[PROJ] Why is Web Marcator EPSG:3857 so popular for web maps (vs EPSG:4326)?

Greg Troxel gdt at lexort.com
Thu Nov 26 06:44:56 PST 2020


Idan Miara <idan at miara.com> writes:

> I was wondering if you can enlighten me regarding why Web Marcator (EPSG:3857)
> is so popular and used by virtually all major online map providers as
> opposed to WGS84 lat/lon (EPSG:4326)?

Two main reasons:

  3857 is at least almost conformal.  Angles are not distorted enough
  for humans to notice.  With 4326, there is serious distortion around
  me (42 N).  Basically nobody displays maps in pure geodetic
  coordinates because of this reason.  Yes, 3857 has variable size
  distortion, but if you zoom in enough then any given view has
  reasonably matching horizontal and vertical scale.

  Google started using 3857, and that evolved into it being part of the
  TMS spec, and the way that everybody does it.  While many think
  spherical pseudo-Mercator is icky (instead of real Mercator), having
  all TMS services use 3857 means that they can be displayed/switched
  without reprojecting tiles.

> 2. Users expect coordinates in 4326 and the transformation from 3857 to
> 4326 is rather fast (in comparison to ellipsoidal mercator).
> But if you save the coordinates in 4326 then you wouldn't need to transform
> at all.

Transforming coordinates for a point is easy.  Reprojecting rasters is
harder and comes out less well.

> 3. 3857 is "almost" conformal - I think that the normal use case is not
> marine navigation so it doesn't seem so important.

Have you looked at maps in 4326?  I hadn't until a few months ago when I
set that as my project CRS9 (or maybe is was NAD83(2011), 6319, same
thing for this discussion).  Square things on the ground don't look
square, and the same distance north-south appears much bigger on th map
than the east-west distance.

If you are on the equator 4326 preserves shapes. But not at higher
latitudes.

> 4. Both 3857 and 4326 have size distortion.

Yes but distortion within the same view is a bigger deal.  Size
distortion overall is not that important because people zoom to what
they want to see and then look at the scale bar.  Yes, looking at
near-global scale has intra-screen serious size distortion, in both.

> 5. You can't measure distances, areas or angles easier in 3857 as far as I
> know.

In the modern world, 'measure' is grab points and do math, so it doesn't
matter.  But looking at a map in 3857 you get approximately the right
idea of shapes, relative areas, and angles, and looking at 4326 (or any
other unprojected geodetic CRS) you don't.

> What am I missing? Is 3857 faster or more useful in any other way than 4326?

It's all about looking at the map and having things be mostly conformal.

Greg
-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 194 bytes
Desc: not available
URL: <http://lists.osgeo.org/pipermail/proj/attachments/20201126/231232fe/attachment.sig>


More information about the PROJ mailing list