Sect_9.4.6

9.4.6 Modifying the Stochastic Model

A PROCEDURE FOR INCORPORATING TRIVIAL BASELINES INTO A SECONDARY ADJUSTMENT

A Procedure for Incorporating Trivial Baselines into a Secondary Adjustment

CRAYMER & BECK (1992) have demonstrated that by including all baselines (reduced in single baseline mode), trivial and independent, in a network adjustment, such a secondary adjustment of baselines is close to being equivalent to a rigorous simultaneous multi-baseline phase reduction in the case of a single session adjustment, or a simultaneous multi-session phase reduction in the campaign network adjustment case. However, there are several conditions that have to be met:

Include all single baseline results.
Scale baseline VCVs by , where n is the number of receivers operating during the session.
The same ambiguities are resolved in single baselines as would have been for the multi-baseline processing.

This certainly confirms the intuitive arguments presented earlier that including all baseline somehow "reconstructs" the missing between-baseline correlation information. The scaling of VCVs by the factor would overcome the problem of over-optimistic network VCVs.

A similar result has been verified by HAN & RIZOS (1995b), except that they suggest that it is the co-factor matrices that must be scaled by n/2, not the VCV matrices. Then each (scaled) co-factor matrix, for each baseline solution, is converted into the corresponding VCV matrix upon multiplication of a single session variance factor. This session variance factor is the mean of all the variance factors. The procedure for scaling the VCV matrices also includes the additional "standardisation" factor arising from the need to take into account the physical correlations between successive epochs of GPS observations (next section).

Back to Chapter 9 Contents / Next Topic / Previous Topic