Talk:Canada Building Import

From OpenStreetMap Wiki
Jump to navigation Jump to search

Task Status

  • Updated the documentation on how to add a reverse filter in JOSM James2432 (talk) 15:11, 20 December 2018 (UTC)

User account?

Just a reminder that if we're following the import guidelines, dedicated import accounts should be used. Let's not get blocked again, as happened in Ottawa. See Import Guidelines - Uploading and Use a dedicated user account for details. --Scruss (talk) 18:50, 20 December 2018 (UTC)

Background section needs cleanup

This currently reads like a lot of independent thoughts squeezed into a section. I'm not familiar with the background myself - can someone who is please help with this? Nate Wessel (talk) 05:01, 19 January 2019 (UTC)

I did recently (today) add the last-paragraph blurb about the "old" wiki, so I can see how you'd think that, Nate. As John Whelan is the "essential author" of this wiki, I think it is incumbent upon him to better clarify this Introduction section. What I'm saying is "he wrote it, I re-wrote an older version he 'chucked' and turned into this one." So, as I'm largely trying to "stay out of the way" (though I'll speak when spoken to) I'd like to see John clarify this section. Stevea (talk) 05:08, 19 January 2019 (UTC)
I rewrote it / cleaned it up. John, do you want to take a look? Stevea (talk) 07:10, 19 January 2019 (UTC)

Name change?

The name of this page gives me a bit of a headache and is very hard to search for. Is there something simpler and more concise that would work for everyone? Might I propose perhaps something like StatsCan Canadian Building Import? The details that are presently embodied in the title here can of course be retained on the actual page. Nate Wessel (talk) 05:05, 19 January 2019 (UTC)

As StatsCan is largely historical now in this project, I'd be OK with "Canadian Building Import" in the interests of simplification. Though as an "outsider" (American), maybe I only get a "slightly diminished" vote?! Stevea (talk) 06:22, 19 January 2019 (UTC)
I'd like to re-emphasize that while Canada Stats might be the publisher of the data, given their license, they no longer have much to do with these data entering OSM. This is an OSM project and "Canada Stats" being in the name seems superfluous. It's true that in the USA we call a massive road-and-rail import we did in the mid-2000s "the TIGER import" (from the name of a flavor of the USA's Census Bureau data, the "the Topologically Integrated Geographic Encoding and Referencing system, a public domain data source which has many geographic features") and in a certain context (that from the perspective of editors in the USA) "we know what we mean by that." Yes, there is a "TIGER fixup" project in OSM, (though it is fairly long in the tooth, or getting old and losing some focus and direction), but I don't think that we want to call the new wiki that is birthed from this Import/Plan "Stats Canada fixup" (or something like that). So, let's simply drop "Stats Canada" altogether and refer to this (emerging WikiProject) as I say above (and fine-tune a bit), WikiProject Canada/Building Import. Of course, other suggestions are welcome. Stevea (talk) 03:08, 20 January 2019 (UTC)

Proposal
There are several sources from which we could import building footprints. ODB (StatCan) is one of them, Natural Resources Canada is another one, Microsoft is also offering buildings and other sources may be available soon.
I propose that “Canada Building Import” be renamed “Canada Building Imports”. This way, we could use the page to define our expectations concerning the quality of imported data, the general import process, etc., and document each import via a link to another dedicated wiki page (e.g. “Canada Building Import - ODB” for current ODB import).
If there is no objection, I could make the changes we will have discussed.
--Jfd553 (talk) 20:02, 1 January 2020 (UTC)

Quality Assurance details

In the Page, the Import Plan section says "Quality assurance will be done by people using the Tasking Manager as well as Osmose and other QA tools to ensure proper compliance." Let's please be more specific about specifically WHAT tools (not "other QA tools") will be used, as well as the step-by-step instructions it will take to complete this testing satisfactorily. Too often, QA gets short shrift, being given only a "hand wavy" high-level description (because some never intend to complete it, and/or others don't know exactly how) and then never gets done. Please, let's not let that happen here. Stevea (talk) 18:08, 24 January 2019 (UTC)

Improving data quality through pre-processing

I developed a tool that can be used to clear some of the problems identified in the data (see "Quality Issues in Source Data" in Data section). However, the tool needs an FME license to be run. I could process ODB data, on a municipal basis and when required, before it is being served through the tasking manager (see "OSM Data Files" in Data section).

In summary, the tool ...

  • orthogonalizes all buildings' corners that are within a given angle tolerance from being at a right angle.
  • preserves the topology of shared buildings' boundaries.
  • tags buildings within a building as building:part=true
  • removes unnecessary vertices.

The complete documentation is available here... https://github.com/jfd553/OrthogonalizingBuildingFootprint
The documantation is detailled, hoping it may motivate a few courageous programmers to translate it in an open source application.

Nice to have: If we go ahead with this approach, I might try to implement a topological check to identify buildings having crossing footprints. In these cases I could tag the feature with fixme=’Topological errors’.
--Jfd553 (talk) 22:35, 31 December 2019 (UTC)

Post-Processing

Once the building are analyzed/corrected, the import participants must be advised to look at the note and fixme tags to properly import the new geometries.

Other Comments About the Wiki Page Content

I just realized that we might want to consider to replace the image Nate used to explain the simplification process. The image shows a building with two different roof heights, which should be entered as explained in the wiki about roof modelling. Such buildings can show up with intermediate vertices as the height tag will be different. However, in the given image the vertices do not correspond to the different roof heights.
--AGeographer (talk) 23:28, 26 March 2019 (UTC)

Import steps need fine-tuning?

How 'bout them Import steps? I think an advanced OSM mapper can use them as-is, do we want to fine-tune them with a bit more detail for intermediate mappers who might want to begin the fun and effort of a medium-to-large scale import? Stevea (talk) 05:20, 26 January 2019 (UTC)

I mostly agree regarding current steps for advance OSM mappers. However, I do not agree with the proposed step 3 (before merging buildings data switch to the OSM layer and see if there are any clusters of buildings without any meaningful tags you can delete to save time when merging…). It assumes that imported data are more up to date than the OSM layer. That might be true in urban area where the data come from local GIS department but in rural areas, StatCan uses CANVEC data when nothing else is available. You should understand it concerns most of the Canadian territory. I have added/modified/deleted thousands of such buildings in southern Québec (Eastern Townships) and the current procedure will simply scrap the work I have done to replace it with the same data I edited over the years. The proposed procedure does not respect the work done by previous mappers. Merging all buildings one at a time is certainly longer but it is also more respectful of other contributors.--Jfd553 (talk) 16:56, 3 February 2019 (UTC)

I totally agree. Even in urban areas, there should be no instruction to thoughtlessly delete anything contributed by users, regardless of the tagging. I've added many buildings with just a building=yes tag, and I'd be quite upset if they were summarily deleted. At least update the geometry and preserve the way history if nothing else. Nate Wessel (talk) 17:01, 4 February 2019 (UTC)

Previously, Canvec data was provided in StatCan's datasets, but looking at their current metadata, I found that CANVEC buildings are no longer included - good! However, the respect for the work done by previous contributors remains a problem in the proposed procedure. --Jfd553 (talk) 17:09, 4 February 2019 (UTC)

Request for update on simplification?

I removed the following text from the simplification section on the main page: "Can somebody offer an update about this? (Newer than January 25, 2019?)." In the context, I wasn't sure what it was asking for, so I figured I would drag it over to here where we can discuss . Nate Wessel (talk) 18:18, 27 January 2019 (UTC)