Don't let yourself be fooled by the similarity in names: these are pretty different topics, with different audiences.

Computer science is a broad subject, with a somewhat misleading names because it's not just about computers: computing science would be a better name. Computing refers to applying systematic treatment to information.
Computational science, also known as scientific computing, is the application of mathematical models to computations for scientific disciplines.
There is some overlap between CS and Scicomp, mainly in the form of algorithms for numerical analysis. But CS is a lot broader than that. Scicomp.SE is more of a vertical: it ventures far more into engineering where CS.SE sticks to science, but sticks to one field.
I don't know what you mean when you write “the 3 question examples seem to fit very well here”. Of the three bullet points that are mentioned as a goot fit:
- Software packages used in computational science are squarely off-topic here.
- Algorithms or methods used to solve problems in applied mathematics are on-topic here (but they are more likely to find a specialist audience on Scicomp.SE, since on CS.SE we're good at algorithms but not always at the applied math behind them).
- Modeling methods are off-topic here: they're a topic for physicists, chemists, biologists, etc.
Scicomp.SE and CS.SE have pretty much disjoint audiences. From data retrieved in November 2011 (courtesy of Piotr Migdal), there were only 5 users who had more than 200 reputation points on both CS.SE and Scicomp.SE, out of 212 on CS and 130 on Scicomp. For comparison, with 195 eligible users on Academia, 16 also have 200 rep on CS and 16 also have 200 rep on Scicomp (with just 1 on all three).