[Imports] Thimphu City data import

Rafael Avila Coya ravilacoya at gmail.com
Thu Jul 7 15:24:09 UTC 2016


Hi, Christoph:

Thank you for your time and input. My comments inline:

On 07/07/16 15:40, Christoph Hormann wrote:
> On Saturday 02 July 2016, Rafael Avila Coya wrote:
>>
>> As there isn't any local OSM community yet, we contacted the HOT
>> mailing list for opinions.
>
> This statement astonishes me - if there is no organized local community
> but obviously locally made mapping you should try and contact mappers
> active in the area for their opinion.
>
> Note also Bhutan is a fairly popular tourist destination and there is
> likely quite a bit of local mapping done by visitors.  Even if you are
> not able to contact any native local mappers for their opinion you can
> certainly try to get in touch with people who have mapped in country
> there and who might be able to make a more competent assessment if such
> import makes sense or not.

Do I need to contact local mappers to trace buildings with aerial 
imagery. If so, I guess most of mapping in many African and Asian 
countries should be reverted. I refer to my answer to Frederik's email 
on how this import (data integration) will be done, and why it is 
respectful of local communities.

>
> Based on a quick look:
>
> - you seem to have among 13610 buildings 1531+2579 obvious errors.  This
> indicates poor quality original data and without any possibility for
> ground verification i think this disqualifies the data from being
> imported in OSM.

I don't think it disqualifies the data in any way whatsoever. It would 
disqualify it for an automatic or semi-automatic import. But building 
footprints will be imported ***manually*** through a project of the 
Tasking Manager, by OSM community. They will align them, and when there 
are two footprints for the same building, they will ignore (delete) the 
worse of the two. It's like mapping buildings with aerial imagery, but 
with a better quality and much easier and faster. Win-win.

> - you seem to intend running Douglas-Peucker on roads but do not provide
> any explanation as to why you think this is a good idea, let alone any
> assessment if the results meet any kind of quality criterion.

I had to go to Google to know what you mean with Douglas-Peucker. People 
will understand you better if you say simply "using simplify way tool".

Anyway, the reason is simple: We reduce the number of nodes. We did it 
for the Tanzania road import, and everybody was happy, to no surprise.

> - The road data lacks any highway tags and largely lacks proper
> connectivity between road segments.

As they will be imported manually, one by one, there isn't any problem 
with that. They will be joined, corrected, ignored, etc, on a one-by-one 
basis.

> - The house numbers are obviously not unique city-wide but you don't
> have any tag except addr:housenumber on the address nodes.

But in many cases it's very simple to see what street they belong to. 
They will be added their add:street. But this is part of the second 
import, and we will be in contact with the people in Bhutan (both locals 
and the Nepalese trainers), so we can assess better the quality of the data.

>
> So in short apparently:
>
> - you have not contacted anyone who has mapped in Bhutan locally

Explained.

> - you have no idea on the local road tagging practice

There isn't local road tagging practice for Bhutan. In fact, there isn't 
local road tagging practices for many countries in the world, and yet 
many people (like me, proudly) map thousands of roads of every type 
everywhere.

But there is a wiki that was written after lots of opinions and 
discussions, that fits very well for most developing countries: the 
Highway Tag Africa wiki [1].

> - you have no knowledge of the local address system

Explained. We aren't starting that second part of the import before 
local community give their opinion, hopefully soon.

> - you have no idea of the existing data in OSM in the area

And you say that, and you are happy. Great.

>
> yet you want to import data there.

I won't import data. It will be the OSM community who will. Quite a 
difference.

>
> Maybe it would be a good idea to have that group of Kathmandu Living
> Labs who are going to work with the local people to find out what is
> necessary to make a proper assessment of the data as part of this work,
> discuss the import possibility with the locals and then - if the
> reaction is positive - manage the import with a solid knowledge of the
> matter.
>
> As it is i can only say i disapprove of this import plan.

I am not surprised of your disapproval.

Cheers,

Rafael.

>

[1] http://wiki.openstreetmap.org/wiki/Highway_Tag_Africa



More information about the Imports mailing list