[LEAPSECS] Crunching Bulletin B numbers

Rob Seaman seaman at noao.edu
Tue Feb 15 17:25:49 EST 2011

On Feb 15, 2011, at 2:31 PM, Warner Losh wrote:

> On 02/15/2011 09:12, Rob Seaman wrote:

>> So, what is the state of the art for long term predictions of UT1? Could the algorithms used by the EOP PCC teams simply be run on the historical Bulletin B numbers to find out?


> At the Torin conference they had a sliding scale for confidence in UT1 predictions, and it was very non-linear. I don't have the reference handy, but it went something like:


> 10 days a few microseconds

> 100 days a few milliseconds

> 1000 days a second or two


> which fits with other statements that said operationally, they can model the earth well enough to predict leap seconds about 2-3 years into the future with a 98% confidence level to conform with the DUT1 .9s requirement.

So in 2003, they estimated that UT1 could be predicted no better than 1s over 1000 days. c. 2008 they actually performed the exercise and found that UT1 could be predicted more than an order of magnitude better than this over 500 days. This raises the ante over what can actually be achieved and what the necessary DUT1 tolerance trade-offs might have to be.

The original November 1999 McCarthy & Kelpczynski GPS World article that started this all off:


outlined five separate options for the "continued maintenance of UTC". Wouldn't it be good to examine one of the other four options using actual scientific methods?

Operationally the IERS is using older procedures that are likely significantly more simplistic algorithmically than the recent EOP PCC while also being entangled with other parts of the workflow. For instance, how much evolution in the UT1 prediction operations has occurred since 1990:


Also, it is a different exercise to predict UT1 separately from all the other Earth orientation parameters, if all you need to constrain is the former.

> Looking at the Bulletin B's might not give a long enough time horizon. Those only predict out 30 days or so.

Correct me if I'm wrong: Bulletin A are the predictions (using methods presumably less advanced than the EOP PCC). Bulletin B are the final observed parameters. My point is that whatever the necessary input data (including such things as atmospheric angular momentum models), the new-and-improved predictions can be compared against the archival data. The data challenge (at least for this purpose) doesn't have to be conducted in real time.

> IERS provides historical aggregated earth orientation data that can be used to see how well predictions match actual data over any given time horizon.


> There's enough grunt work here that you might be able to publish a paper on this. :)

They are publishing papers. The question is whether the papers resulting from efforts like:




can be turned into improved workflows for predicting UT1 as input to a new extended scheduling paradigm.

Apparently the lion share of the effort on EOP predictions is invested in the short term numbers for real time operations. Meanwhile, the 500 day UT1 predictions from EOP PCC appear remarkably good. What improvements can be realized on the long term side of operations? Is the epoch during which the EOP PCC was conducted representative of the results that can be expected?


-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://six.pairlist.net/pipermail/leapsecs/attachments/20110215/18f443de/attachment.html>

More information about the LEAPSECS mailing list