I also love that GIS is a fractal of cursed things.
-
If nothing else, a trivial thing I hadn't considered: tides. Coastlines aren't a constant feature even on a day timescale. OpenStreetMap standardizes on the coastline feature being "mean high water springs", the high tide line in spring averaged over the past 19 years.
There's a dedicated path in the OSM rendering pipeline just for producing the coastline+ocean backdrops, completely separate from the entire rest of the mapping logic, because it's just its own cursed little problem.
-
And then the readme goes to great lengths to say that most of the work it does is to ensure the coastlines are closed polygons, because the rest of the GIS pipeline requires them to be.
And then immediately says "except Antarctica, because in most projections it ends up being an open polygon, so there's hardcoded special cases for Antarctica to allow it to be an open polygon that gets closed using projection-specific hacks"
Mapping is like Unicode, deeply human and intricate and flawed
️ -
Is there a standard "ruler length" map people use for measuring squiggly lines?
-
@danderson are there separate maps for physical and legal coastlines?
-
I think I'm finding that I love these corners of computing that are about engaging with the real world and human systems, with all their history and traditions and exceptions. Where every broad statement comes with at least one asterisk, and that asterisk could be an entire master's thesis by itself (and probably was!). In a time where venture capital is busy trying to make humanity more computable, projects like Unicode and OpenStreetMap try to teach computers about humanity. It's beautiful.
-
@lolcat No idea! I assume a lot of coastline data comes from government survey data, at varying degrees of detail and accuracy. Part of the challenge of coastline representation, I'm told, is how to simplify the geometry as you zoom out without it looking weird or distorting shapes. Like, at what zoom level do you replace a cove with just a straight line? Last time I looked at this people had genuinely spent 10y of research on just that question, and it's still not fully solved.
-
@jann There's a million GIS systems for different purposes, so generally the answer to "is there a thing" is "yes", the question is which one makes sense for the application
OSM standardizes on the coastline being mean high water springs, an average springtime high-tide mark over the last 19 years. That's such a specific definition that I assume it's a somewhat widely accepted definition in cartography, but idk.
-
Agnieszka R. Turczyńskareplied to Dave Anderson last edited by
@danderson How the renderer treats objects like causeway to Lindisfarne?
-
Dave Andersonreplied to Agnieszka R. Turczyńska last edited by
@agturcz That's a human construct, so both OpenStreetMap and Google seem to represent it as a road segment, separate from the continent layer.
... I think, I would have to go digging into the OSM backend to check exactly what the feature is, but it's being rendered as a street.
-
@agturcz But there are plenty of natural features like that, which are part of the complexity of coastlines. Another is archipelagos of islands that are close together, as you zoom out you end up having to lie and distort the shapes of the islands to maintain a visual separation between them, because the eye dislikes "oops these are all one island now" much more than "I'm lying about the shape and area of these islands so that you can still perceive a tiny bit of water between them"
-
I also love that, in general, the articles on OSM wiki in German are 10x more detailed than the ones in English. National stereotypes are silly and unhelpful, but also there is definitely a meta-commentary here about the relative devotion to order and organization in different cultures. OSM owes a great deal to its German contributors.
-
Speaking of, the German OSM article on coastlines has some great little statements in it. (machine translated into English, because my German is quite poor)
> Coastlines are not cartographically trivial.
> When exactly does a coast turn into a river?
> When does an artificial construction belong to the coast? [link to immense article about mapping ports]
-
серафими многоꙮчитїиreplied to Dave Anderson last edited by
@danderson now I'm wondering how routing works for Noirmoutier in France which has a road except at high tides when it doesn't
-
Dave Andersonreplied to серафими многоꙮчитїи last edited by
@djm62 You can check the data! Assuming you mean the Passage du Gois that connects the island to the mainland, it has a bunch of tags:
type: secondary road
road number: D948
access: tidal
description: route submersible
hazard: flood
tidal: yesIt's up to higher level applications to decide how to interpret that
