It's described in a paper entitled Ranking and mapping of universities and research-focused institutions worldwide based on highly-cited papers: A visualization of results from multi-level models by Lutz Bornmann, Moritz Stefaner, Felix de Moya Anegon, Ruediger Mutz.
It just looks at one metric for each institution for each research field: the fraction of papers in the top 10 percentile of papers in that field.
Since this is a field-specific exercise, it is slightly better than university level ranking (even after discounting their unsound/stupid methodologies). I would still rank it as a bad exercise since it encourages people to see the research enterprise essentially as a race to get into the top 10 percentile of papers.
Anywasy, what makes it worth blogging is the
time-sink interactive web app [password provided on demand] created by the authors that allows you to see for yourself where the good places (and the bad places) are. Proceed with caution -- you might end up spending a lot of time over some metric of colossal insignificance. [See this for example].
For the pointer, I
blame thank Doug Natelson.