The map is a gnomic projection icosahedal map. Basically, what you get if you project edges of landmasses onto an icosahedron places inside the sphere, using normal vectors, then unfolding the icosahedron.
The PDF is vectorial, (not a raster wrapped in a pdf. I hate those.) Meaning it is made of 2d line data, and can be enlarged without losing precision. I converted that vector data into a DXF, imported that into my CAD software, and "folded" the icosahedron. Then I projected the the 2d vector data from each facet onto a section of a sherical volume I had created.
The size of the sphere was about 10 inch diameter, because of limitations of the cad software. (On a joke, I tried creating a earth sized sphere once. It brough my 8 core system to its knees, and consumed 12gb of memory before windows force-closed it on a 64bit engineering workstation. Earth sized model just isn't gonna happen.)
I used the isometric projection feature of the drafting workbench to get a 2d lineart rendering of the spherical "earth", saved that as an svg, and brought it into inkscape for poster making purposes. My employer was happy with the result, but unhappy at the time it took me to make. (Hey, they wanted a globe, and didn't want to pay for stock models, so they paid more in terms of my paycheck to make them one instead.
)
I could replicate the work sometime I suppose. I would LOVE to get an ocosahedral gnomic projection topology map that is vectoral! I could produce a geologically accurate earth from one of those!
Sasdly, those are almost always raster images, and never the right projection.