OpenStreetMap

What's the story behind these disconnected streets in Lima?

Posted by lxbarth on 2 May 2012 in English (English)

I'm working with Rub21 right now on improving Lima. We're finding a lot of disconnected streets there. What's the story behind these? I'm seeing that at least some of the remaining streets are by user Telecom IP - I'll be sending him a message. Not sure whether this is license change related, I remember seen these disconnected roads in Lima well before the bots kicked in.

Right now our plan of action is to repair the streets that match nicely with Bing satellite imagery and not touch those that don't. I'd like to eventually repair or remove the ones that appear to be garbage, too.

Location: Calle Las Cruces, Lurigancho, Lima, Peru

Comment from pieleric on 2 May 2012 at 20:51

Hei, I've been working a bit on this (on and off since a year), as a couple of other guys. Actually the Telcom IP import covers the whole Peru. It has many bad sides (e.g., low quality of positioning, missing segments of roads, same road represented by many segments, roads without name named "S/N", names all in upppercase), but it has one big advantage: it has the name of many streets.

So now what I usually do is: 1. map using the satellite imagery 2. name the street using the name of the close by segment from Telcom IP 3. delete the Telcom IP segment

Usually, I leave the name in uppercase, to mark it's coming from Telcom IP. I guess later on, when everything is converted, it will be possible to fix it with a bot.

For Lima precisely, I can give you a couple more advises: * Some places look like already mapped, but the Telcom IP and the satellite imagery haven't been merged (step 2 and 3). It looks ok from a low zoom, but if you zoom in (or use JOSM validator) you can detect that case. * The bing imagery has quite a big shift (~20m), so you want to align it first on existing roads/gps tracks.

Have fun & good luck! √Čric

Hide this comment

Comment from giggls on 3 May 2012 at 08:58

This is an example of a very bad import. Just get rid of that stuff and replace it by local knowledge.

Hide this comment

Comment from compdude on 4 May 2012 at 03:50

That's even worse than the TIGER data that was imported in the US! But the TIGER import, even though it was very inaccurate, was actually connected to all adjacent ways. And because it included names (which were actually in lower case), it was worth it because without the TIGER data, the US would be largely incomplete.

Here, you're only going to be able to salvage the street names from those ways and you might as well redraw the way completely.

Hide this comment

Comment from RM87 on 4 May 2012 at 19:28

I have done some Telcom IP remapping from which I can say that this area has very very few gps tracks. As it is half impossible to align bing aerials without any tracks then there is no easy way for non local mappers to help with initial road network.

Hide this comment

Comment from bri g on 10 May 2012 at 23:44

Quote

'As it is half impossible to align bing aerials without any tracks'

I have no disagreement with this statement at all, may I suggest though that mapping using aerial imagery, even if it is misaligned to absolute co-ordinates is not a waste of time.

In the event that you are newly mapping an area with imagery, but no traces, the whole map could be offset, this is quite true.

You work is not wasted I believe, as within an area, all the roads will be offset by the same amount, and it will just require one upload from an OSM contributor to allow correct alignment the map to the traces, a trivial operation.

So my advice, map to BING imagery, correction and of course all the rich data and corrections that can only be added by people who visit the area, will surely follow.

bri

Hide this comment

Comment from lxbarth on 13 June 2012 at 14:41

Update: Rub21 has done a lot of work in cleaning up Lima and got it to a pretty good point. We got awesome help by WernerP

Here's a rough viz showing the edits in the past month:

Hide this comment

Leave a comment

Parsed with Markdown

  • Headings

    # Heading
    ## Subheading

  • Unordered list

    * First item
    * Second item

  • Ordered list

    1. First item
    2. Second item

  • Link

    [Text](URL)
  • Image

    ![Alt text](URL)

Login to leave a comment