GIVEN THAT THE A-PRIORI VARIANCE OF UNIT WEIGHT AND THE A-POSTERIORI VARIENCE OF UNIT WEIGHT SHOULD BE NEAR 1, WITH THE SCALING FACTOR SET TO 1
I SEEM TO REGUALRLY ENCOUNTER THE FOLLOWING REPORT, WITH THE DIFFERENCE BETWEEN THE TWO VARIENCIES AS ZERO, WHICH APPEARS TO RESULT IN A FAIL CONDITION.
ANY THOUGHTS?
Chi-square test: failed
Lower limit: 1.000000
Upper limit: 1.000000
Chi-square: 0.000000

Hi:

These values in the Chi-square test will happen in Solutions (or during any network adjustment) when there is no redundancy in the network i.e. only "radial" GPS vectors present -- no repeat vectors, and only one GPS vector coming into each GPS point.

The residuals will always be zero in the case of no redundancy as well. This might appear to be a good situation, but there could be undetected problems, because there is no check on the GPS measurements in the network. In some ways, it's similar to having only "sideshots" -- there is no double check on the "sideshot" measurements that have been made

We (Thales Navigation) always recommend the use of redundancy in GPS control networks -- i.e. with at least 2 independent vectors coming into every GPS point, there is a check on the measurements, and the network adjustment statistics can be used to help determine the quality of the GPS measurements.

Ideally, as stated in the previous thread, the variance of unit weight (and also standard deviation of unit weight) should be 1 when the processed vector scaling factor is equal to 1. If not, there may be blunders in one or more GPS vectors that will require data cleaning, and possibly elimination from the network adjustment altogether.

Best regards,

Richard Phelan

Thales Tech Support

Richard, does this mean every GPS point should be occupied twice?
Or does the number of satellites observed enter into it?

I remember the brocure "Introduction to GPS Surveying" that came with our Locus receivers basically said a certain number of observation stations should be occupied twice, but not all of them.

Thanks.