[Tagging] Sewage, tailings, and evaporators

[Minh Nguyen]

Vào lúc 11:57 2021-01-09, Kevin Kenny đã viết:
> On Sat, Jan 9, 2021 at 1:41 PM Brian M. Sperlongano 
> <zelonewolf at gmail.com 
> <mailto:zelonewolf at gmail.com>> wrote:
> 
>     For the case of sewage and tailings at least, I'm thinking that
>     these are quite discernable from waterbodies.  It should be pretty
>     straightforward to come up with a gallery of examples as to what
>     they look like on imagery and how they're characterized.
> 
> 
> Frederik has a point. I'd be hard put to tell in 
> https://kbk.is-a-geek.net/attachments/20210109/Tahawus.jpg 
> <https://kbk.is-a-geek.net/attachments/20210109/Tahawus.jpg> what is 
> natural water and what is tailings - the mine is right on the river. And 
> I'd have an even harder time if the mine has been abandoned for over a 
> century, as at 
> https://kbk.is-a-geek.net/attachments/20210109/Surebridge.jpg 
> <https://kbk.is-a-geek.net/attachments/20210109/Surebridge.jpg>. (I 
> still won't drink from a couple of the streams in the latter image! The 
> water tastes foul and is widely suspected to be contaminated with heavy 
> transition metals, even after all that time.)

This seems to me like an argument to deprecate natural=water in favor of 
landuse=reservoir or something even more generic, not an argument to tag 
tailings ponds as natural=water, as was originally proposed in [1]. 
Since then, Brian has revised the proposal to industrial=tailing_pond. [2]

The landuse=reservoir scheme is focused on the method of construction, 
whereas the natural=water scheme is focused on the contents. A typical 
potable water reservoir could neatly fit under either scheme, so moving 
those from landuse=reservoir to natural=water water=reservoir doesn't 
raise a question about accuracy.

By contrast, retagging a tailings pond as natural=water water=tailings 
would be a stretch, if only because on the other end of the spectrum we 
have tailings ponds that obviously are full of slurry -- "water" is the 
last thing you'd call them, but should they be tagged (and called) any 
differently than tailings ponds that are more settled? What's the cutoff 
between *=unsettled_tailings and natural=water water=settled_tailings? 
Since we seem to be disbanding the reservoir_type=* key, at some point 
we'll also need to contend with reservoir_type=sewage for pig lagoons 
and the like, which are an even poorer fit for natural=water.

> Then again, mapping waterbodies from aerials around here is harder than 
> it looks. Many orthos show either ice or mats of floating vegetation, 
> both of which can be hard to distinguish from adjoining land. To do a 
> really good job, you need to compare imagery from different seasons.
> 
> Nevertheless, I'd personally be perfectly fine with tagging 
> 'natural=water' on any of these objects if mapping from aerials and 
> fixing it when better information is learnt.  It's no worse than other 
> mistaggings I've seen arising from misinterpretation of imagery. I don't 
> think that means that sewage farms, tailings pools, or spoils 
> evaporation basins 
> (https://kbk.is-a-geek.net/attachments/20210109/VischerFerrySpoilsDump.jpg 
> <https://kbk.is-a-geek.net/attachments/20210109/VischerFerrySpoilsDump.jpg> 
> - when dredging is active, the basin is indeed flooded; the orthos 
> caught a dry time) have to remain `water` once the mapper knows better, 
> even if it makes sense to tag them that way initially. It's no worse 
> than mapping `natural=water` and discovering that the imagery you were 
> working from happened to be captured during a flood. (Don't ask me how I 
> know about _that_ problem!)

A tailings pond can often be identified by the mine or quarry nearby. 
Otherwise, I agree that it becomes more difficult. When it comes to 
bodies of liquids/colloids/suspensions, I don't think we can totally 
avoid having to make an educated guess based on context that may 
sometimes be wrong -- that's a key risk associated with armchair mapping.

I've lost count of the times I've had to go back and retag a poorly 
maintained swimming pool I misidentified as an algae-infested pond, or a 
fountain or marsh I misidentified as a pond because I was using leaf-off 
aerial imagery taken in early spring. Despite having made those 
mistakes, I think it would be counterproductive to deprecate 
leisure=swimming_pool, amenity=fountain, and natural=wetland in favor of 
something more generic.

I mentioned this in another thread, but USGS topographic maps symbolize 
tailings ponds completely differently than water ponds. They aren't even 
grouped into the same section of the legend ("Rivers, lakes, and ponds" 
versus "Surface features"). [3] Of course, there's plenty of precedent 
for a renderer adjusting the appearance of a feature based on a subkey 
or secondary key, but in my opinion, to see it listed as a completely 
different kind of feature lends support for an independent tagging scheme.

The revised proposal of industrial=tailing_pond seems reasonable to me, 
setting aside the fact that some existing industrial=* tags are 
secondary tags paired with landuse=* tags, while others are primary tags 
in their own right. That's a far larger issue that shouldn't necessarily 
block this proposal. Besides, I think we'd all prefer to debate whether 
it's "tailing pond" or "tailings pond" or whatever the British 
manufacturers market them as. ;-)

[1] https://wiki.openstreetmap.org/wiki/Special:Diff/2087932
[2] https://wiki.openstreetmap.org/wiki/Proposed_features/Tailing_pond
[3] 
https://lists.openstreetmap.org/pipermail/tagging/2020-December/057059.html

-- 
minh at nguyen.cincinnati.oh.us