[Imports] [OSM-talk] ADD import (Antarctica natural features and POIs)

[Christoph Hormann]

On Tuesday 14 May 2013, Paul Norman wrote:
>
> You are proposing ADD:code, ADD:date, ADD:text, ADD:source, and
> ADD:revision.
>
> The current wisdom is to not include these type of tags.
>
> The code duplicates the other OSM tags and is wrong as soon as
> someone reclassifies a feature.
>
> Date, text and revision are not documented on the wiki page so it's
> hard to see what purpose they serve.
>
> ADD:source would be better off as a source tag.

Hello Paul,

thanks for your suggestions.  Just for clarification of the meaning:

- date is the date of the original data source (i.e. satellite image 
date), often only the year
- revision is the date of the last modification
- code is a feature classification in the ADD
- source is the original source, sometimes quite detailed, like a 
Landsat scene number
- text contains further notes on the techniques used, reliability, etc.

I agree code is redundant, i originally kept it because there are a few 
errors in the polygon data, especially shelf ice areas wrongly 
attributed as land while the line data is correct.  In those cases the 
line attributes would help fixing those.  In hindsight though these 
errors can be quite reliably fixed without this information just by 
careful inspection.

As for the others, i could imagine the following:

ADD:code -> (discard)
ADD:text -> note
ADD:revision -> source:date
ADD:date -> survey:date
ADD:source=* -> source=ADD;*

I would very much like to keep the original date somewhere since this 
could simplify future updates.  Especially with the spreading use of 
LIMA in web maps (both Bing and Mapbox have added it recently) it would 
be important for mappers to have a way to decide which parts should not 
be 'improved' based on outdated images.

> ADD:temporal and ADD:status seem unnecessary given the information is
> in the seasonal and abandoned tags.

All right, looking through it again it seems the only additional useful 
information in those fields is an occasional 'temporarily closed' which 
would translate well into 'disused=yes'.

I will also separate the POIs from the rest since they will need to be 
carefully merged with existing data which can better be done separately 
i suppose.

> Some of the ways could really use a simplify. I'm not sure what the
> appropriate value is, but there are some runs of 20 nodes in 13b a
> straight line which simplify down to 2 nodes. The grounding lines
> seem the most prone to this.

Yes, the problem is simplifying before conversion to OSM format does not 
work well (generates polygons touching without sharing nodes).  And 
simplifying afterwards in a highly distorted coordinate system does not 
seem so good either.  It seems though that JOSM actually simplifies 
based on spherical/ellipsoidal distance - does anyone know for sure?).

> What tags do you plan to use on the changesets, and what size
> changesets are you planning to use?

I have not yet given this much thought but as for the tags i would 
probably suggest:

type=import
source=ADD 6.0
website=http://wiki.openstreetmap.org/wiki/Antarctic_Digital_Database

Changesets should probably be limited to a single tile - the larger ones 
will however require multiple changesets (would be split as required).

Greetings,

-- 
Christoph Hormann
http://www.imagico.de/