There is also the more generic field "Computational Science and Engineering" (CS&E) [1], which is primarily the development of methods and software that is applied in many specific disciplines. I work primarily on the library development side of CS&E, working with many applications to understand their needs, often finding ways to satisfy those needs with better software reuse and better ultimate flexibility, then implementing any components that belong in libraries. Additionally, we do basic research by recognizing unexploited common themes and designing new algorithms that will benefit existing and potential users.
There is currently a great deal of demand for Computational X, for CS&E through the postdoc level, and for CS&E in industry and national labs. It is recognized that a versatile practitioner of CS&E can effectively assist Computational X for many values of X, since there really is a lot in common. Once you've put in 10000 hours of CS&E, plus 1000 hours in a handful of disciplines, you should be able to get orientedy quickly in new applications.
At the faculty level, support for CS&E is highly variable. At some universities, X departments have the view that all their faculty should "X first, computation second", while the mathematics department wants to see theorems and isn't overly concerned with following through to see the method applied to obtain a new science or engineering result. When hiring committees say "we need CS&E, but some other department should have the CS&E faculty", it drives a lot of talent away from teaching roles (e.g., into industry), perpetuating the relative lack of early education in CS&E fundamentals. On the other hand, the trend is for universities to have more direct interest in CS&E (sometimes recognizing it with a dedicated department or inistitute). It helps that computation stimulates many interdisciplinary collaborations, leading to more diverse funding possibilities.
[1] We have a big meeting next week: http://www.siam.org/meetings/cse13/