7 replies to this topic

[Dennis McClendon]

I'm advising a friend who's working on a very cool, easy-to-use application for doing simple choropleth and dot location statistical maps. His market is academics and students in the social sciences making maps for publications and presentations, and newspapers and magazines who might not have the staff or expertise to work with ArcGIS or more robust packages.

Assuming he wants to give users the option of two or three default projections for world maps, which three should he choose?

[Adam Wilbert]

Winkel Tripel
Robinson
Obligatory Mercator

in that order.

[Charles Syrett]

Winkel Tripel
Robinson
Obligatory Mercator

in that order.

I'd buy into all three of these on the basis of their recognizability. National Geographic has made extensive use of the first two, and Google Maps has hauled ol' Mercator out of the cartographic doghouse so that -- once again -- it's the most-viewed projection of the world.

But to use Mercator as a base for statistical and choropleth mapping? It was developed for navigation, but areas are badly distorted (hence its political "incorrectness"). Distributions of points, etc., would give a misleading sense of relative density from one area to another, at least at small scales. I don't know what to suggest as a third (rarely do this kind of mapping); maybe someone else has an idea.

Charles Syrett
Map Graphics
http://www.mapgraphics.com

[François Goulet]

Good choice, but I'd replace Mercator by interrupted Goode homolosine.

Maybe not the most "accessible" of all projections, but I like it.

[Hans van der Maarel]

Good choice, but I'd replace Mercator by interrupted Goode homolosine.

Maybe not the most "accessible" of all projections, but I like it.

Same here, it just looks cool

[DaveB]

Good choice, but I'd replace Mercator by interrupted Goode homolosine.

Maybe not the most "accessible" of all projections, but I like it.

I almost said the same thing. (as case of "great mappers think alike"? )

[natcase]

Good choice, but I'd replace Mercator by interrupted Goode homolosine.

Maybe not the most "accessible" of all projections, but I like it.

I almost said the same thing. (as case of "great mappers think alike"? )

Wow. Yeah, that was my reaction too. Actually for human statistical maps it works quite well (see the Goode's World Atlas for lots of examples)

Do we need both Robinson and Winkel Tripel? I'd vote for the latter. And it may be necessary for graphic reasons to have a cylindrical projection (just gotta have that rectangle...). What about something less extreme than Mercator, like Miller? Or just straightforward Plate-Carée?

[M.Denil]

For clorepleth mapping neither Robinson nor Winkel Tripel are much use.

Neither is equal area, which should be a central concern for such use. True Nat Geo has used both, but that is hardly much of a recommendation: in fact, I generally prefer van der Gritten (which they used before 1988) to those two, but that is another story. Plate-Carée should not be an option.

There should be three equal area maps with distinct graphic shapes. There should be:
one pseudo-cylindrical (with round ends)
one rectangular (with square corners)
one interrupted (good low distortion display of all areas)

There should be a pseudo cylindrical: Eckert IV (4) is very good. It has general proportions like Robinson (it is a little more square) and is equal Area.

Some people have problems with the round ends on pseudocylindricals (like design department directors who want to 'fill in the corners'). Thus, a rectangular option should be included. Lambert's Cylindrical is too long and narrow to work well on a page (except if you have a stack of small multiples) and makes higher latitudes almost impossible to see.
The unmitigated Gall-Peters is too extreme (the 'long underwear on a cloths-line' thing: the shapes are just too distorted).

For the rectangular cylindrical, I use the Behrman’s. It uses 30° standard parallels, as opposed to 0° for Lamberts and 45° for Gall-Peters. Most parts of the world show reasonably good shapes, it is equal area, and it satisfies the people who want corners.

For the third, I would agree with the proposal of the interrupted Goode Homolosine. It would be best if we could include both an un-interrupted-land and an un-interrupted-water version, but that may be asking too much (we were restricted to three, after all). The un-interrupted-land version would likely get precedent.

So, the three are:
pseudo cylindrical: Eckert IV (4)
rectangular: Behrman’s
interrupted: Goode Homolosine (un-interrupted-land)

Mark Denil
National Ice Center