Mercator poisoning

March 1, 2010 12:44 PM

Why did Google Maps and the other popular online map tools pick the Mercator projection? If one scrolls further north or south from the equator — up to northern Canada, or down to Chile, for example — is it prohibitive or nonsensical to calculate a new projection to be drawn based on the viewer's current latitude, a new "great circle" that more correctly shows the area one is looking at? Is this is a usage issue for people who live far from the equator, one which forces them to pick maps with other projections more frequently?
posted by Blazecock Pileon to Science & Nature (7 answers total)

Why did Google Maps and the other popular online map tools pick the Mercator projection?

My hunch is that Mercator is the map format that the bulk of googleMaps' users are accustomed to seeing in general.
posted by EmpressCallipygos at 1:04 PM on March 1, 2010

Best answer: It's vastly more efficient for them to be able to pre-generate a single set of tiles for each zoom level, instead of having to recompute the projection every single time somebody drags their mouse. (Do you know how many hits a day Google Maps gets? I don't, but I bet it's a very large number.) And the projection has to be conformal, or else things will look all squished when you zoom in. Mercator is the simplest projection that meets those requirements, which is why pretty much everyone uses it.

(At one point, IIRC, Google's satellite view used an equirectangular projection instead. They switched that to Mercator as well because it caused a ridiculous amount of horizontal stretching at high latitudes.)

The only real downside to a conformal projection is that it distorts scale over large distances, but I'm sure the vast majority of usage of any of these mapping sites is for street-level or city-level views where the distortion is negligible.
posted by teraflop at 1:11 PM on March 1, 2010

Best answer: According to the Wikipedia article on the Mercator projection:

Google Maps currently uses a Mercator projection for its map images. Despite its obvious scale distortions at small scales, the projection is well-suited as an interactive world map that can be zoomed seamlessly to large-scale (local) maps, where there is relatively little distortion due to the projection's conformal nature. (Google Satellite Maps, on the other hand, used a plate carrée projection until July 22, 2005.)

The Google Maps tiling system displays most of the world at zoom level 0 as a single 256 pixel-square image, excluding the polar regions. Since the Mercator coordinate x varies over 2π, the other coordinate is limited to -π ≤ y ≤ π. Because

[math jargon I can't copy-paste]

the corresponding latitude extrema are φ = ±85.05113°. Latitude values outside this range are mapped using a different relationship that doesn't diverge at φ = ±90°.

Also, a Google employee clarified the reasons last August:

Hi John - Thanks for the feedback. Maps uses Mercator because it preserves angles. The first launch of Maps actually did not use Mercator, and streets in high latitude places like Stockholm did not meet at right angles on the map the way they do in reality. While this distorts a 'zoomed-out view' of the map, it allows close-ups (street level) to appear more like reality. The majority of our users are looking down at the street level for businesses, directions, etc... so we're sticking with this projection for now. In the meantime, you might want to look at our favorite 3D view of the world.

posted by Rhaomi at 1:12 PM on March 1, 2010 [5 favorites]

If you use google earth, the maps are projected onto a sphere. I don't really think it matters what projection they used so long as you could pan without interruption, those projections with gaps. They also would have wanted something without any angular distortion.

As you scroll upwards, the scale changes. If you're zoomed close in, it will change very slowly.

As far as issues for people, i doubt it, because it would only be an issue if you wanted to compare two different areas that were hugely different in latitude (or I guess if you're REALLY close to the poles)
posted by delmoi at 1:18 PM on March 1, 2010

Hmm, I was right about angular distortion, but I waited to long to post. I actually opened up google earth to see what the poles looked like. They actually look pretty bad for some reason, I think they use UV coordinate mapping over the entire earth, rather then having a patch at the poles so the textures are all distorted.
posted by delmoi at 1:22 PM on March 1, 2010

What teraflop said about costlry rendering. Google Maps (and many other services like it) use pre-rendered tilesets at various zoom levels, and the process of scrolling around is really just the process of loading pre-rendered images.

The rendering process is very expensive, computationally: throw a batch of geo data at a beefy server, and you have a 24-48 hour render job. Although Google can bring an awful lot of computational power to the table, re-rendering things in realtime just isn't worth it. Thus, one type of projection tailored to the most common use case, street-and-city level scrolling.
posted by verb at 5:31 PM on March 1, 2010

Latitude, longitude coordinates are fact. But, yes, rendering and maps do get a bit more complicated. Most of my experience is with integrated circuits, but it's the same principle. You've got coordinates in some form, and need to render the images/polygons onto those coordinates. But, at least they're flat. Throw a somewhat circular globe into the equation, things get a lot more complicated. Hence the different mapping models, some are better than others but it's all subjective. Depends on the info you want.
posted by hungrysquirrels at 4:24 AM on March 2, 2010