[Talk-nz] Continuing the import of LINZ data into OSM

Kyle Hensel K.y.l.e at outlook.co.nz
Thu Sep 9 09:47:39 UTC 2021


Hi Mark, unfortunately you can’t use JOSM for this (yet).

I haven’t used JOSM much (I found it rather confusing) but the data is available as geojson files (and soon .osc files), so for anyone interested in using JOSM, this should be an option in the near future.

I’ll update this mailing list once the data is available as .osc files which I think is the most convenient for JOSM users.


Kyle

From: Mark Begbie<mailto:mark at begbie.org>
Sent: Thursday, 9 September 2021 19:48
To: Kyle Hensel<mailto:K.y.l.e at outlook.co.nz>; talk-nz<mailto:talk-nz at openstreetmap.org>
Subject: Re: [Talk-nz] Continuing the import of LINZ data into OSM

This looks great Kyle,

Maybe a beginner question, is there a way to import these layers into JOSM rather than Rapid? Just because I'm used to JOSM.

Thanks,

Mark
On 9 September 2021 6:54:51 pm NZST, Kyle Hensel <K.y.l.e at outlook.co.nz> wrote:
Hi all,

I’ve made a few of the simple datasets available to import.
Anyone is welcome to import this data using a modified version of the RapiD editor: https://osm-nz.github.io/RapiD

I think this system will be too inefficient for importing rivers, but hopefully this is a useful first step for the simple datasets.

Best regards,
Kyle


From: Kyle Hensel<mailto:K.y.l.e at outlook.co.nz>
Sent: Sunday, 22 August 2021 17:11
To: talk-nz<mailto:talk-nz at openstreetmap.org>
Subject: Re: [Talk-nz] Continuing the import of LINZ data into OSM

Hi,

rivers will definitely be the hardest to conflate. I’m proposing to make this data available for anyone to importing using a fork of RapiD, just like street addresses and similar to the building import.

As an example I've made a few layers available to view in the fork of RapiD (https://osm-nz.github.io/RapiD).

We may find that this system is too slow or cumbersome to handle importing some layers like rivers. I haven't used JOSM much myself, but I'd be interested to hear about better ways of doing this.
I know the original linz2osm system had a comprehensive system for this, but I haven’t had much luck trying to revive it.

For other data layers, it should be much more straightforward. There will be manual steps required in some cases, such as connecting roads and rivers to existing features. The saddles layer (available to test) is a good example of where the importer will need to manually merge the data with any existing nodes.

I'm also proposing to add a tag called ref:linz:topo50_id to all imported features. Like addresses and buildings, this will make it much easier to determine what data is already in OSM and will help if the data needs updating in the future.


From: Mike Kittridge<mailto:mgkittridge at gmail.com>
Sent: Sunday, 22 August 2021 13:50
To: Andrew Douglas-Clifford<mailto:adouglasclifford at gmail.com>
Cc: Kyle Hensel<mailto:K.y.l.e at outlook.co.nz>; talk-nz<mailto:talk-nz at openstreetmap.org>
Subject: Re: [Talk-nz] Continuing the import of LINZ data into OSM

Hi y'all,

I just wanted to mention my import processing of the LINZ rivers into OSM.

As Andrew mentioned related to the Mackenzie area, a direct/automated import is not practical and just causes a great deal of mess. Some time ago, I was working on improving the OSM river network by iteratively importing river sections from LINZ into JOSM and making the edits and corrections in JOSM before importing them into OSM. With a little bit of preprocessing in Python, the LINZ rivers didn't take too much work in JOSM to efficiently import rivers into OSM.

I really should get back to doing this again...because I really really like using OSM's river network and API for my hydrological analysis. It's much better than any other existing rivers layer for network analysis and routing.

If anyone is interested in a more detailed river import procedure, I'd be happy to provide it.

Take care,
Mike

On Sun, Aug 22, 2021 at 1:23 PM Andrew Douglas-Clifford <adouglasclifford at gmail.com<mailto:adouglasclifford at gmail.com>> wrote:
Hi Kyle,

Thanks for putting together the details on this. I haven't been involved in some of the older LINZ imports (only the recent building imports via RapiD), but I have seen the results of various older imports.

I think it might be useful to classify datasets by how they should be imported into OSM. Many of these seem simple enough via a bulk import, while others will require varying degrees of review/modification as they are imported (as some of your notes indicate). For these, that might mean taking the approach of importing/reviewing manually via RapiD instead of the previous methods used, to ensure it's done appropriately.

Some thoughts around specific datasets:
·        Rivers: I believe this may be quite complicated for a simple bulk import. I've seen somebody's previous messy import examples (e.g. Mackenzie District around the Tekapo area) that have taken some work to clean up, where the large multitudes of line segments have an incorrect direction, meaning the flow direction is topologically incorrect. Also, river lines do not always have a continuous path through wider bodies of water. Perhaps with the named rivers dataset imported pre-dissolved and with the correct direction will help with this - but this will need manual review, perhaps by catchment?
·        Peaks/Height Points/Place Names: There are quite a few peaks/place names added already, assume the import will only add those that are missing from OSM? I also made an effort to add the geographic LINZ place names (town through to locality) that weren't in OSM previously, so imports would obviously need to conflate with existing data. Don't see why all height points couldn't be imported though, given the use of OSM for many outdoor applications.

Cheers,

Andrew
andrew_dc



On Sat, Aug 21, 2021 at 9:20 PM Kyle Hensel <K.y.l.e at outlook.co.nz<mailto:K.y.l.e at outlook.co.nz>> wrote:
Hi all, I hope everyone is safe during lockdown...

I've noticed that the import of topographic data from LINZ was never completed. This took place between 2008 and 2014. During this time, the majority of topographic data from LINZ was imported into OSM. Since 2014, only addresses and building outlines have been/are being imported.

I've put together a proposal to continue importing 41 of the remaining data layers from LINZ.

I've updated the OSM Wiki page to provide an overview of what data is and isn't in OSM, and the OSM tags used for each layer.
This is available here: https://wiki.osm.org/LINZ The proposed new layers are marked as "🔵 Ready to Import" in the table.

I'd love to hear what everyone thinks about this proposal.
In particular, the proposed OSM tags for the new layers are just a suggestion - some of the choices are not straightforward and definitely need discussion.

Best regards,
Kyle

_______________________________________________
Talk-nz mailing list
Talk-nz at openstreetmap.org<mailto:Talk-nz at openstreetmap.org>
https://lists.openstreetmap.org/listinfo/talk-nz
_______________________________________________
Talk-nz mailing list
Talk-nz at openstreetmap.org<mailto:Talk-nz at openstreetmap.org>
https://lists.openstreetmap.org/listinfo/talk-nz


--
Sent from my Android device with K-9 Mail. Please excuse my brevity.

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.openstreetmap.org/pipermail/talk-nz/attachments/20210909/35468393/attachment-0001.htm>


More information about the Talk-nz mailing list