I recently moved into a neighborhood that’s still under construction. About ⅔ of the neighborhood is built, but only half is visible in available aerial imagery (Esri seems to have the most up-to-date for here).
Since it’s new, I’ve been trying to add everything into OSM bit by bit. I’ve already added all the streets, and traced out the houses in the half of the neighborhood that’s available.
So, I’m wondering what I should do about the houses not yet available on satellite images. I walk the streets with my dogs every day, so I could easily add nodes for all the missing houses in the right spot from my phone. But I’d rather have all the houses as areas.
Could I add them as nodes, and then convert them to areas later once images become available? Does that approach even make sense?
If possible you should add a node with addr:* tags and building=*. If you can’t estimate or extrapolate building outlines I’d suggest to wait until aerial images or other sources (here in Germany it would be cadastre or land register) are available.
Hi,
personally, I would suggest trying to add an approximate area for each building, based on your observation, instead of just having a node.
Maybe also add a note or FIXME highlighting that the shape will need to be updated when the images are available.
Hi Dan, Osm is not all about copying areals, that is what other firms already do. What about surveying with the technology and knowledge of a cadastral member or surveyor and a GPS in your hand ? The visible left and right-side walls of a building could be easily visual stretched to the neighbouring streets so you’ll get the width of the building.
The depth will be an estimated value or common sense. Time consuming yes but the results could be there for years, so go ahead and contribute the objects to OSM.
Ps use a measuring line (8,50 to 20 m) to control your GPS as well to get a even better result.
Have Fun and Keep Mapping
I like this idea. I should be able to approximate the houses from the street while walking around. It’ll be slower and less accurate, but close enough to get them in the right place and roughly the right size. Then I’ll just add a note explaining that they’re approximations that should be improved when imagery is available.
Thanks for the suggestion. This should work pretty well.
I’ve only got my phone for this, but I’ve already added a bunch of roads that aren’t available in the imagery by just walking the streets (straight down the middle) and adding the nodes for the street as I walk.
There is nothing wrong with adding nodes with the information you know (addr:=, building=detached, etc.). That makes an immediate improvement in the map as OSM based navigation apps can now determine locations for routing and guidance.
Once a manageable source of building footprint data, often aerial imagery, becomes available those nodes can be turned into areas pretty easily. At least in JOSM, simply draw the building outline then select both the outline and the older node you put addr:- data on then do the “replace geometry” operation. There is a keyboard short cut for that operation so this goes really, really fast.
If you try to create the building outlines using on survey, say walking by and seeing what walls line up with what landmarks, my experience is the outline is not very good and ought to be replaced later when aerial imagery becomes available. And then you are in the same place about needing to draw the outline from scratch and do “replace geometry” or tediously update the old outline. I have done this in the past and have decided that it is just more work to do a rough outline than use a node and the benefit to OSM is marginal.
Yes, I always wait for at least the aerial images and then update if the cadastral is available. Makes it much more obvious where to look for updated background images.
It sounds like this might not be the best option, in that case. If I replace geometry from a node to an area, the edit history for the building would be on a single node of it, rather than the area itself. That seems less optimal, to me. What are your thoughts on that limitation?
What the edit history of the way will show is the history of way. It will be version 1 when you first create it. However one of the nodes on that way will be the original node that had the addr:, etc. information. The history on that node will show all edits on it (creation, changes, etc.) with the addr:, etc. tags being removed on the same change set that created the way that took on those tags.
Taking all into account, it depends on many different aspects and it is hard to tell which is the best process. As written above, a note=* or FIXME=* is appreciated for inaccurate areas.
In many situations, a node is more accurate than an approximate area. Areas might look nicer on the map but on the other hand they lead to wrong expectations. There are many reasons for breaks in history like splitting ways or converting an area into a multipolygon. You will get used to it.
Depending on the area, addresses might be best represented as node, atm. Buildings might have several addresses with individual entrances or several buildings have one address. In my area, the address is often connected to the parcel and not to the building. Currently the most important building on the parcel is often used for the address but that is inaccurate.
Regarding “Replace Geometry”, please, watch out, it is not perfect. E.g. #9929. Still it is the best we have, right now, and I do not know if other Editors, beside JOSM, have a similar action available. I fear they don’t.