SAF is being developed as part of the Data Models and Formats (DMF) component of the Accelerated Strategic Computing Initiative (ASCI). SAF represents a revolutionary approach to interoperation of high performance, scientific computing applications based upon rigorous, math oriented data modeling principles. Previous technologies have required all applications to use the same data structures and/or mesh objects to represent scientific data or lead to an ever expanding set of incrementally different data structures and/or mesh objects. SAF addresses this problem by providing a small set of mathematical building blocks, sets, relations and fields, out of which a wide variety of scientific data can be characterized. Applications literally model their data by assembling these building blocks. Sets and fields building blocks are at once, both primitive and abstract: * They are primitive enough to model a wide variety of scientific data. * They are abstract enough to model the data in terms of what it represents in a mathematical or physical sense independent of how it is represented in an implementation sense. For example, while there are many ways to represent the airflow over the wing of a supersonic aircraft in a computer program, there is only one mathematical/physical interpretation: amore » field of 3D velocity vectors over a 2D surface. This latter description is immutable. It is independent of any particular representation or implementation choices. Understanding this what versus how relationship, that is what is represented versus how it is represented, is key to developing a solution for large scale integration of scientific software.« less