I am working on a project to try to identify the most threatened populations (on a global scale) of having a limited access to freshwater. I want to compare those most at risk currently and then add in climate change projections to see how it is expected to change. I am trying to justify what projection system to use. With such a global scale what is the best option, if focusing on freshwater geography as well as human demographics? I am a little overwhelmed, but looking at mostly WGS 1984 coordinate systems. How would you justify a certain projection for such a broad analysis area and concentrations.
If you are looking at demographic, I suggest that you use some kind of equal area projection. This will give to each country a size that reflects its true size.
There are several equal area projection, including sinusoidal or cylindrical equal area. Cylindrical equal area is quite distorted, but it is convenient for WebGIS because you can seemlessly turn around the world. However, for a static map you could look at sinusoidal or Mollweide (see a list here) which are less distorted.
Note: With Plate carree (the "default" projection with lat/long coordinates) or Mercator (to give another example), you will increase the importance of countries located at high latitudes, which are mainly located in the northen hemisphere ( there is not much inhabitated land in the Southern high latitude compared with the Northern hemisphere). Because countries of the high latitude usually suffer less from freshwater issues, you will give a more positive signal.
remark : you could also look at the more fancy Waterman projection (a nice compromise between conformity and equal area, and a nice projection name for freshwater mapping)
I would portray the output data on Mollewide or Robinson.
Gall-Peters is used extensively by the UN and would work as well.
Wikipedia has nice Tissot distortion diagrams for all three.
I would politely disagree with just this statement in the first answer
"Viewing anything on a global scale suits geographic projection, WGS84 being one of them."