- Relative weighting of solutions
In a report by P. Sillard (IGN), the robustness of applying Helmert weighting to rescale the variance
matrices of the submitted solutions was shown to be better when based on the the global combination than
on combinations of solutions of the same technique only. It was agreed that the relative weightings should
be based on the global combination.
- Editting and outlier detection
An examination of the preliminary solution residuals shows a number of cases where an individual solution
for some particular site differs from the combination by a relatively large amount. It was proposed that
an editting criteria be applied to reject observed coordinates with residuals exceeding 4 times their
rescaled errors. All such rejections should be reported so that analysts are aware.
In addition, any other data problems and significant residuals for local site ties should be reported in
the final ITRF2000 report. It is particularly important that information about the local ties is fed back
to the station managers.
- Datum: ITRF2000 origin
It was agreed that the specification of the origin and its linear rate of change be defined by the weighted
combination of all the SLR solutions (7).
- Datum: ITRF2000 scale
It was agreed that the specification of the scale and its linear rate of change be defined by the weighted
combination of the all the SLR and VLBI solutions (10).
There was extended discussion of general relativistic time scales and the scale differences that result using
either TDT/TT (terrestrial time) or TCG (geocentric time). G. Petit (BIPM) described the basis for the current
IAU/IUGG recommendation to use TCG. However, for the ITRF, the time scale choice is purely a matter of
convenience with the only observable consequence being a scale difference of ~0.7 ppb.
|
