What happens to Google Maps when tectonic plates move? (2020)
nautil.usFWIW, global drift vectors by NASA: https://sideshow.jpl.nasa.gov/post/series.html
If you want to see an interesting one in my country:
https://sideshow.jpl.nasa.gov/post/links/KAIK.html
2016 Kaikōura earthquake: https://en.wikipedia.org/wiki/2016_Kaik%C5%8Dura_earthquake
There are deltas there of over 70cm.
That's insane when it usually moves 2cm per year at most.
It was insane. There was a massive die off of marine life. One movement they were happily swimming in the sea or crawling on the sand, next minute they were above the sea.
... what's the origin, when the entire surface of the Earth is sliding around?
From the article:
> WGS 84 is a global standard tied to no one plate. In essence, it is fixed to Earth’s deep interior.
Those drift vectors are based on GPS, which the article also says relies on WGS 84z
And on fixed surveying markers around the globe.
You'll see them hither and yon, but my favorite is the miniature Washington Monument buried next to the real one.
For anyone curious about the buried mini-monument: https://www.atlasobscura.com/places/washington-mini-monument
Thank you for turning me on to Atlas Obscura! I downloaded the app, and have already marked several places I want to visit, and others that I've been to. Looking forward to working with it more!
I happened to have been in Thailand recently, and came across these statues holding up traffic lights. I found it fascinating that a city council would devote resources to something like this. I'm looking forward to seeing other treasures that Atlas Obscura has.
That's gonna be a trip for archaeologists in a few thousand years.
> it is fixed to Earth’s deep interior
Which is molten and also not static?
The core of the earth is solid no?
It's based on a center of mass point of the earth.
The system is an example of an "ECEF" or Earth Centered Earth Fixed system.
Good writeup here -
https://en.m.wikipedia.org/wiki/Earth-centered,_Earth-fixed_...
Edit - fixed center statement.
I still don't quite get it. So the center is the barycenter of the planet, but that only gives you the point, not the axes.
The Z axis seems the most clearly defined as it's just aligned to the planet's rotation and the XY plane could be defined as just halving that, but the the rotation of the XY plane doesn't seem to be well defined. The article seems to fix it to the first meridian which is a land reference and subject to drift, so it would be inaccurate. Or there'd be some reference point in the south atlantic where the drift is zero and the rest is relative to it?
Further along this line of discussion, it's always been immensely puzzling to me how sun-relative solar system localization systems handle this sort of thing, since while it's possible to locate the sun and a few stars to track your absolute rotation and one translational axis, where the hell do they get the other two? One reference might be Earth signals I suppose, and possibly Jupiter? But they are all coplanar so the Z axis resolution from triangulation must be complete crap.
The IERS reference median is “the weighted average (in the least squares sense) of the reference meridians of the hundreds of ground stations contributing to the IERS network” [1].
Ah so it's just averaged out, that's a decent solution I suppose. But that still makes it fixed to the surface for the most part.
> still makes it fixed to the surface for the most part
Which makes sense. That’s what we’re trying to measure. You could fix it to a point on the core’s surface, but what wouldn’t be practically useful.
according to hairy banana theorem, at least one surface point won't be sliding sideways
For those looking for a clearer standard it's the International Terrestrial Reference System and Frame [0]. The better datums in use today are tied to a particular year (epoch) reference frame of this.
For instance Australia's GDA2020 [1] is based on ITRF2014 at epoch 2020.0. It replaced GDA94 which was based on ITRF1992 at epoch 1994.0. The difference is around 1.8 metres
[0] https://en.m.wikipedia.org/wiki/International_Terrestrial_Re...
[1] https://www.ga.gov.au/scientific-topics/positioning-navigati...
Discussions on similar submissions:
What Happens to Google Maps When Tectonic Plates Move? https://news.ycombinator.com/item?id=22146454 (January 25, 2020 — 2 points, 0 comments)
What happens to Google Maps when tectonic plates move? https://news.ycombinator.com/item?id=22145303 (January 24, 2020 — 188 points, 53 comments)
What Happens to Google Maps When Tectonic Plates Move? https://news.ycombinator.com/item?id=12216474 (August 3, 2016 — 2 points, 0 comments)
Last one didn't have much of a discussion. 0 comments :D
I get the joke. However to be clear, I included it because it contained a link to another site that was rather interesting because it went to SA instead of Nautil.
If you want to dive even deeper into datums around the world, there's a site dedicated to it: https://www.asprs.org/asprs-publications/grids-and-datums
You'll find many countries in there with a bit of history of their datums and how they arrived at their present state. It's a dry reading that can be fascinating at the same time.
>It's a dry reading that can be fascinating at the same time.
I'm a surveyor by trade and you hit the nail on the head.
One of my favorite parts about this stuff is the history, and how we humans have tried to understand the shape of our world better and better. And then how we've attempted to make that shape conform to a flat map so we can measure on it better.
Fascinating stuff.
Well since the author did not talk to Google we will not find out by reading this article.
In their defense, the tectonic plates will have moved a __lot__ before you can expect to get any response from Google support.
Those will be interesting patch notes for google maps
"we removed most of the gulf islands, part of BC, Washington, Oregon, and California" ...
At least that's guaranteed to be extremely far in the future.
Vs. "we removed Naples, and most other south-central Italian cities, due to the recent VEI 8 eruption of the Campe Flegree caldera complex...".
> At least that's guaranteed to be extremely far in the future.
a subduction zone quake in the PNW is already overdue (they appear to happen regularly with a periodicity of 3-600 years). The famous Atlantic article from several years ago made it clear that it can be expected at any time now.
I wont hold my breath becomes I wont cache my lat/lon.
I wanted to know how it's handled on OpenStreetMap, and here's how [1], on a first of April anyway.
If someone knows more on this, don't hesitate to share :-)
[1] https://blog.openstreetmap.org/2017/03/31/osm-plate-tectonic...
I was wondering if they were going to apply the correction to all objects, but as they hinted at in the post, only objects old enough to cause a half meter shift will probably be corrected.
Edit: Oh, April Fools. Seemed doable.
The post is an April Fools, to be clear.
edit: right? It seemed credible, I only noticed the date at the end. The part where they would hide the edit from history was incredibly shady though, and that's what made me look even closer. I would hope they would never do this, if they even could. Without this part, the post would not have looked like an April Fools, just an actually good idea.
I love how that post supposedly ended up being posted on the wrong day. XD
The article doesn't mention it, but how are the coordinates of property boundaries recorded in South California to take into account the drift?
In my relatively tectonically stable country there is a single coordinate system for the whole country.
I'd imagine you need to record distance to a local landmark or similar? And what happens when there are shifts that end up in roads split in two like in Turkey last year?
https://nationalpost.com/news/world/turkey-syria-earthquake-...
Property boundaries aren't delimited by GPS coordinates, but rather a surveyor's instructions on how to locate the boundary from a set of known landmarks. In most US states, the Public Land Survey System [1] gives "meridians" and "baselines" for each region serving as the regional center. Radiating out, every 6 miles, is a new "township," with its corners serving as the local landmark for property boundaries.
But so that means that you could lose land or gain land if enough earth moves and the relative change to the land marks is in your favor or not. But it's just not a big enough of a deal to do it a different way. I do wonder what the most significant occurrence of that is though.
Surveyors reference monuments on the ground. Since tectonic plates tend to move together then so so short lines and their corners.
It can get complicated on large tracts or lengths, but stuff like rail roads and freeways are composed of many small properties anyway.
The most significant occurrence is probably land loss due to tidal erosion and/or flooding.
I often think about a scifi story where after X million years, somehow familiar humanity survives, and two plates containing land with VERY different natural flora and fauna approach each other - for example, california, home of many native plant defenders, and australia. As they get closer than 50 miles, news would start to mention making sure not to cross-transplant animals and plants. Under 20 miles, the wind and storms would do some work already, but people on both sides may still be resisting. But the moment they actually touch for the first time, some great treaty may shimmer into enforceable existence, changing the future of that whole world.
Or else, they'd just build fences, forbid anyone from living on the coast, and maintain this artificial continental boundary forever...?
There are two cycles that govern human social change: the generational ~10 year cycle and the lifetime ~80 year cycle. Each generation adopts new values, and once a generation dies their values disappear. Tectonic movement is so slow that I think reality will be much more boring than your scifi story. The hundred thousand years where the two regions are "close enough" is more than enough time for the two peoples to both integrate and experience conflict hundreds of times over.
True. Although most science fiction which would take place over such a long timespan might involve methods of locking things down more firmly, and putting at least part of the world into a kind of stasis. This is almost necessary to explain how humanity could still be at a similar tech level to us now yet still be around for such a time period.
This will change if technology is developed which greatly extends human lifespans. Such technology is going to become necessary soon because of plummeting birth rates.
> after X million years [...] two plates containing land with VERY different natural flora and fauna approach each other - for example, california [...] and australia.
Yeah, nah - the flora jumped the gap 150 years ago already . . .
https://www.independent.com/2011/01/15/how-eucalyptus-came-c...
Humanity won't be alive by that time.
It's a interesting setting to explore in fiction, but has no chance of becoming reality.
That’s kind of misanthropic, talking in absolute terms. I could agree in terms of low possibilities but certainly not absolutes. I feel it does a disservice to the few who actually come up with genius solutions to problems humanity faces and the many who dedicate their lives to solving them too.
Is it, really? With current progress I don't see humanity surviving even another 500 yrs, honestly. The main intelligent race at that point is likely going to be something like a symbiont/cyborg that's part machine. At least nothing it'd consider human.
Millions of years is a long period of time. Humanity has only existed for roughly 200-300k. That's less then 1% of the projected time period here.
you must be fun at parties
Aren't parties just a celebration of not being dead yet?
Not being dead yet is well worth celebrating for most people though.
It's difficult to imagine human beings behaving in such a cautious way, when Australian trees have already been so widely transplanted across California that I grew up believing eucalyptus were native.
I remember having a hand-held Garmin GPS unit back in the '00s that I used on a trip to Berkeley from Los Angeles (one bonus was that the speedometer in my car didn’t work¹ so I was able to use it as a speedometer for the trip). The location that it showed on the map was consistently off by about 50 feet. I wonder now if this was a tectonic drift or just an inherent inaccuracy of the unit.
⸻
1. When I think back to that car, it’s amazing what a piece of junk it was. On the other hand, it was a convertible, so a lot of fun although my wife considered it a deathtrap.
The natural (or artificial) inaccuracy of GPS tends to present as an offset which remains relatively stable for many minutes.
About 10 years ago, I did a lot of canyoning/bushwalking in the Blue Mountains just outside of Sydney Australia. I’d have my Garmin GPSMAP switched on for the entire journey including the car trip. For the 20+ journeys, there would be overlapping paths where I was driving the highway through the mountains. Except none of the paths overlap. They were all smooth paths which matched the curvature of the road, but each one had a different offset, typically around 2–5 metres away from the map data. (The median of these 20+ paths was highly consistent with the map data.)
Back in the day GPS was intentionally offset so countries couldn't use it for weapons.. well except it's owner of course!
It seems like the intentional degradation didn't work as intended in the first place and the inaccuracy could be corrected with little effort - at least for someone who had the resources to use it for weapons systems, so it only affected civilian use cases.
> Because SA affects every GPS receiver in a given area almost equally, a fixed station with an accurately known position can measure the SA error values and transmit them to the local GPS receivers so they may correct their position fixes. This is called Differential GPS or DGPS.
https://en.wikipedia.org/wiki/Error_analysis_for_the_Global_...
Yes, differential GPS was used to correct the error. It wasn't useful for military purposes though: it took a LOT of time to set up a base station and allow it to collect readings and measure the SA error values. It wasn't a quick thing. And, you had to know exactly where the base station was in the first place.
If you're a civilian doing surveying work, it was fine: you stick the base station at a known location, let it work for a few hours, and then you had high-precision GPS in that area. If you're a military power attacking someone on their own ground, it's not so useful. Even worse if you're at sea.
related: a market summarizing what I know about China's intentional map offset laws, specifically on whether Google will continue following this (only 4 bettors, not very meaningful yet) https://manifold.markets/Ernie/will-google-maps-stop-followi...
Owning a junk car is a rite of passage to adulthood hood. You learn what to do when a car overheats, how to change a tire, replace a fuse, recognize the difference between burning oil and radiator fluid, how to jump start a car, and how to call a tow truck.
Important life skills
50 ft seems a little large for plate tectonics , plates move on the order of a centimeter per year [1]. (similar to fingernail growth!) [1] https://oceanservice.noaa.gov/facts/tectonics.html
I believe North America/ Australia is off by around 10 feet since GPS became a thing.
Australia's drift was taking into account and standards were updated back in 2020. Since 1994 the entire continent had moved 1.8 meters north-east.
https://www.icsm.gov.au/gda2020/what-changing-and-why
I'm sure the plate has drifted since the 2020 update and another update deployed when appropriate.
The 2011 Tohoku earthquake moved Japan's main island (Honshu) by 2.4m (8ft) in 6 minutes.
Fun facts I remember from looking into this in university some 9 years ago:
It’s not only the sideways movement of plates that causes a shift.
WGS 84 and other reference systems work with an underlying ellipsoid reference model. This ellipsoid is chosen to approximate the earth surface but it’s a simple shape and as such can’t account well for things like mountains and other irregularities in the shape of the earth.
Not only does that mean that your position on a mountain is less accurate, it also means that it becomes less accurate over time due to mountains growing.
These of course are tiny numbers so compared to a few cm shift of the plates it is nothing.
Another fun fact is that the ellipsoid also grows more inaccurate over time due to the earth rotating around a fixed axis.
The rotational force slowly causes the earth to flatten at the top and bottom (where the poles are) and to widen in the middle.
This too I assume is negligible
Speaking of ellipsoid, today I think it is measured via satellite. But in the past things had to be done by hand and were more local in nature. There’s a lot of history here as well since mapping has always been a very important task for governments.
So today in a lot of places a lot of data is still based on systems other than wgs84. Either because it’s historical data (e.g. property boundaries before gps was invented), or because they use a more localized reference system including a more localized ellipsoid that better matches specific country or state needs.
Perhaps one day we will use a big world wide look up table instead of a mathematical representation of the earth shape.
What I did at university was to compare a Germany-wide lookup grid with the mathematical approach of a Germany-wide reference system.
This lookup grid was created by a state agency and was more accurate because it could include more localized reference systems. In technical terms think of this as a pre calculation where each state can choose its most accurate method and then the results are put together into a country wide lookup table.
Last fun fact: The result of this comparison showed that the difference of a German-wide reference system compared to this collection of state-wide reference systems is up to four meters.
Edit: removed word inaccuracy, added difference. It’s all relative. What it mostly shows is that when working with geo data you need to know what source and what target reference systems to use. Otherwise things break
Assuming 30cm/pixel and O(75mm)/year plate motion, a several year update cycle ought to suffice for visual purposes (how frequently does Maps refresh now, anyway?)
Resurvey and publish a new datum. We’ve been doing it for a long time. This is fascinating stuff though because WGS84 predates this era of popular web map usage.
They update the data on a regular basis, like they had to do in Japan in 2011 after the tsunami. They even flew planes to capture imagery in the immediate aftermath.
Well up in canada we have 5 year old google street data, and nobody even updates business hours on 80k+ cities...
I’ll have to add this to my long list of technical problems when people request a teleporting super power.
Some related problems are the planet’s location in orbit / the solar system’s location relative to the entire universe and how landscape changes over time (trees/towns/mountains/rivers appearing and disappearing).
On another related note, it’s worth noting that the ability to pause time is a universe-ending super weapon. Even if it worked like you intended, you wouldn’t be able to use any senses or breathe without moving.
Similarly, super speed would cause every human movement to have deadly force. You would go insane from the relatively slow action of everything operating at standard speed. TV might be hours per frame instead of frames per second, with incomprehensible audio to match.
In the book Treason, the author explores that a tad: a character can make a time bubble around them, effectively freezing others. At one point, the character is being choked, freezes time, opens the assailant's hands and ducks out. When the bubble pops, in the frame of reference for the assailant, his hands moved too fast, shattering bone and destroying tissue. Read that decades ago as a kid and it was illuminating on the practical problems of time alteration
It's so crazy, never thought about this before
Yeah, amazing article
Google cancels the project?
> This screenshot represented my position in Google Maps while I was standing on my back deck.
In the age of OSINT: not smart.
(2020)
google kills google maps
The Australian tectonic plates move about 7cm a year, if you didn't adjust for it they'd be losing 2-5 thousand dollars a year as the fence moved in many suburban homes. Of course you'd gain it on the other side.
Google needs to be dealing with it constantly, you'd be able to observe more errors on Google maps, ie. look at Google directions and how often mistakes are made, if it wasn't updated monthly? I'd guess yearly is not enough.
Most of this article seems to be about other issues.
Yep the monuments on the plates here in USA have actual published velocities.
Which means yo can model what the coordinates should be after some previous survey and go check them.
Pretty cool!