[IGSMAIL-1266] Summary of 1996 Silver Spring AC Workshop

[G.]

******************************************************************************
IGS Electronic Mail      Fri Mar 29  0:33:11 PST 1996      Message Number 1266
******************************************************************************

Author: G. Beutler
Subject: Summary of 1996 Silver Spring AC Workshop

Dear colleagues,

The 1996 IGS Analysis Center Worshop took place last week (19-21 March) in
Silver Spring, MD.  Gerry Mader and Jan Kouba who organized this meeting
arranged it up as a real workshop.  The setup was perfect to focus the
discussion, and I believe, that everybody enjoyed a very fruitful three
days at the NOAA facilities.  On Friday, 22 March, a business meeting of
the IGS Governing Board with the session chairs as guests was organized
with the goal to come up with the appropriate action items.

Let us now try to summarize the sessions and some events of the workshop.
It was a shock for the participants to learn immediately before
the start of the workshop that Jan Kouba, IGS Analysis Coordinator, could
not attend the workshop due to a very sudden health problem, which
virtually immobilized our Coordinator for a week.  I am convinced that
everybody is relieved to hear now that Jan -- according to his own
diagnosis -- is in perfect shape again, and that he continues his
coordinating task for the IGS with the same energy as before.

The following topics were addressed, where each topic was introduced by a
position paper prepared by the session chair persons:

- Orbit/Clock Combination (chair Kouba/Beutler)
- Earth Orientation (chair Ray/McCarthy)
- Antenna Calibration (chair Mader/Rothacher)
- SINEX, densification of the ITRF using the GPS (chair Blewitt)
- Receiver standards and Performance (chair Zumberge/Gurtner)
- Atmospheric Topics (chair Feltens/Gendt)

The position papers were available before the beginning of the workshop.
They will serve as a first draft for the session summary, including all
recommendations and decisions, which will be included into the workshop
proceedings. Having seen the efficiency and professional spirit with which
the workshop was organized I am convinced to see the proceedings in the
very near future. Let me go through the individual sessions, now.

ORBIT/CLOCK COMBINATION (chair Kouba/Beutler)

Currently the best ACs and the combined IGS solutions are approaching the
5cm(orbits)/0.5ns(clocks) precision level.  Combinations, comparisons,
evaluations and free exchange of information within the IGS and amongst the
IGS ACs are essential to the health and growth of the IGS.

The development of the IGS orbit quality showed that orbit
parameterization became an important issue even if the arc length is only
one day.  The weekly analyses of the IGS coordinator made it also clear
that different orbit modeling techniques led to different estimates (or
realizations) of the ITRF origin.  This is why it was recommended that "all
ACs make every effort to align their orbit, station and EOP solution to
conform to the ITRF origin.  It was shown that this could be effectively
achieved by means of stochastic orbit modeling or radiation pressure
modeling."

Recently the ITRF94 was made available by the ITRF section of the IERS
(Boucher and Altamimi). It was recommended that the ITRF94 should replace
the ITRF93 within the IGS, provided the tests performed by the IGS ACs in
collaboration with the IERS clearly indicate the superiority of the ITRF94.
The IGS AC Coordinator will coordinate these activities with the IERS.

Today all IGS ACs take part in yet another IGS combination, called the "IGS
Preliminary orbit/clock combination" which is now approaching a precision
of about 10cm/1ns and is made available with a delay of 38 h only.  In
order "to economize and to minimize the IGS combination effort and to speed
up the delivery of the IGS Final orbits/clocks it is recommended that
starting on 30 June, 1996 (doy 182, start of GPS week 860) the IGS Final
combination be discontinued, the current Rapid IGS combination becomes the
IGS Final and the IGS Preliminary (IGP) becomes the IGS Rapid (IGR)
combination.  This way the most precise Final orbits/clocks will become
available within 11 days and the IGS Rapid orbits/clocks will be available
within about 1 day." It was moreover decided that the 38h deadline for the
(now really) rapid orbit will be replaced by a 23h deadline, allowing it to
make available the official IGS Rapid Orbit with a delay of 24h.  This is
of course only possible if the data are available at the ACs about 6 hours
after midnight UT (!).  Again these changes shall be implemented on 30
June, 1996.

It became clear at the workshop that there is considerable interest in 1-2
day predicted orbits.  This is why IGS Analysis centers will start
producing 1- and 2- day predicted orbits.  The interest in predictions
became even more apparent at the business meeting, which is why the IGS
AC-coordinator will be asked to study options leading to the production of
an official IGS predicted orbit.

Mike Watkins from JPL presented a very encouraging agreement of few
centimeters of SLR measurements to GPS satellites (PRNs 5 and 6 are
equipped with a Laser reflector) with distances derived from individual
and the combined IGS orbits. He addressed in particular the importance of
modeling the actual attitude of the GPS satellites during eclipse phases.
It was also agreed that SLR data at present would have little impact on IGS
orbits, but that more SLR data would be most desirable for calibration
purposes. There were indications that a concentrated and coordinated SLR
observation campaign of PRNs 5 and 6 might take place in fall 1996.

Clyde Goad from OSU presented a very elegant and most efficient triple
difference algorithm which was successfully used for orbit determination
and estimation of erp-series. It was pointed out that the approach is
equivalent to a correct double difference scheme (without ambiguity
resolution) because mathematical correlations of the triple differences are
modeled correctly.

Tim Springer from CODE presented first experiences using the "new" orbit
model developed in Bern. There are indications that the model is
particularly well suited for orbit predictions.

EARTH ORIENTATION (chair Ray/McCarthy)

The session was opened with a review of the method developed and applied by
the IGS Analysis Coordinator to produce the combined IGS EOP series. The
review was presented by Pierre Tetreault. In the next presentation by
Marshall Eubanks we were reminded that the IGS combined EOP series agree
very well with the VLBI derived values. Periodic variations seen in the
differences "IGS - IERS EOP series" could be attributed to smoothing
effects in the IERS series which disappeared after a review of the IERS
algorithms to produce the combined series.

Only the x- and y- components of IGS polar motion series have been
extensively used by the IERS.  The GPS-based length of day (lod) or UT1-UTC
drift values were not given much weight by the IERS so far.  The
presentations by Jim Ray and Daniel Gambis revealed that much more
attention is given to that topic now.  It became clear that GPS-derived
lod values are biased (because of correlations with the dynamical orbit
parameters); it became also clear on the other hand that much very valuable
information is contained in the IGS-derived lod series.  We will
undoubtedly observe in future that these IGS products will play a more
important role in the determination (and the prediction) of the IERS
UT1-UTC series.  This might become true in particular if the correlations
between these drift parameters and the (empirical) radiation pressure
parameters will be more clearly understood.

Dennis McCarthy and Tom Herring pointed out that sub-diurnal EOP variations
play a crucial role for the EOP series derived by the IGS Analysis Centers.
It is true on the one hand that the effects are minimized if constant EOP
values are derived for time intervals covering one or several of these
periods. But in view of the fact that the amplitudes may reach the 1 mas
level, biases of the order of 0.1 - 0.3 mas still may remain in such
series. Tom Herring also pointed out that such effects are difficult to be
seen by IGS analysts because they may be absorbed by the estimated
radiation pressure parameters. It was argued that the well established
diurnal and semi-diurnal terms should be applied by all IGS analysis
centers.

The oral presentations were concluded by a review of the existing and a
preview of the new IERS standards. It was argued that IGS Analysis centers
should follow more closely the IERS standards. If departures from these
standards cannot be avoided this fact should emerge from the ACs'
processing specifications (AC questionnaire).

The recommendations of this session really emerged from the oral
presentations: All analysis centers are asked to follow the IERS
conventions (standards) to the extent possible (something which is
facilitated by making available software source code), all ACs are
urged to update their AC Questionnaire (available at the IGS Central Bureau
Information System) at least once per year and the AC coordinator is asked
to review these schemes in the IGS annual report.  It was further
recommended that the general users use the IGS rapid and preliminary polar
motion series (in future rapid and final) together with the corresponding
final and rapid, resp. IGS orbits.  The IERS further asks the IGS Analysis
Coordinator to develop
a method to combine submitted LOD/UT1 results with the goal to form an
official IGS series of such values.  The series of recommendations was
concluded by the requests to take into account 12h- and 24h- terms in EOP
series using the latest tidal model of Richard Ray (to be made available by
the IERS) and to document the actual procedures of the ACs (which is of
particular importance in this case).  Of course such terms have to be taken
into account in all transformations between the terrestrial and the
celestial frames.


ANTENNA CALIBRATION (chair Mader/Rothacher)

The "state of the art" in anechoic chamber measurements was introduced by
two papers, namely Chuck Meertens from UNAVCO and Bruce Schupler from
NASA/GSFC.  This underlines the broad interest in absolute precise phase
center information.  These presentations were complemented by discussions
of the "in situ" techniques focusing on the differential antenna behaviour
(relative to one antenna or one antenna type) by Gerry Mader from NOAA and
by Markus Rothacher from CODE.

It became apparent that "in situ" calibrations from different groups are in
good agreement and are well suited to correct relative antenna biases.
Some inconsistencies still exist, however, between these in situ and the
chamber test results.  Using the (absolute) chamber test models to correct
the phase center of the Dorne-Margolin antennas leads to an unexplained and
significant scale bias of about 0.015 ppm in global GPS analyses.

It was therefore recommended to make available to all parties interested
the relative antenna phase center models for (if possible) all commercially
available geodetic antenna types stemming from in situ measurements.
Provided that the final tests performed with this set are successful the
IGS Analysis Centers will start using these relative models on 30 June,
1996 at the latest.  This will remove obvious discrepancies e.g.
for sites equipped with Trimble antennas in solutions which did not yet
account for such relative models.  The amount of work invested by
all involved parties is amazing and it was acknowledged that all efforts
are necessary to come to a satisfactory model, eventually.

It was acknowledged that the scale effect resulting from the use of
absolute chamber tests needs to be directly addressed in the future.

SINEX, DENSIFICATION OF THE ITRF (chair Blewitt)

We are now in the middle of the IGS pilot project "densification of the
ITRF through regional GPS networks".  Since September 1995 the IGS ACs
started submitting to the global data centers so-called free network
solutions in a still experimental version of the SINEX format
(Software Independent EXchange format).  Today such series are available
from all seven ACs.  Three institutions (JPL, MIT, University of Newcastle)
are analyzing and combining these weekly products.  The procedures of each
of the centers was presented and discussed in the session.  It became
obvious that the philosophy and the actual procedures were quite different.
The "final" products, on the other hand, agree amazingly well.  Is this
possibly a consequence of the Central Limit Theorem formulated by C.F.
Gauss?  The consequence of these weekly analyses is remarkable:  a
consistent set of coordinates (referring to the ITRF) for all the sites
analyzed by at least one AC are openly available!  It is generally expected
that these activities will make the updating of the ITRF (GPS part) much
easier.

In any case one could draw the conclusion that this first phase of the
pilot project was quite successful. The second phase, where the products of
regional associate analysis centers will be included into these weekly
comparisons, too, is scheduled to start on 30 June, 1996. It was initiated
by the call for participation in January 1996. The proposals are now
evaluated, the "new players" will be introduced by IGS mail soon.

It also became clear that there was too much flexibility in the SINEX
format in the past. A working group is now revising the SINEX format with
the goal to have the weekly AC and AAC contributions transmitted in the
SINEX Version 1.0, starting 30 June, 1996. The AC coordinator is
responsible for finalizing this version, of course in close contact with
the ACs and the new associates!

All recommendations of this session were related to the SINEX format, most
of them were very technical in nature. There was, however, the
recommendation to include the EOP information into the SINEX file which
will require some additional thoughts. In view of the variety of methods
used by the ACs to implement a priori information and to parameterize the
EOP series, the implementation seems to be non-trivial at first sight.
There is little doubt, on the other hand, that the AC coordinator in
collaboration with the ACs will come up with a solution that makes sense.
It was the general understanding, that the inclusion of this information
shall NOT serve the generation of a "new" IGS polar motion series, but
allow it to remove reference frame inconsistencies between solutions in a
more rigorous way.

RECEIVER STANDARDS AND PERFORMANCE (chair Zumberge/Gurtner)

The network performance and in particular data latency was reviewed by
Werner Gurtner and Jim Zumberge. These analyses were based on statistics
routinely made at the global data centers and at some of the ACs. The
result was encouraging in the sense that with a "minor" organizational
effort, it actually should be possible to make the observations (at
least of a sub-net) available to the ACs early in the morning (UT) which
actually would allow them to turn out rapid products within 24h.

Data quality was not well monitored so far within the network. The goal,
to my understanding, still is to have a short information concerning
quality available together with the RINEX data files coming in. Such tools
are prepared right now.

A very interesting and (at least for me) surprising presentation was given
by Dr. Hatanaka from the Geographical Survey Institute (GSI) of Japan. He
presented an algorithm (based on forming differences of the observables)
allowing it to compress the data before transmission by about a factor of
2.4 (in addition to the compression that is already used today). First
experience with the algorithm made by some of the IGS components are
positive.

The following recommendations concluded the session:

A set of stations will be indentified by the Central Bureau together with
the AC Coordinator and the ACs for which data have to be be available at
the ACs at 6 a.m.  UT.  This implies that the data must be available at all
Global Data Centers before 5 a.m.  UT.  Obviously such stations have to be
operated in a fully automatic way.  Data of sites which are used for the
final products must be made available to the ACs within 48 hours.  Should
this step be successful, we would undoubtedly see the (currently rapid,
but future) final IGS orbits and eops with a delay of much less than the 11
days guaranteed so far.

The Central Bureau prepares procedures for the "Hatanaka compression" to be
made available for extensive tests.

The need for improved mechanisms for problem detection and reporting was
clearly seen by the network specialists. A routine monitoring of the entire
network must be put in place by the CB. It was also requested that not only
negative, but also positive feedback should flow back to operating
agencies.

ATMOSPHERIC TOPICS (chair Gendt/Feltens)

The issue of using the IGS network for modeling the troposphere and the
ionosphere was first addressed within the IGS at the 1995 Potsdam Workshop.
Meanwhile a broad discussion of this topic inside and outside the IGS was
taking place. It becomes clear by now that the IGS actually must play an
active role in these fields.

TROPOSPHERE aspects were first looked at from the user's point of view:
Eugenia Kalnay from the USA National Centers for Environmental Prediction
was particularly interested in data stemming from satellites of the type
GPS-MET.  It seems clear that temperature profiles with a high spatial
density are of greatest use in meteorology.

The IGS is not actively involved in the GPS-MET experiment, at present. It
may make available total tropospheric delays which, if accompanied by high
accuracy barometer and temperature measurements may be transformed into the
"total precipitable water content". The IGS network and the IGS analysis
centers have the potential to make available such information with a high
temporal and spatial resolution (of its tracking network) on a routine
basis to the atmospheric physicists. Many are convinced that such time
series are relevant for climatological purposes, and, if indeed rapid
orbits of the "new kind" (see above) will become routinely available, for
weather prediction.

That the IGS is "in principle" ready for such a development was one
conclusion from Neil Weston's presentation about the CORS network.  Met
data are transmitted in near real time for selected sites within this
US-wide GPS network, the met data are processed together with the
receivers' code and phase observations to generate the information required
by the meteorologists. A presentation prepared by Rocken (and presented by
Meertens) demonstrated how well GPS-derived water contents agree with WVR
results, Gerd Gendt's analysis showed that the tropospheric delays as
derived by different IGS ACs are consistent on the level of a few
millimeters now. That the issue of weather prediction is taken serious by
IGS ACs was underlined by presentations from JPL (Bar Sever discussing
methods to use predicted orbits for meteorological studies) and SIO (Fang
presenting methods for near real time meteorology and crustal deformation
using GPS).

It was recommended that MET stations of a defined high quality should be
deployed -- at least in a part of the IGS network.  Met information already
available at the stations or becoming available in the near future shall be
sent routinely in Met Rinex files to the IGS data centers, where they will
be available for scientific purpose.  Steps leading to the deployment of
the appropriate met equipment will be taken before the end of 1996.

It was also recommended that IGS tropospheric delay estimates should be
studied and combined by special associate analysis centers. GFZ is ready to
build up such a center (hopefully) by the end of 1996. Other parties will
be invited through a call for participation.

IONOSPHERE MODELS using data from the IGS network were developed by Schaer
et al from CODE, by Wilson et al. from JPL, by Feltens et al. from ESA,
by Komjathi et al from University of New Brunswick, and by Jakowski et al.
from DLR Neustrelitz. A data set of five weeks of the year 1995 was used
(and is still used) by the "ionosphere groups".  It became clear that
different groups have different goals in mind:  pure GPS-internal use (to
correct e.g. single frequency data or to help ambiguity resolution) is
one goal, calibration of altimetry data another, pure ionosphere research a
third goal.  Methods and models are very different, too.  It seems,
however, that we are now reaching a state where the models of different
groups may be effectively compared.  Such comparisons were presented by the
ESA- and the Neustrelitz- groups using a two-dimensional grid in the single
layer electron shell.  I personally believe that the level of agreement
(few TECUs) and not (yet) unexplained biases are amazing.  It is fair on
the other hand to state that we are still far from the consistency level we
have reached e.g. in modeling the troposphere. Consequently the
recommendations are more modest for the near future: In a first step the
analyses of the 5 weeks event and the comparisons emerging from it will be
concluded. In a next step a common format for the exchange of ionosphere
models is created. A first draft for this format, with the tentative
name IONEX, will be available soon. Only in the more distant future (one
year from now?) a pre-operational production of IGS ionosphere products is
envisaged.

FINAL REMARKS

Although the above summary is rather long it can only give an incomplete
picture of the workshop. It was a meeting deserving the name (label)
workshop and I have no doubt that important decisions and new directions
were the direct result of the considerable amount of work invested in the
preparation of this workshop.  Let me therefore thank all the contributors
to this workshop, and let me congratulate Gerry Mader and Jan Kouba for the
organization of this fine IGS event.


                                                    Gerhard Beutler
                                                    Chair,
                                                    IGS Governing Board



[Mailed From: Gerhard Beutler <BEUTLER at aiub.unibe.ch>]