[LEAPSECS] real-time UT1 realisation

[Tom Van Baak]

> Yea, the hard part is keeping that table up to date, but

> I guess that's the job of a crontab entry not ntpd itself.


Warner,

Yes, a daily crontab entry would take care of it. Note the kernel
table really only needs two entries (today and tomorrow).

With two entries you have a full 24 hours in which to get the next
value from IERS or USNO or your favorite DUT1 source. If there's
a risk you aren't net connected for a day then using a full week or
month-long table would work too, with almost no loss of precision.

Really, the DUT1 prediction numbers don't change that much, so
for this application especially, there is no urgency getting fresh
values as often as once a day.

For example, if I used an ancient ser7.dat table from Jan 1, 2011
it predicted that DUT1 today would be -0.26741. The actual today
is -0.30862; so that's an error of only 0.04 second with a 9 month
out of date table.

If you want to make a nice project out of this, grab ser7.dat once
a day for a month and then make a nice graph of worst-case
error vs. prediction interval.

/tvb