I'm pretty sure that as the system is a strap down system, and alignment is carried out mathematically, and as variation is a known quantity, with a known annual rate of change,then computing magnetic north shouldn't be too difficult. The hard bit has already been done on initial alignment.
Any other ideas out there...