[Spice_discussion] Best alternative to ITRF93 outside its time range + odd error

[Barry Carter]

I have these kernels loaded:

KERNELS_TO_LOAD = (
'/home/barrycarter/SPICE/KERNELS/de430.bsp',
'/home/barrycarter/SPICE/KERNELS/jup310.bsp',
'/home/barrycarter/SPICE/KERNELS/sat365.bsp',
'/home/user/SPICE/KERNELS/ura111.bsp',
'/home/user/SPICE/KERNELS/nep081.bsp',
'/home/barrycarter/SPICE/KERNELS/pck00010.tpc',
'/home/barrycarter/SPICE/KERNELS/naif0011.tls',
'/home/barrycarter/SPICE/KERNELS/earth_720101_070426.bpc',
'/home/barrycarter/SPICE/KERNELS/earth_070425_370426_predict.bpc',
'/home/barrycarter/SPICE/KERNELS/earth_latest_high_prec.bpc'
)

If I understand correctly, the last three files mean:

   - earth_720101_070426.bpc means I can compute ITRF93 from 1972-01-01 to 
2007-04-26.

   - earth_070425_370426_predict.bpc means I can compute ITRF93 from 
2007-04-25 to 2037-04-26.

   - earth_latest_high_prec.bpc is another name for 
earth_000101_160314_151222.bpc which means I can compute ITRF93 from 
2000-01-01 to 2016-03-14, and this data is accurate as of 2015-12-22.

This should not work:

spkezp_c(10,-1167626622,"ITRF93","CN+S",399,v,&lt);

because et of -1167626622 is approximately 1963-01-01, and none of the 
kernels I have should compute ITRF93 at that time.

If I go back another year (to UTC 1962-01-01), it fails as expected.

So I'm curious as to why this works at all.

More importantly though: what kernel should I use outside the range 
supported by ITRF93? If I want to compute Earth-based positions in the 
Middle Ages, for example, should I use IAU_EARTH, or is there something 
more accurate than IAU_EARTH (but presumably less accurate than ITRF93)?