No need for a pre-fill. When arrays are created, they will be filled with zero value. "Null" values isn't a concept that the O/S supports or recognises. Something must have corrupted one or more values since it was created.
The only way to corrupt a REAL array is to FLL, COP or CPS something that isn't a properly-encoded REAL into it. This is because those instructions take no notice of data-type, and are just fast "byte-for-byte" copiers.
EDIT : Actually 1.$ is "infinity", usually preceded by + or -, and the only way I can figure how you might get that as a result is if you have the LENgth specified as longer than the array, and the instruction is picking up stuff from memory outside of the array tag