NGED Import Project/Discord Discussion

[5:27 PM]Kits [UK-WLS]: Hi all, i'm considering importing Uk 11kv substations data. The data i got is compatable with OSM license and from my testing the locations appear to be very accurate. the worst miss i've seen so far is about 7m. most are accurate to 2m tho. the data is seperated into OS grid squares.

[5:29 PM]Kits [UK-WLS]: i'm not sure the correct procedure for these kind of import so would appreciate any input.

[5:34 PM]jmarchon [US-OR]: Have you already seen this wiki page? https://wiki.openstreetmap.org/wiki/Import Import

[5:34 PM]jmarchon [US-OR]: I assume so, but just in case I'm posting it here

[6:06 PM]Kits [UK-WLS]: thats looks like a lot of effort i instantly lost all moral

[7:32 PM]watmildon [US-WA]: How many locations are we talking about? It's often easier/faster to do a "manual review and conflation" import in JOSM rather than a blind upload which would require the full review. Would also let you fix the worst offenders for being off on distance.

[7:32 PM]watmildon [US-WA]: You can get though a LOT of data review pretty quickly in JOSM. Way more than you would initially expect.

[8:11 PM]Kits [UK-WLS]: it is a lot. i have been adding them manually. thats now i know their accuracy is rather good. let me get you a number, one sec

[8:13 PM]Kits [UK-WLS]: opening them all takes a age so not gonna do that atm but one rural grid square has 1109 and i believe there is something like 200 grid squares total

[8:15 PM]Kits [UK-WLS]: but i was also think about adding other related data, all the data for 11kv underground and overhead cables is also available but this is not as simple since all the lines are seperated and in osm ideally paraell lines should be bundled into a single way with circuits=X tag

[8:17 PM]Kits [UK-WLS]: https://openinframap.org/#13.71/51.75488/-3.2641/L,P,S these 11kv lines and substations are what i've done manually so far for reference Open Infrastructure Map Map of the world's electricity, telecoms, oil, and gas infrastructure, using data from OpenStreetMap

[8:22 PM]Kits [UK-WLS]: I suppose if its acceptable i dont think its unrealistic to manually import one grid, cables and all and do all the tweaking like combining multiple lines into a single way and such. i presume this is what your implying

[8:27 PM]Kits [UK-WLS]: i guess that would work, its not far off what i'm been doing already

[9:04 PM]watmildon [US-WA]: Yeah. It’s usually faster than you expect and most of the fixup you find while doing it SOMEONE would have to do eventually.

[12:00 AM]jmarchon [US-OR]: Yeah that is a bit of a problem...

[12:59 AM]APP will: There is definitely a process but it's a little less awful than it sounds especially if the license is definitely OSM compatible. The remaining issue the reviewers will want to know about is your plan to clean up the data to OSM standards, and merge it with existing data without causing problems. As you've pointed out there's a few issues right away that need figuring out (which they may be happy to help with, many Importers have some great scripts and techniques!) regardless of any formalized process.One suggestion I have is, it sounds like the data may be far more detailed than you need. You could consider running some script that perhaps removes all intermediary poles and just maps corners. Or a script that creates one line between poles instead of many. Or even if that's too messy maybe a MapRoulette challenge that says "is there a power pole here connected to poles Here and Here?" for slower human-centric creation and verification.In my import experience often it comes down to human judgment anyway when you've got a cleaned file but an existing pole that needs conflation. Computers can't always tell "oh there's actually multiple poles here" or "this was all incorrect and it's actually like this" so aggressive "80/20" logic where you try and get as close as possible but still have human eyes on the edge cases can be a good approach even if it's not as quick as we'd like

[8:18 AM]Kits [UK-WLS]: Thanks for the input. I got a group motorcycle ride today so I'm not gonna touch it and just think today. Atm from my experience manually it all, the slowest part is just adding the transformer information. So what I might do is import all the transformers,substations and possibly power poles then trace over the lines manually to correct offset and group them correctly. Out of all the data the power poles tend to drift the most so I've seen one be a good 30m off. So I may skip that data set and just look for them on satalite as I'm tracing. As I say I'll be giving it more thought today

[5:19 PM]Hidoo [SE]: Please don’t skip any poles if they’re accurate in some places! They are quite difficult to spot in forests from imagery, so it would be a big help having them there at least