[Talk-us-massachusetts] Hello

Fri Dec 18 00:43:58 UTC 2020

Hey all - jumping in with a bit of commentary from "south of the border" -
there aren't really any other regular mappers in RI, so talk-us-mass is
effectively my "local" list!  You guys are serious pioneers up there with
the MassGIS imports.  Sorry if this runs long but there's a lot here.

I had a chuckle about the "16 hour course on IUCN rules", because that is a
spot on criticism of the protect_class key, and it's why I got involved in
all of this.  All I wanted to do was to tag my local conservation areas and
state parks, and I found the wiki pages on the topic *wicked* confusing.

I've been working on a concerted campaign to fix it, with lots of help from
stevea whom many of you know and others in the community.  The goal is to
replace all of the insane parts of the current tagging scheme and get us to
a place where mere humans can map this stuff.  However, it's not enough to
say "protect_class sucks, let's deprecate it" -- you have to come up with a
replacement for all the different things people are using protect_class
for, and make it better!  So I've been working to dismantle protect_class
one value at a time.  Unfortunately each of those protect_class values
represents its own collection of nuances, corner cases, etc., and there's
always some community of interest that has a particular stake.  Here's
what's been accomplished so far:

1. Massive cleanup to the boundary=protected_area and protect_class pages
to figure out how these values were actually being used and clearly, and
cleanly, document those values, so that they can be systematically
deconstructed one by one, and replaced with sane tagging.

2. I started by proposing boundary=special_economic_zone as a replacement
for protect_class=23.  That passed 25-2, with the two "no" votes from a
couple of protect_class enthusiasts that are strong proponents of the
numbering system (and would love it if there were even MORE protect_class
values!)

3. Next, I went after protect_class=16.  There was an abandoned proposal
from 2007 for "hazards", and the tagging was popular enough that it
actually had 30K usages.  I cleaned that up and proposed it along with
deprecating protect_class=16.  That proposal[1] is currently voting, so
feel free to add your vote!

4. A mapper in Australia is currently working on proposal to improve the
tagging of military bases.  I asked him to include "...and deprecate
protect_class=25", which was readily agreed to, and that proposal is moving
to a vote soon.  Boom, another one bites the dust.

The next one that I'd like to go after is protect_class=12, which covers
water protection areas.  When I found out that some mappers up in
Massachusetts had created landuse=reservoir_watershed...I shed a tear that
day.  There is now a beautifully crafted wiki page[2] discussing the
particulars of how to tag these areas.  The only thing holding me back
right now is that I'd like to get more places to start using that tag.
It's almost exclusively used in Mass and RI and it would be helpful in the
proposal to have more geographic diversity so it's not just a local thing.

Anyways -- if you've read this far and you agree that protect_class is an
unholy mess that we should replace, the one thing that would help is to
stop using the "fictional" values -- namely protect_class values greater
than 6.  What would _really_ help is to replace those fictional values with
plain-English tagging.  Here's an OP query with the 170 or so of these that
are still left in Mass: http://overpass-turbo.eu/s/11j2

P.S.

With regard to landuse=conservation - that tag showed up as deprecated on
the wiki before I started mucking about with public land tagging.
However...if you look at the history of that page on the wiki, it looks
like someone just unilaterally marked it as deprecated.  It looks like it
was invented in Massachusetts, but it IS used worldwide.  There was a
posting[3] in the tagging list about this specific question.  If the Mass
mapping community believes that landuse=conservation is still a correct
tagging, I would encourage you to take the description in the Mass
conservation wiki page and make landuse=conservation say that.  There was
never a vote to deprecate it, so you're totally in the clear to define and
document  landuse=conservation in the way you're using it.

[1] https://wiki.openstreetmap.org/wiki/Proposed_features/hazard
[2] https://wiki.openstreetmap.org/wiki/Tag:landuse%3Dreservoir_watershed
[3]
https://lists.openstreetmap.org/pipermail/tagging/2020-December/thread.html

On Thu, Dec 17, 2020 at 9:15 AM Greg Troxel <gdt at lexort.com> wrote:

>
> Tom Parent <tomparent at gmail.com> writes:
>
> > I'm a new but active mapper out of Belmont.  TomPar on OSM.  I've chatted
> > with some folks on the OSM Slack site but didn't think to subscribe to
> > this.  Seems there are different people on this list?
>
> Welcome to the list.  Yes, each OSM communications mechanism has
> different pepple, with overlap of course.  Personally, I don't think
> open data/open source projects should use proprietary tools, so I do not
> deal with Slack.
>
> (I used to live in Belmont pre-OSM, and used to be there a lot, but not
> so much recently.  You'll probably see my userid (gdt) in changesets.  I
> map around Stow mostly now, and have done most of the trail mapping in
> our conservation lands.)
>
> > my mapping passion is making public accessible parcels available for
> > outdoor activities more clear and dataset/visually consistent across New
> > England...mostly eastern MA and VT that I'm most familiar with after
> > decades of traipsing the countryside.
>
> A caution when you mention visually consistent.  There are fairly few
> points of doctrine in OSM, but a big one is having the database
> represent reality using tags that are defined by consensus, and not
> adjusting tags to achieve a particular rendering.  Various renderers
> make choices about what to show based on what the renderer maintainers
> think is important.
>
> Specfically, my town has a lot of land that is "conservation land",
> which means maintained in a natural state with trails for walking/etc.
> This is tagged "landuse=conservation" and "leisure=nature_reserve"
> following longstanding practices.  It also now has
> "boundary=protected_area" which is a modern Euro convention foisted on
> the rest of OSM.  While tags are just tokens, the use of that as
> boundary leads to confused thinking that the polygon feature around the
> edge is the important thing, and this bleeds into the default render.
> In reality, all those tags describe a property of the area, not the
> edge.
>
> The maintainers of the standard style, that appears on
> openstreetmap.org, appear not to think that open space mapping is
> particularly important.  Things like golf courses render more
> prominently than conservation land, for example.
>
> However, the great thing about OSM is that anyone can create a render
> how they want, and show what matters.  Besides the other styles on the
> main map, there are other map displays on the web.  Also OsmAnd does its
> own rendering, and you can tweak it - but it respects
> landuse=conservation.
>
> [I'm reordering your text to reply.]
>
> > I'd like to help contribute to tagging/mapping convention
> standardization.
> > I'm particularly interested in burgeoning efforts like:
> > https://wiki.openstreetmap.org/wiki/United_States/Public_lands
>
> We have documented the local consensus (as reached here) on conservation
> mapping:
>
>   https://wiki.openstreetmap.org/wiki/Massachusetts/Conservation
>
> On a quick skim of the Public_lands page, things seem mostly
> consistent.  A few points:
>
>   The big difference is that we do not recognize the deprecation of
>   landuse=conservation, and ask that all parcels for which "preserving
>   the land in a natural state indefinitely", more or less, is the
>   primary purpose, have a landuse=conservation tag.  However this is not
>   in conflict as the other pages do not specify a landuse.
>
>   To understand the public_lands page you need to take a 16-hour course
>   in IUCN rules.  That's a joke, but the whole scheme of numeric code
>   points is contrary to larger OSM norms of tags that while technically
>   tokens with defined meaning are somewhat understandable.  I'm a long
>   term mapper (since 2009) with a degree in computer science -- and I
>   find it way too confusing.
>
>   There is a real focus on government with protected_area (not
>   surprising as it seems to be a UN thing).  But in MA, there are a
>   tremendous number of places that are legally protected from
>   development by Conservation Restrictions, which are easements recorded
>   at the registry of deeds, some held by towns, and some held by land
>   trusts.  We consider those entire sufficient for protected_area
>   status, but don't use that tag for general municipal land without a CR
>   or legal conservation status even if it is currently used as a
>   nature_reserve.
>
> Also landuse=forest means the principal use is forestry as opposed to
> principally conservation with some forestry that is not inconsistent
> with long-term open space conservation.  Some people use landuse=forest
> to show that there are trees, or because the parcel is called e.g. "Stow
> Town Forest"; I think both of these usages are incorrect.
>
> > I hope I've already made some positive contributions to parcels in
> > Belmont/Arlington/Lincoln.  I haven't received any negative changeset
> > comments yet so presume everything is OK.  I did take some time to learn
> > the culture of OSM, get up to speed with JOSM, before really digging in.
> > Thanks to whoever made available the MassGIS orthos and parcel layers in
> > JOSM.  It's been extremely useful.
>
> You're welcome - that was a group effort over time by several of us on
> this list.  There is also an easement layer from MassGIS but they don't
> have a tile service.  That is only occasionally useful; e.g. one
> conservation parcel in Stow does not have any frontage and there is an
> access easement that in included in the leisure=nature_reserve polygon.
>
> I haven't really been paying attention recently to Belmont.  The only
> thing I would throw out from a quick glance at the standard render is
> that there is a surprising amount of solid green.  If you have tagged
> things according to the guidelines and that's how it is, great --
> perhaps the standard render render is changing.  But if you are choosing
> tags to make things green, that's more or less not ok.  I believe a fair
> bit of Belmont conservation land should be leisure=recreation_ground
> rather than park, but I don't tnink I'm seeing park green.
>
> And, if you have figured out IUCN tagging, and put in code points that
> are right, and that makes conservation green, then thank you and please
> explain the key points!
>
> Another big issue is the difference between land use and land cover.  In
> geography these are distinct concepts, and in OSM they mostly are but
> the default render tries to do both.  That's hard, and it does a decent
> job at an impossible task.  Plus, OSM is confused in that landuse is
> denoted sometimes by landuse= and sometimes by other tags.  Similarly
> for landcover; it's not landcover=foo, but some mix of various things
> which are all essentially landcover.  So semantically it's a bit of a
> mess.
>
> There's another difficult issue, which is natural=wood sorts of tags
> (landcover) and areal extent.  This denotes that an area is tree
> covered.  Around me (compared to Belmont which is the city :-) trees do
> not stop at conservation land boundaries, and looking at imagery you
> cannot figure out where conservation land stops.  So putting landcover
> tags on parcels is in my view basically wrong, unless the landcover
> really is limited to the parcel and uniform within it.  I basically
> don't believe that will ever be the case.  But I can believe in Belmont
> that this will be far more true than it is in Stow, given the local
> culture and zoning in terms of adjacent land use, where non-conservation
> land is split into 1/8 acre parcels, completely cleared, and having
> houses, driveway, lawn and pools only with a few trees.
>
> If you have micromapped grass vs trees from imagery, and gone over the
> parcel line when the landcover crosses, so that your landcover tagging
> is disconnected from ownership, then I salute you for amazing work.  And
> if so, I'd encourage you to do that micromapping of the whole town
> regardless of ownership.  That's the great thing about OSM, how each
> person can take really good care of a few areas and together we have an
> amazing map.
>
> Also, if you haven't put in stone walls those are great to add
> (barrier=wall wall=dry_stone).
>
> I'll try to take a look in more detail and I'm sure a few others will
> too.
>
> MassGIS publishes an open space layer and you can see it on Oliver.  The
> quality of the geometry is variable but it's great to use as a clue what
> to go look at.  An earlier version was imported long ago (2008?).
>
>   http://maps.massgis.state.ma.us/map_ol/oliver.php
>
> Finally, I want to point out a site run by Jason Remillard, one of the
> members of this list, that does a conservation-focused render and
> analysis.
>
>   https://www.mass-trails.org/
>   https://www.mass-trails.org/towns/Belmont.html
>
> You may find this useful not only to look at, but to find things that
> maybe you should tag more or differently.
>
> Long ago we had a few geo meetups in Cambridge (at CBC), and it would be
> good to do that again, when we reach Phase 4.
>
> Greg
> _______________________________________________
> Talk-us-massachusetts mailing list
> Talk-us-massachusetts at openstreetmap.org
> https://lists.openstreetmap.org/listinfo/talk-us-massachusetts
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.openstreetmap.org/pipermail/talk-us-massachusetts/attachments/20201217/21c98694/attachment-0001.htm>