[talk-au] PSMA Administrative Boundaries

[Andrew Harvey]

On Tue, 9 Oct 2018 at 20:51, Andrew Davidson <theswavu at gmail.com> wrote:
> One question would be should we be putting a place tag on admin
> boundaries? According to taginfo only 34% of admin_level=10 has a place
> tag.

I say yes since these suburbs are an area not a single point.

> > - how we'll manage that upload in terms of breaking it up into smaller pieces
>
> I'd assumed that you'd be uploading them using an import account. How
> small to break them up into would depend on what's used to upload them.
> State-by-state does have the advantage that the existing state borders
> are going to have to be manual snapped to the imported boundaries anyway.
>
> > - if there is any re-using of existing ways or not
>
> As several others have pointed out if we import the boundaries
> completely then mappers can manually snap them to natural features,
> roads, other boundaries etc to their heart's content. This step is
> optional as the boundaries will still be valid even if no one does any
> follow up work.
>
> As a minimum the boundaries would have to be matched to the existing
> state boundaries, otherwise there will be overlaps or slivers.

+1

> > - how to ensure we're retaining the existing LGA/Suburb data in OSM.
>
> So I'm assuming here that we are not talking about NSW/SA/ACT. These
> have already been done from state data and if the PSMA data is different
> then we should be going back to the state data to check (at which point
> we're using the state data).

+1

> To keep history I thought we'd could keep record of the current relation
> numbers of the existing admin boundaries, scrub off the no longer needed
> ways, and once the new boundaries are imported swap the new for old. For
> boundaries that are currently closed ways we'd need to convert them to a
> one-member relation first.

That should be possible.

On Tue, 9 Oct 2018 at 21:14, Lee Mason <lee.mason at outlook.com.au> wrote:
> I didn’t know what to make of QLD, but I’ve just found this definition which suggests that the OSM coastline should be sufficient:
> http://qldspatial.information.qld.gov.au/catalogue/custom/viewMetadataDetails.page?uuid=%7B3F3DBD69-647B-4833-B0A5-CC43D5E70699%7D
> “For coastal areas other than Brisbane, the LGA comprises the mainland and all islands above their respective sea-shores within the encompassed area.”

Thanks Lee!

On Tue, 9 Oct 2018 at 21:53, Joel H. <joelh at disroot.org> wrote:
>
> Good, doing country wide I think would have been a pain. I just tried
> QLD and all is looking good except the boundaries which extend into the
> ocean.
>
> Are you going to do the upload? And when should we do it?

I'm happy to do it, but I don't think we're at the point where it's
ready just yet.

As others have pointed out, in terms of LGAs,
- WA has three already in OSM of unknown origin, some don't match the
PSMA data. I've added a changeset comment to try to determine if we
leave those in place or replace them.
- NT has one which is from 3 years ago originally from GA, which
doesn't account for a new LGA from 2014, so I'm okay to replace this.
- TAS have no LGAs in OSM currently
- Chrismas Island/Cocos Island have no LGAs in OSM currently
- SA has some LGAs, we'll just add in those missing and leave the
existing ones in place
- ACT is N/A
- NSW already in OSM, so let's skip this
- VIC already in OSM, so let's skip this
- QLD has some rough approximations for a few LGA, I propose to
replace these with the more detailed boundaries

for Suburb/Localities, VIC has a lot of big differences so some local
knowledge on that would be great, NSW is mostly complete, SA is just
missing Adelaide.