That mountain ridge trick, how reliable of an identifier is that? I'm not really questioning the validity of its use in this situation, but I'd be curious to know more about the nature of the ridge to gps coordinates mapping. How many ridges map to more than one gps coordinate (albeit at potentially very distant locations, one of which could be excluded if more information is available)? Or is the mapping completely injective? That'd be really cool, a bit surprising, but not infeasible.

Sounds like a fun data science project. (Draw a ridge and get a google street view image whose horizon matches the ridge as closely as possible, maybe? Might be fun to play with!)