[talk-au] Bulk loading all the Australian Statistical Geography Standard into the OSM - a query from the Australian Bureau of Statistics [SEC=UNCLASSIFIED]
andrew.harvey4 at gmail.com
Wed Feb 23 11:26:23 GMT 2011
Thank you for engaging with the community on this matter.
My comments here are referring to the Mesh Blocks data I'm seeing, which
looks like this http://andrew.liway.net/ABS_MB_2011_Landuse.png
>From what I can tell...
= 1. ASGS data into the OSM DB =
== The ABS mesh block boundaries... ==
These boundaries are defined arbitrarily by the ABS solely for the use of
tieing your statistics with geographical regions, as such contributors can't
really improve the data. It would just sit there in the database.
Also the data is truly a separate layer from most of the other data in OSM
(okay, it may tie to LGA or Postcode, but then what more information do they
provide). It seems no nodes or ways are shared between existing OSM data and
the mesh blocks, and no relations could the two data sets together. ie. the
mesh blocks would just sit on top of the rest of the data.
An import of mesh blocks would also add lots of clutter to the OSM data.
I think these first two reasons (and to a lesser extent the third) are big
reasons for not-importing this data into the main OSM database. The ABS or
the community can make .osm files of your current data. This would make it
easier for people who wish to use ABS data in an OSM focused environment,
with tools designed to work with OSM data without having to add the data to
the main OSM database. As you already have made the 1st step toward open
data (ie. using an open license), anyone can publish .osm files from your
I'll have to see what other people think. One could almost use these as
arguments against importing LGA data, except that LGA boundaries can be
surveyed by going around and checking signs. In NSW we don't currently have
the LGA's in OSM, I can't seem to find them on the ABS site though (only on
the $99 CD from ABS2006, which unfortunately isn't CC licensed)... This is
something we can potentially import too.
== The landuse attribute... ==
However, I do think the landuse attribute could be useful. I would advice
against one mass import of this landuse data, because there are areas that
are already tagged with landuse. (I'm thinking convert the data to OSM
format, and a contributor could go around use these landuse polygons as a
source, refining them if needed, and uploading only what the contributor
feels in ready and now going to mess up existing data. There will need to be
more discussion here.)
== Postcode, LGA, suburb, etc Boundaries... ==
I'm not sure, but if this ASGS data can provide for updates to the existing
postcode and lga boundaries in OSM, it would be best for these datasets to
be published as on OSM file, contributors can then check this against the
current database and manually make improvements if need be.
= 2. License =
As far as I can tell, unfortunately only members who have not agreed to the
Contributor Terms can upload anything that requires attribution. Although as
others have mentioned, forks of the current OSM database are happening which
will hopefully provide a smooth transition to a CC-BY-SA database with no
Contributor Terms requiring additional rights soon.
I think adding any such attribution should be done against a source tag on
the individual nodes, ways or relations.
= 3. Data =
I tried converting the a shp file from
the OSM format using ogr2osm.py, but it makes complex multipolygons. I
would rather it stick to simple closed ways. I suppose we could try
converting to OSM with FME and see what it looks like from there.
Questions. I do have a question. Where does the SSC (suburbs) data that the
ABS published for the 2006 census come from?
As mentioned in this thread
has been some uncertainty over what to do about previous ABS imports
which provided some administrative boundaries and SSC names.
In some cases it was obvious that the SSC admin boundary was following the
coastline, which appeared to have been traced from some aerial imagery. The
problem was should we use the same location for this admin boundary and the
coastline, or should we move the admin boundary to match the coastline as we
improve the quality of the coastline data, and/or as the coastline moves.
Its hard to draw conclusions when we are unsure where the data came from
originally and how it was made.
On Wed, Feb 23, 2011 at 12:01 PM, Marcus Blake <marcus.blake at abs.gov.au>wrote:
> To the Australian OSM community,
> The Australian Bureau of Statistics has recent published the first part of
> a new statistical geography, the Australia Statistical Geography Standard or
> ASGS for short. The boundaries are based on a new basic spatial unit called
> a mesh block which have been aggregated to create efficient spatial units
> for the dissemination and analysis of statistical data. They have been
> released in advanced of the 2011 Australian census and are fixed for the
> next 5 years. The attached links and PDF file provide additional
> The ABS Geography section is presently investigating the possibility of
> loaded the new Australian Statistical Geography Standard into the OSM
> As a starting point, I'd like to start a discussion about how this could be
> achieved, if it is possible at all.
> From the ABS point of view the principle reason for doing this is that an
> the OSM database would hold a copy of the official version of the
> boundaries and that this point of truth would be available for all OSM users
> and downstream distributors. It would therefore become one of the channels
> by which the ABS distributes the ASGS boundaries and associated coding
> There are three main issues I can see need addressing (and probably a large
> number of other issues I'm not yet aware of )
> 1. Is the OSM database a suitable location for the ASGS*
> The ABS would like to facilitate the use of the new ASGS as much as
> possible and the OSM database looks to be an efficient mechanism for the
> distribution of the spatial boundaries and codes. But what does the
> community think??...
> 2. Licensing*
> Even though ABS data (including all spatial data) is released under a CC
> license it does require attribution (Attribution 2.5 Australia CC BY 2.5).
> How is this license model handled under OSM. Is there a means to associated
> attribution with particular "layers" within the OSM database?
> 3. The practicality's of loading load.*
> I note previous posts on loading the ABS Postal Areas and the technical
> problems involved. What is the most efficient and best way of load a
> categorising these data within the database? Our preference would be to bulk
> upload through an FME process. Perhaps this is a question for the imports
> Any Questions for the ABS?*
> Lastly if there are any questions people have on the new ASGS (and the old
> ASGC) or anything on the definition or application of statistical boundaries
> I am happy to answer specific queries and contribute to discussions.
> Marcus Blake
> Marcus.Blake at abs.gov.au
> Assistant Director
> Geography Section
> Australian Bureau of Statistics
> Additional Information*
> *2911.0.55.003 - Census of Population and Housing: Outcomes from the
> 2011 Census Output Geography Discussion Paper, 2011*
> This publication a good diagram of the ASGS
> ABS License details: CC Attribution 2.5 Australia*
> The first volume of the ASGS. *
> This includes all the electronic boundaries in MID/MIF and Shape formats
> **Australian Statistical Geography Standard (ASGS): Volume 1 - Main
> Structure and Greater Capital City Statistical Areas, July 2011 (cat no.
> 1270.0.55.001)* <http://www.abs.gov.au/ausstats/abs@.nsf/mf/1270.0.55.001>
> The ABS Geography website*
> **http://www.abs.gov.au/Geography* <http://www.abs.gov.au/Geography>
> Geography Frequently Asked Questions*
> Free publications and statistics available on www.abs.gov.au
> Talk-au mailing list
> Talk-au at openstreetmap.org
-------------- next part --------------
An HTML attachment was scrubbed...
More information about the Talk-au