to restore those trees without reverting any changes made in that area after that, that would be ideal, because I of course do not wish to remove your work either if a restoration is to happen.
This is a problem for all map data, but the solution is to encourage more people to contribute to OSM, not to delete data.
They are useful, they tell you where you will find shade, and what type of landscape to expect.
Any app or data consumer is free to filter out data that they are not interested in.
If a mapper chooses to map feature type A, they are not prohibited from doing so because feature type B is of poor quality. People can map what they want.
I’d add to that as well that I have put an effort to remodel and reposition a bunch of house models to make them accurately represent the data. this does take time though because there are a lot of variety to these houses.
I even went to the effort to colour code the walls and roofs and add roof shapes to the house models that I updated myself.
This does take a lot more time than mapping the trees though, so those will be updated eventually, but will likely take a lot more time for me to do.
Thanks everyone. I will first contact the author next time when I see discreancies.
Next.
I would like to insist of NOT restoring the trees in this area. The reasons are the same (expect for the #1):
N/A.
Useless if they all are just average, and not notable. One can name it as, for example, a managed forest or a park, it is much faster. No need to spend 100-200 hours.
Create noise and lags for navigation, editing, adding w/o providing any support. There were more trees than all the other elements in summary. Btw, they are partially located on the private territory, I doubt if any editor had an access to it.
Have nothing to do with the quality. Currently, in that area, there are/were missing walls, houses, buildings, paths, house numbers, crossings; there are/were walls marked as paths, false crossings, rows of houses marked as a single house, street names doubling. But there were thousands of trees.
They are useful, they tell you where you will find shade, and what type of landscape to expect.
Managed forest will work instead of mapping 10 trees.
If a mapper chooses to map feature type A, they are not prohibited from doing so because feature type B is of poor quality. People can map what they want.
Managed forest would not work for most these areas because they simply are NOT managed forests. Labelling private gardens or roadside trees as “gardens” is very misleading and incorrect.
Also, just to show another example, this is still present. Why can’t I have the trees that I also added restored since there are clearly examples of large parks and roadsides which have these trees mapped. Wouldn’t it be unfair and unjust to remove mine if there are already so many similar ones, all of which are undisputed?
Would you say that all these trees must also be removed?
To listen to the arguments on all sides and to follow the guidance of the balance of opinion in the thread? This might not match your personal view, but then OSM is not your personal map (nor is it mine or anyone else’s in this thread).
