From jimr@maia.usno.navy.mil Fri Oct 23 17:52:13 EDT 1998
Received: (from jimr@localhost)
	by maia.usno.navy.mil (8.8.6 (PHNE_14041)/8.8.6) id RAA08226
	for gpst@maia; Fri, 23 Oct 1998 17:49:32 -0400 (EDT)
From: Jim Ray (USNO 202-762-1444) 
Message-Id: <199810232149.RAA08226@maia.usno.navy.mil>
Subject: IGS Clock Analysis plan
To: gpst@maia.usno.navy.mil
Date: Fri, 23 Oct 1998 17:49:31 EDT
X-Mailer: Elm [revision: 212.4]
Status: RO

Dear Colleagues,

For your information, below is a copy of the IGS clock analysis plan that
has been developed among the IGS Analysis Centers.  The initial steps 
have begun and discussions with Jan Kouba and Tim Springer indicate that
the IGS probably will be distributing clock products for GPS stations
(in addition to the current GPS satellite clocks) by the start of 1999.
This plan provides details about how the process will work.

--Jim


Goal 1. -- Submit station (& satellite) clock results
-----------------------------------------------------
There are several important reasons for the IGS to add station clock products
to its output, in addition to these being part of the Terms of Reference:
(1) As noted by Springer, Zumberge, and Kouba (1998) in their Darmstadt
position paper, the availability of clock estimates from stations equipped
with high-performance frequency standards (H-masers mostly) would allow a
more precise alignment of the clock solutions (currently including only
satellites) from the ACs relative to one another and would allow better
relative weightings.  The current restriction of only using satellite clocks
for weighting and alignment limits the IGS combined precision to roughly
0.2 ns.  (2) As discussed at the IGS/BIPM meeting at Sevres in June, 
linkage of the IGS clock products to TAI/UTC is highly desirable (just as
station coordinates are linked to ITRF).  This is most directly done by
using IGS station clock results from timing lab receivers together with
local time transfer (calibration) data.  Until this is feasible (the Timing
Project must develop the calibration techniques with the labs and BIPM), the
IGS must continue to link its clock products to GPS time.  (3) The station
clock results have intrinsic value and interest to a wide user community
for a variety of applications.

* Action 1.A -- Develop clock data exchange format [DONE]

  Within the Timing Project, the initial RINEX extension has been develop
  and is available at http://maia.usno.navy.mil/gpst/clock-format .

* Action 1.B -- ACs adopt common satellite antenna offsets [ASAP]

  In consultation with the ACs, the AC Coordinator should distribute a
  set of values for the antenna phase center offsets from the center of
  mass of each satellite.  The ACs are urged to adopt these values in
  their analysis solutions as soon as feasible in order that all the IGS
  clock solutions are referred to a common set of physical points within
  the spacecraft.

* Action 1.C -- ACs begin to submit weekly clock solutions [ASAP]

  ACs are encouraged to develop clock RINEX software and start submitting
  "cccwwwwd.clk" solution files, weekly with other IGS Finals submissions.
  The results should be for the same 15-minute epochs used in the orbit
  sp3 files and should include the (redundant) satellite clock information
  in order to simplify the IGS combination effort.

  Later, after the procedures for handling the new clock information are
  fully developed for the IGS Finals products, the ACs would be asked to
  also start submitting daily station and satellite clock results with
  their other IGR submissions.

* Action 1.D -- AC Coordinator develops tools to use clock solutions [ASAP]

  As soon as clock solution files become available the Coordinator needs
  to begin to test and develop the ability to read and manipulate the
  clock information from the ACs.


Goal 2. -- Use station clocks to improve IGS alignment/weighting
----------------------------------------------------------------
The improved inter-AC clock alignment and weighting described above can
only be accomplished if there is a sufficient number of station clocks in
common among the ACs.  If too few stations are common, then the results
will not improve over the current use of ~26 satellite clocks.  To improve
the alignment strategy, ACs are encouraged to include as many sites as
possible from a list of "fiducial clock" stations.  The list should
include primarily sites from the set of RF47 which have stable frequency
standards.  In addition, suitable IGS sites at timing labs should be 
included to prepare for eventual linkage of the IGS clock results to TAI.

* Action 2.A -- Develop list of "fiducial clock" sites [DONE]

  The following are RF47 sites equipped with H-masers:

  ALGO,  DRAO,  FAIR,  FORT,  GODE,  GOL2,     PRIMARY FIDUCIAL CLOCKS
  HOB2,  IRKT,  KOKB,  MAD2,  MATE,  NLIB,     PRIMARY FIDUCIAL CLOCKS
  NYAL,  ONSA,  PIE1,  TID2,  WES2,  WTZR,     PRIMARY FIDUCIAL CLOCKS
  YELL                                         PRIMARY FIDUCIAL CLOCKS

  The following H-maser sites are at timing labs:

  BRUS,  NRC1,  USNO                           PRIMARY FIDUCIAL CLOCKS

  Together, the 22 sites listed above form the "primary fiducial clock"
  stations and ACs are encouraged to include as many of these as possible
  in their solutions for clock products.  In addition, ACs are urged to
  use one of these stations as their fixed clock reference.

  For reference, the following additional IGS sites are also equipped
  with H-masers but are not among the RF47 set:

  HRAO,  MDVO,  MEDI,  METS,  MKEA,  NOTO     SECONDARY FIDUCIAL CLOCKS

  and the following are RF47 sites which are equipped with Cs frequency
  standards:

  GRAZ,  MAS1,  MDO1,  TSKB,  VILL,  YAR1     SECONDARY FIDUCIAL CLOCKS

  These latter 12 sites form the "secondary fiducial clock" stations
  which ACs should consider including in their solutions for clock
  products where feasible.

* Action 2.B -- ACs consider using fiducial clock sites [DONE]

  ACs are asked to examine the list above and consider whether it is 
  reasonable for their operational procedures.  If it is not, then
  a revised list or plan will be needed.  Also, ACs should consider
  whether other fiducial clocks sites should be added (or removed).

  When a fiducial plan is adopted, the ACs are encouraged to use as
  many primary fiducial clock sites as possible in their solutions and
  to fix one of them as their clock reference.  Additional secondary
  fiducial clock sites would be beneficial.
  
* Action 2.C -- AC Coordinator evaluates effectiveness [ASAP]

  Based on this approach, the AC Coordinator will need to evaluate the
  effectiveness and usefulness of the fiducial selection.  The Coordinator
  will need to develop new combination software, in close cooperation with
  the ACs.  If the set of common stable clock sites is insufficient, then
  the use of satellite clock results for inter-AC alignment and weighting
  must continue.


Goal 3. -- Distribute IGS clock products
----------------------------------------
Once Goals 1 and 2 have been met, then the IGS can begin to distribute
its combined station clock products (initially as part of the Final
products but soon afterwards also with the Rapids), as well as improved
satellite clocks.  The satellite clocks will continue to be adjusted to
account for differences between AC orbit solutions and the IGS combined
orbits in order to ensure that consistency is maintained.  All of the
clock products (stations and satellites) will be consistent with other
IGS products, particularly the orbits, and will be expressed in a common
timescale (eventually, linked to TAI/UTC).

It should be made very clear to the user community that the epoch offsets
of these station clock products refer to the internal "clock" of each GPS
receiver and not to the external standards (at least not until local time
transfer calibrations become available).

* Action 3.A -- AC Coordinator adds new clock products [after Goals 1,2]


Goal 4. -- Linkage of IGS clock products to TAI/UTC
---------------------------------------------------
For now, the IGS clock products will continue to be aligned with the GPS
timescale.  However, it is desirable for the IGS to become more closely
connected to the UTC/TAI timescales maintain by the BIPM.  Indeed, the
IGS is poised to contribute significantly to the maintenance of UTC by
improving global time/frequency transfer.  However, to accomplish this
will require the development of accurate local calibration techniques that
can be applied at timing labs which contribute to TAI.  This is a major
effort of the IGS/BIPM Timing Project, which must also validate the
calibrations using independent techniques.

* Action 4.A -- Develop local GPS time calibration techniques 
                [IGS/BIPM Timing Project]

* Action 4.B -- IGS apply calibrations to effect TAI link [later]


Goal 5. -- High-rate satellite clocks
-------------------------------------
As also described by Springer, Zumberge, and Kouba (1998) in their Darmstadt
position paper, there are applications which require satellite clock results
more frequently than the 15-minute interval currently provided by the IGS.
To permit interpolation of the satellite clocks affected by SA, it is
necessary to have clock results tabulated every 30 seconds.  The new 
RINEX clock format now allows such results to be exchanged.

* Action 5.A -- ACs indicate willingness to contribute 30-s clocks [ASAP]

  To determine how well ACs are prepared for this, each is asked to
  respond to Jan Kouba with an estimate of when/if they expect to
  contribute satellite clock solutions at 30-s intervals in RINEX
  format.

* Action 5.B -- AC Coordinator assess combination impact [???]

  The task for the Coordinator to handle this new task must be assessed
  against other IGS priorities.  Based on the ACs' responses and this
  assessment, a plan for IGS implementation can be developed.

* Action 5.C -- 30-s satellite clocks as IGS products [???]

  Depending on Actions 5.A and 5.B, 30-second satellite clock values will
  eventually become a standard product of the IGS.