Possible import: NGI refs for power pylons

Communities België / Belgique / Belgien (Belgium)
1

Hi,

I’m considering an import of power pylon references from IGN/NGI to OSM. Currently, as a data user, I’m using power lines from OSM, and power towers from NGI. This is because in NGI they (almost all) have a unique reference. This is very useful for field operators to match what they see on the field vs what is shown on the map. Of course this makes our data layer a bit messy, as the points are not exactly on the lines and the data is only available for BE.

Now that this data is finally open, it would be a nice test case for an import. However, the CC-BY SA 4.0 that they distribute it under is not enough, so we will have to ask for a special permission for use in OSM.

I would develop a script that matches power pylons from both datasets and prepares a few changesets to add the ref when there is none present in OSM. Then a second script to automatically change refs where there are no conflicts. And a third script to offer the rest for review in MapRoulette. Looking at the data, it also seems useful to review towers that are some distance away from each other in both datasets. I already spotted a few towers that were really badly placed in OSM.

There is just one “minor” issue… NGI systematically uses values like IW226 - 40, so first the line reference, then the pylon number. In OSM you see that too from time to time, as well as things like IW226 40 or 40 IW226. But the majority would just state the number 40.

For my purposes, the full string in NGI format would the most practical, as it summarizes all the info. This is also what the wiki seems to say: The reference of the tower as seen on ground., and I have the impression that usually the full identifier is on the tower. However, in theory the IW226 number is the reference of the line, so you could argue that you shouldn’t duplicate that on all the towers. Internationally, just having the most basic pylon number possible seems to be the more common approach.

Minor detail is that the vast majority of the lines in Belgium do not have this information present. So we would need a second exercise to validate line references too (and I would need an extra step in my data use process to extract the line info and add it to the tower - relatively simple in itself, except where a tower belongs to two different lines).

NGI data can be downloaded here: Top10Vector (it’s a BIG file, as it contains all of the top 10 vectors, I have not found a separate download link yet)

1 Like
2

From the chat:

  • 3481 power=tower with ref and 16611 without a ref
  • best to check geometric accuracy of NGI pylons a bit more
  • during manual steps of mapping, in Flanders, use GRB background for positioning
3

Why not make the things simple and unique for every potential data consumer? Define a new key such as ref:BE:ngi (to indicate that it’s a Belgium-specific register called NGI, see Key:ref#Country-specific, which already lists some Belgium-specific codes related with the De Lijn import). On import, assign the NGI codes to the pylons as specified in the original. Afterwards, you can examine and merge existing ref tags into the newly imported ref:BE:ngi.

1 Like
4

It’s an idea, but that would mean even more duplication. So you would maybe have ref=12 or even ref=L204 12, as well asref:BE:NGI=L204 12 on the tower as well asref=L204 on the line.
I would never drop ref=* after cleaning, because all data consumers expect to find this value on a tower.