[Spice_discussion] Long-term accuracy of the IAU_MOON frame

[John Irwin]

> Prior to 1500, the oscillations look like they have a period 
> about equal to a saros.

You're not far wrong.

A quick FFT shows the main component has a period of 18.6 years which is
close to the (current) regression period of the lunar orbital nodes around
the ecliptic. 

The saros period is slightly different at ~18.03 years because it includes
other secular variations of the lunar orbit and also variations in the
Earth's orbit. But the nodal variation is the dominant component.

Not sure why the Moon's nodal period appears so clearly in the IAU_MOON
discrepancy. Maybe it's a reflection of the fact that the secular variations
are not constant over long periods of time. If the "simple" IAU_MOON assumes
constant secular changes (likely) then the periods associated with them will
start to manifest themselves far from current epochs when compared to a more
accurate model (such as DE431).

John.