[Imports] Hydrographic data import from LINZ

[Kyle Hensel]

> is glacier area data incorrect?

Yes I think that’s part of the issue. The crevasse data agrees with the aerial imagery from the USGS, whereas the glaciers in OSM seem to be less precise or match a different snow line.


> is remaining creevace data of the same low quality? Just not obviously detectable?
> If this was missed - what else was not properly reviewed?

As above, it seems like the crevasses are fine, and the glacier outlines are slightly different to the lastest aerial imagery (which is expected to a certain degree).
Here’s a screenshot from iD of the way you mentioned earlier (in green) and the glacier outline (blue)
https://user-images.githubusercontent.com/16009897/138617640-473840ad-c43d-4ba7-8796-b8a23af02dea.png


> So OSM would be reduced to mirroring some official dataset?

No, it will only flag cases that haven’t been edited by anyone since they were imported (or last updated).
This system will never undo anyone’s hard work. It just makes sure that we fix imported data which is now out-of-date and was never touched by anyone else.


> Primary problems caused by low quality imports range from [...]

Thanks for the detailed explanation. This is definitely an issue in OSM, although I don’t think these crevasses pose a significant issue:
If you’ve surveyed an area, it shouldn’t be too difficult to select and delete the small amount of crevasses in that area if you want to re-draw them.
Bear in mind that there are (at most) ~5 ways per 100m^2. Compared to a town with a house every 10 metres, the data is actually quiet spare.

Are there any other ways we can improve imports to make it less intimidating for new contributors to edit the area?
I know that avoiding confusing and useless tags like TIGER:*, LINZ:*, etc. is one important point.


> I am asking about announcement on imports mailing list

Thanks for pointing that out. I will send a seperate email to this mailing list to formally introduce the decade-old import.
From: Mateusz Konieczny via Imports<mailto:imports at openstreetmap.org>
Sent: Monday, 25 October 2021 06:52
Cc: imports at openstreetmap.org<mailto:imports at openstreetmap.org>
Subject: Re: [Imports] Hydrographic data import from LINZ




Oct 24, 2021, 03:45 by K.y.l.e at outlook.co.nz:
> How crevasses can be in rock outside glacier?

I think the extent of the glaciers from the ADD import don’t quite match the glaciers that LINZ has surveyed.
There seem to be only a few cases like this, right at the edge of the glacier, we can delete these if you like.
The question is
- is glacier area data incorrect?
- is remaining creevace data of the same low quality? Just not obviously detectable?
If this was missed - what else was not properly reviewed?
> How you plan to update this data?

Using the same system as the ongoing address update, you can try it yourself from https://osm-nz.github.io/RapiD
It can handle the complex aspects like moving, editing, and deleting data.
So OSM would be reduced to mirroring some official dataset?
With the most glaring errors hidden but remaining data hidden?
> Primary effect is to make this area uneditable.

Not sure what you mean by this. This area is just as editable as any other part of the world, bearing in mind the lack of good quality aerial imagery.
Primary problems caused by low quality imports range from
- many mappers especially new are unaware that low quality data mismatching reality,
especially imported data, can be freely deleted
- even if they aware of this - it is often very tedious to do this, especially for new people
- in general people are often far more motivated by mapping something new
over repairing garbage

Obviously all of that applies only to bad imported data, imports of data that is as
good as people would map are extremely helpful.

But I was editing in some places where for example people imported massive
forest multipolygons (thousands of elements) or import of the same features was done
multiple times or import utterly mismatched reality or someone imported data that
actually no longer existed.

It was confusing and miserable experience and for someone new it would be
likely enough to completely discourage them from contributing.

I also encountered many requests, even from long term mappers to
help with areas especially affected by imports that made editing nearly impossible
in a given area.

If you are unaware that bad import can heavily damage editability and
you are not taking this into account then I would strongly encourage
to avoid doing this Hydrographic data import from LINZ.
> Also, I want to repeat request for linking specific posting on mailing list that announced plan to import this features.
> Changesets that mass added this features miss links to relevant import documentation.

I think I answered all this in our other conversation, in addition, here is the link to talk-nz which I didn’t include https://lists.openstreetmap.org/pipermail/talk-nz/2021/thread.html
All changesets made after 6/10/21 have source=https://wiki.osm.org/LINZ, in addition to attribution=https://wiki.openstreetmap.org/wiki/Contributors#LINZ which they always had. I hope this addresses that problem.
I am asking about announcement on imports mailing list. It is obligatory and gives
people to stop harmful imports. And as far as I know imports that failed this requirement
can be freely reverted.

(I understand that small local and uncontroversial imports are often not doing this,
and in such case such reverts would be unhelpful, but it is not one of such cases)

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.openstreetmap.org/pipermail/imports/attachments/20211024/546b0103/attachment.htm>