9 ms·

I Hate Coordinate Systems

365 points by edent 6y ago

meibo 6y ago
I think what this mostly shows is the unimaginable complexity that many parts of life carry, even if I myself may not have insight into them. And it's probably an alright source for people whose geo data isn't looking right.
- _jal 6y ago
  I try to keep things like this in mind when I have that "I don't get why don't they just..." thought about a topic I know very little about.
solipsism 6y ago
This is a really unfortunate title that seems like it might be interesting to general math-inclined people, but instead is only interesting to a very small group of engineers working in a very specific domain.
- dmahr 6y ago
  You’re totally right, I made it for GIS people, not general math people. No idea why it’s on HN tbh
  - ecommerceguy 6y ago
    Lots of GIS people on HN it seems. My favorite projects were always raster based model projections for various infrastructure and demographics.
  - iudqnolq 6y ago
    As someone who isn't a GIS person but is trying to make a map-centric app (hiking/backpacking, with activity tracking and route planning), any suggestions for resources? I just finished implementing enough of the WMTS spec interact with the one provider I'm using so far (UK ordinance survey). The only library I could find to help that works on Android was proj4j, which I'm using to handle converting points to different CRSs. I'm making incredibly rookie mistakes like just realizing I need to project vector data into the same crs as the raster base map. I want to be able to transparently support changing the base map to one with a different crs and lines and points recorded in the previous crs show up in the "right places", but I can't figure out how that's possible. I think I've seen other apps do it.
- ajdlinux 6y ago
  I don't work anywhere near geospatial stuff, but this would have been helpful to me a few years ago when I did some hackathon projects that involved Google Maps, frantically trying to integrate a number of different data sources that used coordinate systems I'd never heard of. It's worth having at least a vague awareness of this stuff if you might ever touch anything map-related.
wyldfire 6y ago
[The Author] Hates Geographic Coordinate Systems
- gumby 6y ago
  Yeah, this wasn’t really what I expected but it was interesting nonetheless.
- michaelhoffman 6y ago
  As someone who works in genomics, all these issues apply to genomic coordinate systems as well—same concepts, different words.
  - LeifCarrotson 6y ago
    And as someone who works in CAD/CAM/robotics software, these issues apply to robotic coordinate systems, too: World coordinates, joint coordinates, polar coordinates (not the same as high-latitude GIS coordinates, think angle + distance instead of X and Y), tool coordinates, tool offsets, work offsets - all of those are just vectors of 3-6 numbers. Instead of esoteric international standards, and huge modern software packages, we have a lot of old machines, a lot of old code, and a lot more tendency towards 80s-style strategies: Post a sign stating "Do not dumb here!" and leave it all up to the operator to get it right. There's nothing like telling your 6-figure CNC or multi-axis robot to make a piece of carbide occupy some space where you currently have precision-ground hardened steel because someone left a G92 somewhere you didn't expect it to be. Though I suppose that errors can be at least that expensive in GIS if you're quoting civil engineering or real estate...
    - ktpsns 6y ago
      As someone in computational physics, let me assure you it's the same there. The one thing is doing math right and having the correct idea about coordinates. The other is being able to tell the computer about these ideas! Many people cannot code, so are not aware of array or vector facilities of their programming language/ecosystem they use (or rather have to use, because it's always have been like that). This Phd comic is pretty acurate about the situation: http://phdcomics.com/comics.php?f=1689 http://phdcomics.com/comics.php?f=1689 Now imagine being a newcomer with neither programming experience nor domain specific knowledge and asked to add some feature to such a legacy codebase. This is the standard case in science.
anonytrary 6y ago
Yes, the Earth is curved. Assuming it is flat is incompatible with having a robust and accurate geospatial search. You must use sines and cosines if you want to do geospatial indexing correctly!
- techbio 6y ago
  Sines, cosines, and the Haversine formula only get you so far (or approximately close). The Earth is neither flat, nor perfectly spherical, but an oblate spheroid. The coordinate systems referenced in TFA are defined to adapt for this, but in different ways depending on the geospatial application, source data collection, map projection, and other factors. A reusable coordinate dataset will include not just latitude and longitude, but also the reference ellipsoid to which it approximates the Earth's surface, and the direction of North used at the location to calibrate sufficient dimensions to pinpoint relative location. Not in a word, ideal, and thus, TFA.
  - bigiain 6y ago
    Mt Everest and The Marianas Trench would like a word with you about The Earth being an oblate spheroid... Everything is an approximation.
    - techbio 6y ago
      True the Earth is not flat in many different ways, and also, the parts are moving. Edit: if you can put either of those extraordinary places in touch with me, that's fine also.
bagels 6y ago
Okay, now do space reference frames. https://naif.jpl.nasa.gov/pub/naif/toolkit_docs/Tutorials/pdf/individual_docs/17_frames_and_coordinate_systems.pdf https://naif.jpl.nasa.gov/pub/naif/toolkit_docs/Tutorials/pd... https://en.wikipedia.org/wiki/Earth-centered_inertial https://en.wikipedia.org/wiki/Earth-centered_inertial vs https://en.wikipedia.org/wiki/ECEF https://en.wikipedia.org/wiki/ECEF
- cossatot 6y ago
  And then think about tectonic plate motions and that any point affixed to a continent is moving with regards to an ECEF coordinate system.
- Sharlin 6y ago
  Ugh, I get flashbacks to one of my astronomy intro courses and having to memorize conversion formulas between horizontal (azimuth/altitude), equatorial (RA or hour angle/declination), and ecliptic (ecliptic long/lat) coordinate systems. Fun times.
- panzagl 6y ago
  No kidding, this article is someone complaining about easy mode...
- accurrent 6y ago
  While in uni I once ended up having to chase an autonomous boat as it decided to head out too deep waters and do a hoky-poky dance because one guy wrote controls code in NED coordinate convention and another one issued commands in ENU convention.
  - segfaultbuserr 6y ago
    Write a blog post if you can. Chasing a boat on the water because the program used the wrong coordinate system - sounds like an amusing debugging story.
perrygeo 6y ago
The problem with a self-described "problem-based guide" to coordinate systems is that it relies on the user to identify the problem in the first place. Reality is that these are expensive errors that can be avoided by systematic planning but made far too easy by default software options. I've seen weeks/hundreds of thousands of dollars wasted by proceeding with data analysis without properly considering the coordinate reference system. If spatial accuracy is a concern and - your data is vector-based lines or a polygons, there is no guarantee that the edges between your vertices will follow the same trajectory in a different coordinate system. This means that the spatial relationship (intersection, overlap, etc) between two geographic objects depends on the coordinate system and the relative density of points. - your data is raster-based, you need to resample pixels values to translate between coordinate systems. If your pixels represent absolute/measured values (eg population), you need to consider this carefully. - In both vector and raster, you need to deal with dateline and pole issues. The earth is not a cartesian plane. - In both cases, you also need to consider more than the projection. You also deal with datums which define the relationship between the actual earth surface and the theoretical sphere/spheroid the projections are based on. Get this wrong and your results might "look" OK but be meters off - you might not see it visually but it will cause problems for high-precision analysis.
- jeffbee 6y ago
  I only have a little bit of GIS abilities but I found the page interesting. I think it's suitable for people as long as they have non-zero exposure to geographic coordinate systems. One thing about this topic that makes it hard to learn is software ergonomics. It seems like a lot of tools can work with GeoJSON now, and as I understand it GeoJSON has simply declared that there is only one coordinate system and every file uses that one. Declaring a different CRS is not allowed in the standard. However, ESRI ArcGIS will emit GeoJSON with any CRS you want. In fact if you don't tell it specifically to use the GeoJSON standard, it will instead emit GeoJSON in EPSG 3857, which is not just irritating but plainly incorrect. It further confuses the issue that the 3857 system has six other alternate names, each seemingly as well known as any of the others. It doesn't help that geopandas will cheerfully perform a spatial join of two dataframes despite the fact that it knows full well that the two frames have incompatible coordinate reference systems. It will warn about this condition but it will do the join anyway and you might not see the warning after a million lines of console spew.
garaetjjte 6y ago
Related xkcd: https://xkcd.com/977/ https://xkcd.com/977/
me2i81 6y ago
The most common error I've seen in this area is that most GIS systems use (Lat, Lng) but most systems written by non-GIS people use (Lng, Lat), which they think of as (x, y).
- iudqnolq 6y ago
  I think you have that backwards? In my very limited experience most GIS systems use ENU ordering (east, north up), and easting corresponds to longitude.
  - hurrrrrrrr 6y ago
    I think he has that correct. Ime it's split on east/north vs. north/east but wgs84 is very dominant with lat/lon.
    - iudqnolq 6y ago
      Huh, thanks. Researching this more, I think you're right. I thought the Open Geospatial Consortium's standards would reflect dominant practice, that turns out to have been the wrong assumption [1] The only geospatial software I've used is Proj4J, which normalizes with a function called toENU. I guess I've only encountered weird stuff then. [1]: https://lists.opengeospatial.org/pipermail/coordtran.wg/2006-December/001877.html https://lists.opengeospatial.org/pipermail/coordtran.wg/2006...
Waterluvian 6y ago
I love geographic coordinate systems. My one wish is if the GeoJSON spec as loosened so that it technically allows non lat/long coordinates. I use it all the time for indoor flatland scenarios where units are metres. It all works fine. But it's technically not allowed. On rare occasion a GIS tool will completely balk at the data being CRS.Simple
rurounijones 6y ago
My main hatred is the simple one Lat/Lon or Lon/Lat, and what about altitude? especially when libraries have parameters like x, y, z and you have to triple-check which value is going into which vaguely named, not well documented, field. Because there are different conventions on which value goes to which depending on your background.
- Rebelgecko 6y ago
  Another fun one is altitude vs elevation (then of course it comes down to whether you're using a WGS84 ellipsoid or something more esoteric)
- yxhuvud 6y ago
  Not to mention that different coordinate systems also use different orderings.
- cjlm 6y ago
  Good read on this: https://macwright.com/lonlat/ https://macwright.com/lonlat/
aimor 6y ago
Does anyone here know a good source that gives details on how to redefine coordinate systems without relying on existing software implementations?
- dmahr 6y ago
  Snyder is the canonical source that most use. Free download from USGS. Once you see the math you’ll understand why you want to use libraries. https://pubs.er.usgs.gov/publication/pp1395 https://pubs.er.usgs.gov/publication/pp1395
  - ngcc_hk 6y ago
    Same when I found out the need to use fortran for astro calc ...
  - to11mtm 6y ago
    HA! The upshot of Snyder is that it includes -samples- for everything, which is really useful if you ever wind up having to deal with an Albers Conic over 3 states in the US, that also is using an elipsoid typically used in India. Oh, also it's 2010 and you can't afford actual GIS tools. You have Google Earth, Visual Studio, AutoCAD, and Microstation. Good Luck.
ConradKilroy 6y ago
Wow, thanks for posting this about geospatial Coordinate reference systems (CRSs)!
ecommerceguy 6y ago
Back in the day.... I tried to stay in HARN as much as I could, or at the very least something local. There were some minor precision problems but never anything drastic.
jandrewrogers 6y ago
The tragedy is that good implementations are rare because correct implementations are really difficult. Everyone grossly underestimates how difficult it is to implement precise, correct behavior globally in software without weird emergent problems. Odd edge cases are the default state of reality for geospatial unless you are extremely careful. Even if you are diligent and assume a global high-precision ellipsoid as your Earth model, you still have to ensure that the computational geometry is correct in that context for every possible use case. Operations on real surfaces are difficult to implement on discrete computers. As a simple example, correct ellipsoid computation in double precision sometimes can't even be computed in quad precision, never mind double, yet most efforts at computational geometry limit themselves to double precision because that is what computers support natively. Manipulating relationships in real-world coordinate systems is not for amateurs.
- jeffbee 6y ago
  My impression has been that basically ESRI seems to have more correctness and the open source stuff has done the first 90% and left the second 90% on the table. What do you think of libraries like shapely, if you've tried it? I make maps of the county where I live and there are parcels that ESRI doesn't seem to mind but shapely refuses to ponder. An example: 1: https://drive.google.com/file/d/1Fw5SdZ-aAiHrZERnIkVgll_OFntm5IC5/view?usp=sharing https://drive.google.com/file/d/1Fw5SdZ-aAiHrZERnIkVgll_OFnt...
  - terryf 6y ago
    The open source stuff is mostly really fidgety and works correctly with a rather narrow set of inputs. To make most libraries be actually easily usable for a wide range of inputs is a tremendous amount of work. The core of them mostly do the thing they claim to do, but if you feed them data that's just slightly off, then it'll break in interesting ways. Speaking based on a lot of experience[1]. Nothing in the geospatial or environmental data world is easy and there will always be an unexpected problem lurking right behind the corner, regardless of how many of them you've already fixed or worked around. Still, the complexity of the problem space also makes it fascinating. As an example, the definition and measurement of something that's seemingly very simple: mean seal level is actually tremendously complicated if you think about it. The Wikipedia article talks about averaging over decades of measurements to get it right for example. [1] I'm a part of the team behind https://data.planetos.com/datasets https://data.planetos.com/datasets
sadfev 6y ago
Me too, confusing to transform from one to another, error prone and computationally awkward
ajmarcic 6y ago
I expected this to be another post about geometric algebra (or maybe tensor algebra)
nyanpasu64 6y ago
I expected this to be about coordinate systems used for video game models (y up vs z up). Turns out it's geospatial positioning.
- kaoD 6y ago
  Same here. I absolutely hate 3D coordinate systems, more specifically handedness and how easy it is to have correct results but wrong math due to two wrongs canceling each other.
stellersjay 6y ago
My two cents. Missed opportunity for the domain ihatecoordinate.systems
furstenheim 6y ago
This has really struck home.I'm missing another one. Your dataset is tilted 90° and in some other part of the globe: you probably swapped latitude and longitude. Tip: latitude sounds like "altitude" so it has to be the y coordinate
- longtom 6y ago
  It's such misnomers. You'd think the "lat" prefix comes from lateral and means an axis that goes sideways, but that's not the case.
  - jeromenerf 6y ago
    Not to mention longitude and latitude are most often used for spherical coordinate systems, as angles, where x, y, z are usually meant for Cartesian coordinates, as distances, in a projected representation. As a French nitpick, the "sideways axis" is actually not going sideways, but pointing right at us :)
  - gowld 6y ago
    There's an unavoidable duality (An axis is definitionally perpendicular to the lines of constant value) The core of the problem is that the system talks about "lines" too much. In regular math, we have axes, but we don't think much about a point being at the "intersection of the x=3 line and the y=5 line", it's just "horizontally 3 and vertically 5". But in GIS we visualize a full grid, with "lines of latitude" instead of "lateral deviation".
- yxhuvud 6y ago
  Very easy to do, especially if you are dealing with both lat/lon and other coordinate systems that may use a different order.
at_a_remove 6y ago
This sort of thing bites me every so often. A client wants something in KML, which I have used ... once? The export files look suspiciously small. Converting back gives zero results. No error on the export. Mentioned nowhere is that you need WGS 1984 in the mix or you get nothing. That was fun.
n_kr 6y ago
Can I add another, personal gripe to this? US surveys use their own definition of a foot, which is different from everyone else's : https://en.wikipedia.org/wiki/Foot_(unit)#US_survey_foot https://en.wikipedia.org/wiki/Foot_(unit)#US_survey_foot
nneonneo 6y ago
You haven't seen pain until you've had to deal with crappy embedded systems that output "decimal DMS" data - i.e. ones in which 126.5904 means 126° 59' 04, rather than 126° 35' 25.44" as you might expect. Plus, the crappy documentation doesn't bother to tell you what coordinate system they're using so you get to guess! I wish I'd known about the little calculator they have on this site - which guesses the coordinate system in use - back when I was doing that project. Would have saved me a lot of headaches. The next pain point has got to be China's insane "GCJ-02" obfuscated coordinate system, as alluded to and linked on this site. I've had a lot of trouble getting basically anything GPS-related to spit out meaningful things while in that country (more accurately, matching basically anything up with maps while there). While it has been reverse-engineered, you always get to wonder whether the authorities are going to come after you for having GPS data that's too accurate... Coordinate systems are hard. GIS stuff is hard. I'm glad this website exists. Even if it's hard for people to identify the problem, this site at least makes one aware of the problems in the first place.
- jiggawatts 6y ago
  Ostensibly, GCJ-02 is for national security. Given that all of China's potential enemies have detailed satellite maps of the region and aren't in the slightest bit inconvenienced in their targeting accuracy because of GCJ-02, that's obviously false. Clearly, GCJ-02 is just simple, ordinary corruption, the type endemic to China. Someone convinced some politician with a bag of money to keep a legally mandated monopoly on mapping, locking out foreign competitors. PS: Japan does similar things, where for example Google Maps refuses to "download" maps of Japan because by law they are not allowed to make their own maps of Japan and must license the data from a local provider. That local provider does not allow offline use. The world of mapping is nuts.
  - maxerickson 6y ago
    Is it illegal for Google to make maps of Japan or just not a good investment? (that the licensed data they use has some restrictions is clear enough)
    - 6y ago
tigerlily 6y ago
Here was me thinking this would be a primer on differential forms.
dt3ft 6y ago
Clickbait, he doesn't hate coordinate systems, he loves them.
donquichotte 6y ago
I'm working on a system that transform data from Long/Lat/Alt (WGS84) to a local tangent plane (LTP) East/North/Up coordinate system. To go from WGS84 to LTP, the coordinates first need to be transformed to Earth-Centered Earth-Fixed (ECEF) coordinates and then translated and rotated. I was quite surprised to find that while it is straightforward to go from WGS84 to ECEF with a closed form, the other way, from ECEF to WGS84, needs to be computed iteratively [1] and is not trivial to find. [1] https://en.wikipedia.org/wiki/Geographic_coordinate_conversion#From_ECEF_to_ENU https://en.wikipedia.org/wiki/Geographic_coordinate_conversi...
- tonyarkles 6y ago
  Quick tip from someone else who was working on similar stuff last week: there’s a wonderful paper by Bowring from 1976 that has a one-step “iterative” version of the conversion. To verify that it worked right, I compared the output with the LLA output from my receiver and it matched to as many sigfigs as the receiver output.
sideshowb 6y ago
The rather shouty title makes me wonder if someone put Descartes before the hoarse
mcv 6y ago
Null Island is hardly the only problem with bad quality data. Quite often, when no exact location is known, but the country or region is known, the data may give coordinates at the center of that country or region, which can easily be hundreds or thousands of kilometers away from what you're looking for. Case in point: https://web.archive.org/web/20171201234812/https://consumerist.com/2016/04/11/this-farm-in-kansas-is-the-default-address-for-all-american-internet-users/ https://web.archive.org/web/20171201234812/https://consumeri...
winrid 6y ago
Man coordinate systems are hard, especially mapping between systems. I had to do this for the backend of a game I'm building - built my own geohash so I could add some optimizations I wanted (like storing nearby hashes for certain nodes etc). Created some fun graphics while testing the bounding box queries, like: https://jvm-gaming.org/uploads/default/optimized/2X/d/d7c48976d97b8da9ab0a3a17c2513ca28e9dc426_2_1380x680.png https://jvm-gaming.org/uploads/default/optimized/2X/d/d7c489...
necovek 6y ago
The thing that annoys me as a software engineer with background in mathematics is that it's hard to find pure transformation formulae between these geodesic coordinate systems, or at least was 5+ years ago when I last played with it. I get all the gotchas, but formula is much easier for me to get my head around rather than these articles which are centered around me having to learn a lot of terminology just to be able to do a transformation between coordinate systems. Even if that formula was accompanied with a correction table for particular use cases.

365 points by edent 6y ago

I think what this mostly shows is the unimaginable complexity that many parts of life carry, even if I myself may not have insight into them. And it's probably an alright source for people whose geo data isn't looking right.

_jal 6y ago
I try to keep things like this in mind when I have that "I don't get why don't they just..." thought about a topic I know very little about.

solipsism 6y ago

This is a really unfortunate title that seems like it might be interesting to general math-inclined people, but instead is only interesting to a very small group of engineers working in a very specific domain.

dmahr 6y ago
You’re totally right, I made it for GIS people, not general math people. No idea why it’s on HN tbh
- ecommerceguy 6y ago
  Lots of GIS people on HN it seems. My favorite projects were always raster based model projections for various infrastructure and demographics.
- iudqnolq 6y ago
  As someone who isn't a GIS person but is trying to make a map-centric app (hiking/backpacking, with activity tracking and route planning), any suggestions for resources? I just finished implementing enough of the WMTS spec interact with the one provider I'm using so far (UK ordinance survey). The only library I could find to help that works on Android was proj4j, which I'm using to handle converting points to different CRSs. I'm making incredibly rookie mistakes like just realizing I need to project vector data into the same crs as the raster base map. I want to be able to transparently support changing the base map to one with a different crs and lines and points recorded in the previous crs show up in the "right places", but I can't figure out how that's possible. I think I've seen other apps do it.
ajdlinux 6y ago
I don't work anywhere near geospatial stuff, but this would have been helpful to me a few years ago when I did some hackathon projects that involved Google Maps, frantically trying to integrate a number of different data sources that used coordinate systems I'd never heard of. It's worth having at least a vague awareness of this stuff if you might ever touch anything map-related.

wyldfire 6y ago

[The Author] Hates Geographic Coordinate Systems

gumby 6y ago
Yeah, this wasn’t really what I expected but it was interesting nonetheless.
michaelhoffman 6y ago
As someone who works in genomics, all these issues apply to genomic coordinate systems as well—same concepts, different words.
- LeifCarrotson 6y ago
  And as someone who works in CAD/CAM/robotics software, these issues apply to robotic coordinate systems, too: World coordinates, joint coordinates, polar coordinates (not the same as high-latitude GIS coordinates, think angle + distance instead of X and Y), tool coordinates, tool offsets, work offsets - all of those are just vectors of 3-6 numbers. Instead of esoteric international standards, and huge modern software packages, we have a lot of old machines, a lot of old code, and a lot more tendency towards 80s-style strategies: Post a sign stating "Do not dumb here!" and leave it all up to the operator to get it right. There's nothing like telling your 6-figure CNC or multi-axis robot to make a piece of carbide occupy some space where you currently have precision-ground hardened steel because someone left a G92 somewhere you didn't expect it to be. Though I suppose that errors can be at least that expensive in GIS if you're quoting civil engineering or real estate...
  - ktpsns 6y ago
    As someone in computational physics, let me assure you it's the same there. The one thing is doing math right and having the correct idea about coordinates. The other is being able to tell the computer about these ideas! Many people cannot code, so are not aware of array or vector facilities of their programming language/ecosystem they use (or rather have to use, because it's always have been like that). This Phd comic is pretty acurate about the situation: http://phdcomics.com/comics.php?f=1689 http://phdcomics.com/comics.php?f=1689 Now imagine being a newcomer with neither programming experience nor domain specific knowledge and asked to add some feature to such a legacy codebase. This is the standard case in science.

anonytrary 6y ago

Yes, the Earth is curved. Assuming it is flat is incompatible with having a robust and accurate geospatial search. You must use sines and cosines if you want to do geospatial indexing correctly!

techbio 6y ago
Sines, cosines, and the Haversine formula only get you so far (or approximately close). The Earth is neither flat, nor perfectly spherical, but an oblate spheroid. The coordinate systems referenced in TFA are defined to adapt for this, but in different ways depending on the geospatial application, source data collection, map projection, and other factors. A reusable coordinate dataset will include not just latitude and longitude, but also the reference ellipsoid to which it approximates the Earth's surface, and the direction of North used at the location to calibrate sufficient dimensions to pinpoint relative location. Not in a word, ideal, and thus, TFA.
- bigiain 6y ago
  Mt Everest and The Marianas Trench would like a word with you about The Earth being an oblate spheroid... Everything is an approximation.
  - techbio 6y ago
    True the Earth is not flat in many different ways, and also, the parts are moving. Edit: if you can put either of those extraordinary places in touch with me, that's fine also.

bagels 6y ago

Okay, now do space reference frames. https://naif.jpl.nasa.gov/pub/naif/toolkit_docs/Tutorials/pdf/individual_docs/17_frames_and_coordinate_systems.pdf https://naif.jpl.nasa.gov/pub/naif/toolkit_docs/Tutorials/pd... https://en.wikipedia.org/wiki/Earth-centered_inertial https://en.wikipedia.org/wiki/Earth-centered_inertial vs https://en.wikipedia.org/wiki/ECEF https://en.wikipedia.org/wiki/ECEF

cossatot 6y ago
And then think about tectonic plate motions and that any point affixed to a continent is moving with regards to an ECEF coordinate system.
Sharlin 6y ago
Ugh, I get flashbacks to one of my astronomy intro courses and having to memorize conversion formulas between horizontal (azimuth/altitude), equatorial (RA or hour angle/declination), and ecliptic (ecliptic long/lat) coordinate systems. Fun times.
panzagl 6y ago
No kidding, this article is someone complaining about easy mode...
accurrent 6y ago
While in uni I once ended up having to chase an autonomous boat as it decided to head out too deep waters and do a hoky-poky dance because one guy wrote controls code in NED coordinate convention and another one issued commands in ENU convention.
- segfaultbuserr 6y ago
  Write a blog post if you can. Chasing a boat on the water because the program used the wrong coordinate system - sounds like an amusing debugging story.

perrygeo 6y ago

The problem with a self-described "problem-based guide" to coordinate systems is that it relies on the user to identify the problem in the first place. Reality is that these are expensive errors that can be avoided by systematic planning but made far too easy by default software options. I've seen weeks/hundreds of thousands of dollars wasted by proceeding with data analysis without properly considering the coordinate reference system. If spatial accuracy is a concern and - your data is vector-based lines or a polygons, there is no guarantee that the edges between your vertices will follow the same trajectory in a different coordinate system. This means that the spatial relationship (intersection, overlap, etc) between two geographic objects depends on the coordinate system and the relative density of points. - your data is raster-based, you need to resample pixels values to translate between coordinate systems. If your pixels represent absolute/measured values (eg population), you need to consider this carefully. - In both vector and raster, you need to deal with dateline and pole issues. The earth is not a cartesian plane. - In both cases, you also need to consider more than the projection. You also deal with datums which define the relationship between the actual earth surface and the theoretical sphere/spheroid the projections are based on. Get this wrong and your results might "look" OK but be meters off - you might not see it visually but it will cause problems for high-precision analysis.

jeffbee 6y ago
I only have a little bit of GIS abilities but I found the page interesting. I think it's suitable for people as long as they have non-zero exposure to geographic coordinate systems. One thing about this topic that makes it hard to learn is software ergonomics. It seems like a lot of tools can work with GeoJSON now, and as I understand it GeoJSON has simply declared that there is only one coordinate system and every file uses that one. Declaring a different CRS is not allowed in the standard. However, ESRI ArcGIS will emit GeoJSON with any CRS you want. In fact if you don't tell it specifically to use the GeoJSON standard, it will instead emit GeoJSON in EPSG 3857, which is not just irritating but plainly incorrect. It further confuses the issue that the 3857 system has six other alternate names, each seemingly as well known as any of the others. It doesn't help that geopandas will cheerfully perform a spatial join of two dataframes despite the fact that it knows full well that the two frames have incompatible coordinate reference systems. It will warn about this condition but it will do the join anyway and you might not see the warning after a million lines of console spew.

garaetjjte 6y ago

Related xkcd: https://xkcd.com/977/ https://xkcd.com/977/

me2i81 6y ago

The most common error I've seen in this area is that most GIS systems use (Lat, Lng) but most systems written by non-GIS people use (Lng, Lat), which they think of as (x, y).

iudqnolq 6y ago
I think you have that backwards? In my very limited experience most GIS systems use ENU ordering (east, north up), and easting corresponds to longitude.
- hurrrrrrrr 6y ago
  I think he has that correct. Ime it's split on east/north vs. north/east but wgs84 is very dominant with lat/lon.
  - iudqnolq 6y ago
    Huh, thanks. Researching this more, I think you're right. I thought the Open Geospatial Consortium's standards would reflect dominant practice, that turns out to have been the wrong assumption [1] The only geospatial software I've used is Proj4J, which normalizes with a function called toENU. I guess I've only encountered weird stuff then. [1]: https://lists.opengeospatial.org/pipermail/coordtran.wg/2006-December/001877.html https://lists.opengeospatial.org/pipermail/coordtran.wg/2006...

Waterluvian 6y ago

I love geographic coordinate systems. My one wish is if the GeoJSON spec as loosened so that it technically allows non lat/long coordinates. I use it all the time for indoor flatland scenarios where units are metres. It all works fine. But it's technically not allowed. On rare occasion a GIS tool will completely balk at the data being CRS.Simple

rurounijones 6y ago

My main hatred is the simple one Lat/Lon or Lon/Lat, and what about altitude? especially when libraries have parameters like x, y, z and you have to triple-check which value is going into which vaguely named, not well documented, field. Because there are different conventions on which value goes to which depending on your background.

Rebelgecko 6y ago
Another fun one is altitude vs elevation (then of course it comes down to whether you're using a WGS84 ellipsoid or something more esoteric)
yxhuvud 6y ago
Not to mention that different coordinate systems also use different orderings.
cjlm 6y ago
Good read on this: https://macwright.com/lonlat/ https://macwright.com/lonlat/

aimor 6y ago

Does anyone here know a good source that gives details on how to redefine coordinate systems without relying on existing software implementations?

dmahr 6y ago
Snyder is the canonical source that most use. Free download from USGS. Once you see the math you’ll understand why you want to use libraries. https://pubs.er.usgs.gov/publication/pp1395 https://pubs.er.usgs.gov/publication/pp1395
- ngcc_hk 6y ago
  Same when I found out the need to use fortran for astro calc ...
- to11mtm 6y ago
  HA! The upshot of Snyder is that it includes -samples- for everything, which is really useful if you ever wind up having to deal with an Albers Conic over 3 states in the US, that also is using an elipsoid typically used in India. Oh, also it's 2010 and you can't afford actual GIS tools. You have Google Earth, Visual Studio, AutoCAD, and Microstation. Good Luck.

ConradKilroy 6y ago

Wow, thanks for posting this about geospatial Coordinate reference systems (CRSs)!

ecommerceguy 6y ago

Back in the day.... I tried to stay in HARN as much as I could, or at the very least something local. There were some minor precision problems but never anything drastic.

jandrewrogers 6y ago

The tragedy is that good implementations are rare because correct implementations are really difficult. Everyone grossly underestimates how difficult it is to implement precise, correct behavior globally in software without weird emergent problems. Odd edge cases are the default state of reality for geospatial unless you are extremely careful. Even if you are diligent and assume a global high-precision ellipsoid as your Earth model, you still have to ensure that the computational geometry is correct in that context for every possible use case. Operations on real surfaces are difficult to implement on discrete computers. As a simple example, correct ellipsoid computation in double precision sometimes can't even be computed in quad precision, never mind double, yet most efforts at computational geometry limit themselves to double precision because that is what computers support natively. Manipulating relationships in real-world coordinate systems is not for amateurs.

jeffbee 6y ago
My impression has been that basically ESRI seems to have more correctness and the open source stuff has done the first 90% and left the second 90% on the table. What do you think of libraries like shapely, if you've tried it? I make maps of the county where I live and there are parcels that ESRI doesn't seem to mind but shapely refuses to ponder. An example: 1: https://drive.google.com/file/d/1Fw5SdZ-aAiHrZERnIkVgll_OFntm5IC5/view?usp=sharing https://drive.google.com/file/d/1Fw5SdZ-aAiHrZERnIkVgll_OFnt...
- terryf 6y ago
  The open source stuff is mostly really fidgety and works correctly with a rather narrow set of inputs. To make most libraries be actually easily usable for a wide range of inputs is a tremendous amount of work. The core of them mostly do the thing they claim to do, but if you feed them data that's just slightly off, then it'll break in interesting ways. Speaking based on a lot of experience[1]. Nothing in the geospatial or environmental data world is easy and there will always be an unexpected problem lurking right behind the corner, regardless of how many of them you've already fixed or worked around. Still, the complexity of the problem space also makes it fascinating. As an example, the definition and measurement of something that's seemingly very simple: mean seal level is actually tremendously complicated if you think about it. The Wikipedia article talks about averaging over decades of measurements to get it right for example. [1] I'm a part of the team behind https://data.planetos.com/datasets https://data.planetos.com/datasets

sadfev 6y ago

Me too, confusing to transform from one to another, error prone and computationally awkward

ajmarcic 6y ago

I expected this to be another post about geometric algebra (or maybe tensor algebra)

nyanpasu64 6y ago

I expected this to be about coordinate systems used for video game models (y up vs z up). Turns out it's geospatial positioning.

kaoD 6y ago
Same here. I absolutely hate 3D coordinate systems, more specifically handedness and how easy it is to have correct results but wrong math due to two wrongs canceling each other.

stellersjay 6y ago

My two cents. Missed opportunity for the domain ihatecoordinate.systems

furstenheim 6y ago

This has really struck home.I'm missing another one. Your dataset is tilted 90° and in some other part of the globe: you probably swapped latitude and longitude. Tip: latitude sounds like "altitude" so it has to be the y coordinate

longtom 6y ago
It's such misnomers. You'd think the "lat" prefix comes from lateral and means an axis that goes sideways, but that's not the case.
- jeromenerf 6y ago
  Not to mention longitude and latitude are most often used for spherical coordinate systems, as angles, where x, y, z are usually meant for Cartesian coordinates, as distances, in a projected representation. As a French nitpick, the "sideways axis" is actually not going sideways, but pointing right at us :)
- gowld 6y ago
  There's an unavoidable duality (An axis is definitionally perpendicular to the lines of constant value) The core of the problem is that the system talks about "lines" too much. In regular math, we have axes, but we don't think much about a point being at the "intersection of the x=3 line and the y=5 line", it's just "horizontally 3 and vertically 5". But in GIS we visualize a full grid, with "lines of latitude" instead of "lateral deviation".
yxhuvud 6y ago
Very easy to do, especially if you are dealing with both lat/lon and other coordinate systems that may use a different order.

at_a_remove 6y ago

This sort of thing bites me every so often. A client wants something in KML, which I have used ... once? The export files look suspiciously small. Converting back gives zero results. No error on the export. Mentioned nowhere is that you need WGS 1984 in the mix or you get nothing. That was fun.

n_kr 6y ago

Can I add another, personal gripe to this? US surveys use their own definition of a foot, which is different from everyone else's : https://en.wikipedia.org/wiki/Foot_(unit)#US_survey_foot https://en.wikipedia.org/wiki/Foot_(unit)#US_survey_foot

nneonneo 6y ago

You haven't seen pain until you've had to deal with crappy embedded systems that output "decimal DMS" data - i.e. ones in which 126.5904 means 126° 59' 04, rather than 126° 35' 25.44" as you might expect. Plus, the crappy documentation doesn't bother to tell you what coordinate system they're using so you get to guess! I wish I'd known about the little calculator they have on this site - which guesses the coordinate system in use - back when I was doing that project. Would have saved me a lot of headaches. The next pain point has got to be China's insane "GCJ-02" obfuscated coordinate system, as alluded to and linked on this site. I've had a lot of trouble getting basically anything GPS-related to spit out meaningful things while in that country (more accurately, matching basically anything up with maps while there). While it has been reverse-engineered, you always get to wonder whether the authorities are going to come after you for having GPS data that's too accurate... Coordinate systems are hard. GIS stuff is hard. I'm glad this website exists. Even if it's hard for people to identify the problem, this site at least makes one aware of the problems in the first place.

jiggawatts 6y ago
Ostensibly, GCJ-02 is for national security. Given that all of China's potential enemies have detailed satellite maps of the region and aren't in the slightest bit inconvenienced in their targeting accuracy because of GCJ-02, that's obviously false. Clearly, GCJ-02 is just simple, ordinary corruption, the type endemic to China. Someone convinced some politician with a bag of money to keep a legally mandated monopoly on mapping, locking out foreign competitors. PS: Japan does similar things, where for example Google Maps refuses to "download" maps of Japan because by law they are not allowed to make their own maps of Japan and must license the data from a local provider. That local provider does not allow offline use. The world of mapping is nuts.
- maxerickson 6y ago
  Is it illegal for Google to make maps of Japan or just not a good investment? (that the licensed data they use has some restrictions is clear enough)
  - 6y ago

tigerlily 6y ago

Here was me thinking this would be a primer on differential forms.

dt3ft 6y ago

Clickbait, he doesn't hate coordinate systems, he loves them.

donquichotte 6y ago

I'm working on a system that transform data from Long/Lat/Alt (WGS84) to a local tangent plane (LTP) East/North/Up coordinate system. To go from WGS84 to LTP, the coordinates first need to be transformed to Earth-Centered Earth-Fixed (ECEF) coordinates and then translated and rotated. I was quite surprised to find that while it is straightforward to go from WGS84 to ECEF with a closed form, the other way, from ECEF to WGS84, needs to be computed iteratively [1] and is not trivial to find. [1] https://en.wikipedia.org/wiki/Geographic_coordinate_conversion#From_ECEF_to_ENU https://en.wikipedia.org/wiki/Geographic_coordinate_conversi...

tonyarkles 6y ago
Quick tip from someone else who was working on similar stuff last week: there’s a wonderful paper by Bowring from 1976 that has a one-step “iterative” version of the conversion. To verify that it worked right, I compared the output with the LLA output from my receiver and it matched to as many sigfigs as the receiver output.

sideshowb 6y ago

The rather shouty title makes me wonder if someone put Descartes before the hoarse

mcv 6y ago

Null Island is hardly the only problem with bad quality data. Quite often, when no exact location is known, but the country or region is known, the data may give coordinates at the center of that country or region, which can easily be hundreds or thousands of kilometers away from what you're looking for. Case in point: https://web.archive.org/web/20171201234812/https://consumerist.com/2016/04/11/this-farm-in-kansas-is-the-default-address-for-all-american-internet-users/ https://web.archive.org/web/20171201234812/https://consumeri...

winrid 6y ago

Man coordinate systems are hard, especially mapping between systems. I had to do this for the backend of a game I'm building - built my own geohash so I could add some optimizations I wanted (like storing nearby hashes for certain nodes etc). Created some fun graphics while testing the bounding box queries, like: https://jvm-gaming.org/uploads/default/optimized/2X/d/d7c48976d97b8da9ab0a3a17c2513ca28e9dc426_2_1380x680.png https://jvm-gaming.org/uploads/default/optimized/2X/d/d7c489...

necovek 6y ago

The thing that annoys me as a software engineer with background in mathematics is that it's hard to find pure transformation formulae between these geodesic coordinate systems, or at least was 5+ years ago when I last played with it. I get all the gotchas, but formula is much easier for me to get my head around rather than these articles which are centered around me having to learn a lot of terminology just to be able to do a transformation between coordinate systems. Even if that formula was accompanied with a correction table for particular use cases.