Elevations are tricky to evaluate in this scenario. The positions were all on a concrete slab, but there was an actual elevation change of 1cm+ according to the RTK positions. From the dataset, there appears to be a 60cm difference in the vertical datums between HAS and PointPerfect Solutions at this specific area (expected).
Once that Vertical Shift is removed between the 2 correction systems, we get the following Vertical Comparison between HAS and PointPerfect:
Vertical Residuals using fixed +60cm elevation bias between datums (Each pair, in cm)
- -29.53
- -6.95
- -5.98
- -4.96
- +3.46
- +4.72
- +2.39
- +2.58
- -1.08
- -1.11
That’s probably a perfect example of how multi-path influences Elevations more than X and Y. The HAS Elevations tighten-up significantly as we move away from the building, even with a choke ring antenna.
I understand these results are a lot better than the HAS service can technically provide…but it’s an honest test. I do not have an explanation. The same test might produce worse results in the future under different conditions.