[IGSMAIL-3470]: new clock products available

Thu Aug 16 11:34:17 PDT 2001

******************************************************************************
IGS Electronic Mail      16 Aug 11:34:19 PDT 2001      Message Number 3470
******************************************************************************

Author: Jim Ray & Ken Senior

Dear Colleagues,

Since October 2000 the IGS is distributing clock products for both
satellites and tracking receivers, which are tabulated at 5-minute
intervals and distributed in .clk RINEX-format daily files.  Together
with the .sp3 orbit files, these products provide users with sufficient
information to determine consistent coordinates and clock values for an
isolated GPS receiver every 5-minutes (kinematic mode) with an accuracy
at the 5-cm level or so.

However, because the underlying time scale for the IGS combined clocks is
based on a linear alignment to broadcast GPS time for each day separately,
the day-to-day stability of the IGS clocks is very poor.  This is due to
the limited stability of GPS time itself, which has an Allan deviation of
about 2 parts in 10^14 at 1-day intervals.  The ending of SA on 02 May
2000 has not improved the performance of GPS time at these intervals.

The IGS, working closely with the BIPM through the IGS/BIPM Pilot Project,
aims to improve the day-to-day stability of its clock products while
providing an underlying time scale linked as closely as possible to UTC.
The next step is to realize an improved internal IGS time scale by using
the current clock products to form a new reference frequency scale from
a dynamically weighted ensemble of the stable frequency standards in the
IGS network.  Eventually, it is expected that this more stable IGS time
scale will be traceable directly to UTC through the data of calibrated
GPS receiver links at timing laboratories.

Ken Senior (USNO/TS, senior.ken at usno.navy.mil) has developed such a time
scale which is now available for evaluation.  It is loosely steered to
broadcast GPS time, with a time constant longer than a month.  Results
indicate that the new time scale is stable to about 1 part in 10^15 at
1 day.  Importantly, the differential relationship between any pair of
clocks is unchanged from the official IGS products, so the overall
consistency is also unchanged.  For a detailed description of Ken's
methodology, please refer to the preprint by K. Senior, P. Koppang, D.
Matsakis, and J. Ray, "Developing an IGS time scale", Proc. 2001 IEEE
International Frequency Control Symposium, in press 2001, at
  http://maia.usno.navy.mil/gpst/refs/senior-fcs01.pdf

Ken has generated re-referenced .clk files and plots (time domain and Allan
deviation) starting with week 1086 (29 Oct 2000), which you are encouraged
to examine.  They are available at
  http://clockdev.usno.navy.mil/igst/final/   or
  ftp://clockdev.usno.navy.mil/igst/final/
Note that in the time plots quadratic trends are removed for display and
only every 3rd point is shown.  The clock labelled "GPST" is the realization
of the current IGS time scale compared to the new scale; it is not an
observed clock.  The GPST values have been removed from the current IGS
clocks to produce the new re-referenced clocks.

Please send any comments, suggestions, or questions to Ken at
<senior.ken at usno.navy.mil>.

Sincerely yours,
--Jim Ray & Ken Senior