(academic level) including math, physics, biology, chemistry, statistics and maybe even philosophy. It's highly important that those books will be self-contained.
Hm, well, you'd have to go through the major undergraduate (and 1st-year graduate level) textbooks of each and every subject from freshman to senior level (you can find them through the list of textbooks used by any particular university, or by Amazon guides/listmanias (http://www.ehow.com/how_2341922_...), or simply by asking undergrads or people here). It's a process that I'm still going through (I've pretty much browsed many of the major textbooks for all the subjects you've mentioned *plus* astronomy, atmospheric science, applied math, philosophy, psychology, history, and computer science. In order to *really* learn some subjects, you *have* to do problem sets, and doing them is really what takes up the vast majority of your time). And in some fields like astronomy and biology, you don't really know the field unless you've read the research papers, since even 5-year old textbooks are practically outdated. If you're smart, not too inefficient, and don't spend much time on other things, then it only takes around half a decade. But I think there's a very important role for play too, so don't spend all your time studying.
You generally don't really need to go hardcore into mathematical proofs, and you generally don't need real analysis, abstract algebra, or topology. These do matter for theoretical economics, theoretical CS, mathematical physics, and some parts of theoretical biology [e.g. Carl Bergstrom's research] but in general most scientists don't care too much for them. A graduate level applied math textbook would be far more useful. Also, for modern science, recent topics like digital signal processing and data mining are very useful and cross scientific domains, but usually ignored in undergraduate-level curricula (in fact, a lot of research isn't in theory/experiment anymore - it's in simulation/data mining). You find resources for learning those at How do I become a data scientist?
Anyways, the most important books (IMO) would be Consilience, The Structure of Scientific Revolutions, Scientific Genius, and The Psychology of Science and the Origins of the Scientific Mind.
Here are some resources I collected: Collection of educational resources for the sciences. You should also look at Educational Resources for more general resource collections. You should also learn how to find course webpages on your own: - good google keywords would include "name of textbook" + site:edu.
By the way, I don't think it's important to learn everything when young. Learning technologies are improving so dramatically these days. I can learn things from Coursera in a fraction of the time it took for me to learn things 5 years ago. When obtaining a broad scientific background, time is god (because you have to spend more time learning than most other people). So it's extremely important to be efficient when learning.