Suggested by Tim. Nature has a nice viz on where science happens, in terms of cities, and how publication output interacts with publication impact. Their analysis - provided to Nature by Elsevier - of Scopus data shows how absolute counts of publications per city and relative citation impact have changed from 2000 to 2008. The interactive graphics supplement the news feature 'Building the best cities for science' part of Nature's special package on Science and the city. Berkeley does quite well, by the way. Go Bears!
They also have the same information as a google earth viz. Very cool.