[mseebach]

Hmm, not sure I understand what you're saying. Figure 5 in that paper is the situation I'm describing: at a certain latitude (52 deg north in this case, incidentally that latitude is my 'hood) there are no satellites visible to the north. Figure 2(b) illustrates why that's bad for precision.

[jessriedel]

Sure, that's why I said I wasn't contradicting you, just clarifying. One could mistakenly read your comment and think that the coverage goes to zero at the poles ("the further from the equator you get"), but of course GPS still works pretty well there. When you get far enough north, satellites from the "other side" of the Earth come into view, so the accuracy levels off and (I think) starts increasing again with latitude.

(And I corrected my comment to refer to figure 4, not 2.)