The map up top is the 2008 race between Mark Begich and Ted Stevens, in which Begich prevailed by 1.25%.
If you guessed that correctly (without cheating), 10 points for Gryffindor. If you did cheat and look at the file name, boo on you too, but you can look at the Anchorage inset anyways:
Here's a redux of the Murkowski-Miller race (blue for Murk, Red for Miller; Absentees not included):
And you can judge for yourself similarities between that at the 2008 GOP primary, Young-Parnell (Young in blue, Parnell in red):
I'm not that optimistic about Scott McAdams' chances in November, but there does seem to be a path for him:
Areas of strong Begich performance are decently correlated with areas of strong Murkowski performance - or put differently - weaker Miller performance. Given that, this seems to bode somewhat better for McAdams, in that he could piece together the Begich coalition of Anchorage + Outlying Areas + Juneau for a win, pulling in disaffected Murkowski GOPers. Those areas (notably, GOP voters in those areas) weren't exactly hopping for Miller.