Technical description of ivsd04it_dsnx:
=======================================

  VLBI group of the Geodetic Institute of the University of Bonn
         (Markus Vennebusch, Vennebusch@uni-bonn.de)
                         April 12, 2005


The ivsd04it_dsnx series is a combination product by accumulation of normal 
equation matrices. The normal equation matrices in these files are 
unconstrained (datum-free).
Input series are currently contributed by BKG, DGFI, GSFC, MAO, SHA and USNO.

Analysis center:                                                   | Software:
-----+-------------------------------------------------------------+-----------
BKG  | Bundesamt fuer Kartographie und Geodaesie, Leipzig, Germany | CALC/SOLVE
DGFI | Deutsches Geodaetisches Forschungsinstitut, Munich, Germany | OCCAM
GSFC | Goddard Space Flight Center, Washington D.C., USA           | CALC/SOLVE
MAO  | Main Astronomical Observatory, Kiev, Ukraine                | SteelBreeze
SHA  | Shanghai Astronomical Observatory, China                    | CALC/SOLVE
USNO | U.S. Naval Observatory, Washington D.C., USA                | CALC/SOLVE
-----+-------------------------------------------------------------+-----------


Purpose of this solution:
=========================
The main objective of ivsd04it_dsnx is to provide IVS' contribution to the
IERS ITRF2004 project for the combination of different geodetic space 
techniques and the rigorous determination of station positions and earth 
orientation parameters.

The so-called long-term solution (data from 1980 till 2004) is used as
a basis for routine combination.


Detailed description:
=====================
This approach of combining VLBI solutions is based on the principle of 
addition of normal equation matrices and thus begins at an earlier step of 
analysis. This approach is also known as "Adjustment of groups of observations".
This approach uses datum-free normal equation matrices of VLBI sessions which 
are provided as SINEX files. 

The combination if performed in the following way:

Apriori values:
---------------
Before accumulating the individual normal equation matrices the matrices are
transformed to a common set of apriori values. Therefore one particular analysis
center is chosen whose apriori values are used as apriori values for the normal 
equation matrices of all analysis centers.
These values are listed in SOLUTION/APRIORI-block.

Scaling of solutions:
---------------------
Scaling factors are used to balance the magnitudes of the individual matrices so
that all solutions have equal weights.         

In a first step a mean is computed for all main diagonal elements of station 
coordinate components (sub-trace) for each normal equation matrix.
The average of the means over all analysis centers contributing input to the
session is then used to calculate the scaling factor for each input matrix from 
the respective ratio:

  a. For each analysis center (AC): 
     (m = number of analysis centers, 
      n = number of stations included in this solution)

     The mean of the main diagonal elements for station components is 
     computed by: 

    TR                                                    SUM_{i=1}^{3*n} Nii 
      AC  = TR_AC_i = Mean trace (coordinate elements) = --------------------
        i                                                       n * 3

  b. The mean of these traces is computed by:

              SUM_{i=1}^{m} TR_AC_i
     MEANTR = ---------------------
                       m

  c. Each normal equation (incl. its right hand side) is scaled 
     by the ratio MEANTR / TR_AC_i:

     N      N        MEANTR
      AC  =  AC   * ---------
        i      i     TR_AC_i


Generation of SINEX file for accumulated solution:
--------------------------------------------------
After applying scaling factors the normal equation matrices of the individual
solutions are accumulated. The resulting normal equation matrix is then written
into a new SINEX file (DATABASE_CODE_ivsd0001.snx). 
Since the original normal equation matrices are datum-free or singular, the 
accumulated normal equation matrices are still datum-free. So there is no 
SOLUTION/ESTIMATE section in this file, since no datum information has been 
imposed. 

Outlier detection:
------------------
In order to detect outliers, EOP from all individual solutions have been
computed and have been compared with the combined EOPs.
Here, all stations are fixed to a best estimate TRF. A solution has been 
marked as an outlier if its difference with respect to the combined solution 
exceeded 0.450 mas (for polar motion) or 0.030 ms (for UT1).
The respective input solution has then been excluded from the combination!


Estimated parameters:
=====================
Parameters combined are:
ALL station coordinates, x-pole, y-pole, UT1-TAI, x-pole-dot, y-pole-dot, LOD. 

As mentioned before there are no estimated parameters at all since there is
no datum information imposed.


Software:
=========
- The processing of the individual matrices is done using a software from
  the German Geodetic Research Institute (DGFI), Munich, called DOGS-CS. 
- A software called combine_sinex has been developed at the Geodetic Institut
  of the University of Bonn and is used to process the input SINEX files, 
  perform checks of the input data, to compute scaling factors and to visualize
  results.


Contact:
========
VLBI Group of the Geodetic Institute of the University of Bonn
Nussallee 17
D-53115 Bonn
Germany
eMail: Vennebusch@uni-bonn.de