Did a kinematic survey for a bathymetry survey along a beach front property last week.
My setup was 2 base located ~2km apart & a rover fixed on a small boat picking up sonar data. All units were recording every 1 second.
Post processed using GNSS Solutions. Usually I just use 1 base + 1 rover but in this case had extra unit + operator not needed elsewhere so might as well use resources available for redunduncy.
In GNSS Solutions, I only used vectors that had a "fixed" report & discarded all vector results with "float" results as well as those with a confidence level error above 0.30m.
So I used those points in computing the depth points. Had some strange results for the contour so I compared the z values from the 2 bases. I only used the results from 1 base in the computation for the rover points.
When I compared the points that are problematic, I found out that the z points from the 2 bases had discrepancies of 3-4 meters in the resulting ellipsoidal heights! Rechecked if these were "fixed" vectors and indeedd they were "fixed" and confidence level error is less than 0.30m.
It was a good thing that I had 2 bases to compare results. How do you guys check for abnormal z value results using only 1 base + rover?
The difference between fixed and float is simply a statistical confidence, not a hard line GOOD or BAD. A float solution may have the proper integer resolution and be correct, while a fixed solution could have the wrong integer and be incorrect.
For kinematic processing, there is software available that will process the session forwards in time and then reverse in time and compare the coordinates at each epoch. This is one of the best ways of confirming kinematic accuracy and is independent of a float/fix solution. The idea is that ambiguity resolution is resolved at opposing times in the session and verified to an epoch.
Novatel GrafNav is one of the better software packages in that regard.
Without this software, your solution may be what you have already done: processing the data to two different bases, comparing results and rejecting anything that does not match within tolerance.
Thanks for the advice on the software. Visited their website & there is no demo version.
I tried to check the data from the 2 bases to the rover that had errors in Z values >4m &
could not really find degradations in PDOP values. Very strange to find "fixed" solutions from
2 bases that would have these much variances.
Usually I rely on just 1 base + 1 rover setup for kinematic surveys. Now I am wondering the errors
made in past surveys wherein I rely on 1 base + 1 rover setup.
Are there any 'Repeat Vectors' listed?
M