str stores the address (and therefore points) to a literal string, "Hello", which is stored somewhere in a read-only segment of memory.
ptr points to the same address as 'str'.
rptr basically points to 'str' (not to the same pointee as str, but to str itself). It might be unusual to use 'points to' for references but they're really very much like pointers themselves (in this case a pointer to a pointer) except with slight syntactical differences and the restriction that they cannot point to any other address during their lifetime.
It would be analogous to:
char** const rptr = &str;
Like a reference, rptr above cannot be assigned a new address (it cannot change what it's pointing to), but it can be free to change its pointee (which happens to be a pointer in this case to 'str').
*rptr = 0; // after this, str == 0
References are pretty much the same as a read-only pointer (not a mutable pointer to read-only pointee) only they don't require a dereferencing operator to get at the pointee (the referenced data):
char *str = "Hello";
char *&rptr = str;
rptr = 0; // after this, str == 0
The only difference from the above example with a read-only pointer to pointer is that we didn't have to use the operator*.
const references also have the unique property of being able to extend the lifetime of temporaries, but that's probably beyond the scope of the discussion.