[LEAPSECS] Keeping the day constant

[Zefram]

Tom Van Baak (lab) wrote:

>1) To make Earth more accurate and better match the SI second you need

>a one-time increase in rotation rate. This rate correction is parts in

>ten to the 8th. Energy goes as the square of rate. See Earth energy here:

>http://en.wikipedia.org/wiki/Rotational_energy


That states approximate rotational kinetic energy of 2.138e29 J.
The current Bulletin A <http://maia.usno.navy.mil/ser7/ser7.dat>
gives LOD averaging 86400.00007 s over the week 2011-07-15 to
2011-07-21.  (That's low for recent decades.)  This implies a required
proportional correction of 8e-10.  If performed by uniformly speeding
up the entire bulk of the Earth, that would require an injection of
1.7e20 J.  Current worldwide rate of electrical energy production, per
<http://en.wikipedia.org/wiki/World_energy_resources_and_consumption>,
is about 15 TW.  So performing the one-time correction that way would
require the entire world's electricity production for some 3.6 gigayears.
Of course, it would run straight into the need for continuous correction
for the secular slowing.

We'd better look at ways that don't involve speeding the entire Earth up,
and more appropriate power sources.  It's only the crust that needs to
be synchronised, so some kind of transfer of rotaional energy from the
mantle is an obvious route.  (Indeed, transfers between these components
seems to be a major cause of decadal LOD fluctuation.)  The mantle may
also be the obvious power source: it's fission heated, and there are some
big temperature differences available.  Any geophysicists in the house?


>2) To make Earth more stable and better follow the SI second you would

>need to dynamically increase or decrease the rotational rate to exactly

>compensate for ever present internal and external perturbations.


I think this is best split into two distinct aspects.  There should
be a long-term correction that matches the natural secular slowing.
This can get you a bounded DUT1 (bounded to a couple of seconds?) with
no leap seconds.  The long-term correction varies very little over time,
and only needs to be steered on a timescale of decades.

Secondly, you can apply short-term corrections to keep DUT1 within a much
smaller bound.  This has the trickier control issues that you mentioned.
I imagine the IERS would have to continually invest in research to
gradually improve the DUT1 bound.

I, for one, would be quite satisfied with corrections only for the one-off
frequency offset and the secular drift.  I can cope with a couple of
seconds of DUT1.

-zefram