Hi Samuel,

I agree with Richard.

The main reason for choosing a projection system is to be found in the geometric characteristics of the map  in relation to it's purpose.

Well, there might be other special criteria for choosing a map-projection.

E.g., we are currently setting up a walkable map for multiple users to provide an experiental visualization. Since the viewing direction will be different for each user, we prefer projections that are not oriented towards an exclusive direction (usually North). Moreover, we would like to arouse the user's curiosity. So we decided to use Buckminster Fuller's Dymaxion projection (which, by the way, avoids a West-centric presentation) in this project.

However, as Richard and Greta pointed out, usually geometric characterics play a more important role than those special criteria I listed.

 Thank you, Richard, Greta, and Benno. This is helpful. 

Sincerely, Sam

The simplest would be to use WGS 84. (World Geodetic System) due to wide use of GPS and navigation systems everywhere.  WGS 84 uses oblate spheroid as datum surface.

As to what projections are the best we can formulate this question into: which projections are most widely used? Mercator projection including UTM, and Lambert Conformal Conic projections are most widely used projections. Therefore the practicality is to be also considered. People like to compare different areas that they can identify, especially the ones that are known to them (home country);  in another words, it is good to have a map that is easy compatible with the rest of the maps that we see.

There is no perfect projection in the world, therefore you have to chose based on your area and what you want to preserve as accurate. The simplest answer would be to use WGS 84. (World Geodetic System) due to wide use of GPS and navigation systems everywhere.  WGS 84 uses oblate spheroid as datum surface.

The statistics of usage of  projections are as follows: The Mercator projection and the Lambert Conformal Conic projections are by far the most widely used projections in the world.

Maps are projected in a number of ways, each serving a different purpose. All attempt to find the best way to transform the three-dimensional world into a two dimensional plane. Thus Mercator was interested in a map that permitted maritime travel while the Peters Projection wished to make clear a more accurate method of showing land masses in relation to each other. So the choice of a projection is about the idea behind the mapping. For a discussion of the Peters Projection, and the idea of mapping as an idea rather than a reflecton of some reality, see Denis Wood's The Power of Maps.

I agree with Richard, the use of projection depends on the use case, i.e. for which purposes the map or the geoinformation system is prepared. In navigation the most important is the preservation of the angles, while in statistics the right representation of area. In national and large scales (base maps) the reference and projection systems are optimised so that not only angels and areas, but also distances become a good approximation of reality.

Just a small clarification: the name of the projection system used in Russia is Gauss-Krüger, which is very similar to UTM. Krasovsky is the name of the reference elipsoid, while Pulkovo is the elevation datum.

There are two books that will help: Datums and map projections (Iliffe) and How to lie with maps (Monmonier). 

There are two books that will help: Datums and map projections (Iliffe) and How to lie with maps (Monmonier). 

There are two books that will help: Datums and map projections (Iliffe) and How to lie with maps (Monmonier). 

I advocate B.J.S. Cahill's octahedral world map projection, first published in 1909; and I have compiled a resource site of his maps and drafts and papers (plus my re-design): http://www.genekeyes.com/B.J.S._CAHILL_RESOURCE.html. See especially his article, "A World Map to End World Maps" http://www.genekeyes.com/Cahill-To-End-World-Maps/To-End-World-Maps.html

http://www.genekeyes.com/B.J.S._CAHILL_RESOURCE.html

http://www.genekeyes.com/Cahill-To-End-World-Maps/To-End-World-Maps.html

Several folks here have said already that projection choice should be made with respect to the sorts of geometric distortions the projection does or does not introduce and how that serves the map purpose. Note that distortion can be quantified; you can have a map that perfectly preserves area, or maps that distort area to varying degrees - it's not just a binary. Maps of the whole world always introduce some sort of serious distortion. There are several geometric properties that can be preserved or distorted (e.g., azimuth [i.e., angles and directions], relative distance, etc.). Also, sometimes a projection preserves a quality but only in a certain area on the map (e.g., relative distance is true on a Mercator projection along the latitude parallels only).

Conformality (i.e., the quality of local angles, and therefore, overall shapes, remaining true to their global counterparts) and equivalence (i.e., the quality of areas remaining in true relative proportion to each other) are perhaps the two most important qualities to consider in most non-specialized settings where you have to choose a projection. These two, to my knowledge, are the only two properties that are mutually exclusive - you cannot possibly have both, having one thoroughly distorts the other, though you can have compromises between the two. The Peters projection (i.e., equal-area cylindrical) retains area perfectly, but severely distorts shape. The Mercator map is the opposite. Interestingly, I think, both are derived in much the same way, with a cylindrical "developable surface", but with different mathematical extensions/contractions as one moves from the Equator to the poles. The Mercator does this to allow for straight-line navigation routes that actually get to where they're drawn to get to; the equal-area cylindrical to preserve areas.

Many world map projections (e.g., Robinson, Winkel-Trippel, Kavraisky) are compromise projections, meaning they distort different geometric properties to varying degrees to strike some sort of balance of distortion types. Sometimes the reasons for doing so are aesthetic and not scientific (e.g., Robinson).

I think it's also worth noting that the projection (i.e., the actual geometric/mathematical transformation from sphere or sheroid to flat surface), and the aspect are different - the latter is things like orientation (e.g., whether North is at the top, whether the Equator runs left-to right, where the map is centered on the world, etc.). You could center any projection anywhere. I attach an example of a Mercator projection I centered on southern Africa, for example.

Note also that some projections aren't suited to the whole world. For example, UTM or US State Plane systems are meant only for smaller areas (e.g., roughly one US State or European country), because they introduce very little distortion across such that smaller area - but they severely distort the rest of the world.

The best discussion of projections that assumes no prior knowledge that I've ever seen is in the user's manual for GeoCart, a piece of software that explicitly modifies GIS data projections (Chapter 3): http://www.mapthematics.com/Downloads/Geocart_Manual.pdf. Highly recommended reading.

Cheers, hope this helps :)

P

Answering a question about the "best" projection is like answering a question about the "best" religion or the "best" brand of pick-up truck.  It will always illicit a plethora of opinions.  First, consider the size of the region.  If the entire world for a single map; the use of a datum and/or ellipsoid is a waste of time.  Second, the purpose of the display is the controlling factor in that - are you interested in preserving shapes, or areas, or what?  Each family of world projections has been designed with a specific type of distortion optimized for a specific purpose.  Decide on your cartographic objective as your constraint and with that established your choices will be narrowed down to just a few dozen choices to whatever happens to "float your boat."  If you are in graduate school, the absolutely best choice possible is whatever projection your Major Professor prefers ... Good luck!  :-)

I agree Cliff, but I would not recommend to mix religion in selection of "best"  projection