Devoti gave an overview of the proposed service (Appendix 15). The fundamental elements will be (i) preprocessing, (ii) combination and (iii) QC and delivery. At the moment, about 70% of the software is ready. Testing is being performed with the 1999 “A” series of solutions. These monthly solutions show an rms difference w.r.t. ITRF2000 of 5-9 mm. The combined EOPs fit better to the IERS C04 series than the individual ones, although the CSR solution performs slightly better. Weighting factors are applied to each of the 5 input solutions such that the χ-square for each is more-or-less identical (resulting in scaling factors applied to the covariance matrices of between 1 and 100; for linear uncertainties the square root is to be taken).
Kelm (Appendix 16) discussed the status of the DGFI combination procedures: the level of automation (fully), its status (100% ready), test results (based on the 1999 “5A” solutions, plus IERS SINEX campaign results). The following steps are taken: (i) deconstrain the normal equations and compute eigenvalues, (ii) reduce bias parameters and obtain a reduced normal equation, and (iii) apply the minimal constraint and solve for the parameters. Weight factors range from 0.12 to 1.5 (on a linear scale).
Pavlis is in the process of automating the procedures, and expects to be “up and running” in the course of May (Appendix 17).
Nurutdinov sketched the procedures in his TANYA software, which has been updated to also include EOPs: (i) validation, (ii) unconstraining, (iii) estimation of the GNET combined solution, (iv) Helmert transformation, (v) outlier removal, (vi) variance component estimation, (vii) product generation, and (viii) reporting. Tests have been performed with the “BB4” solutions covering the period January-June 2002 (Appendix 18).