I have a fixed gps base control point with assumed coordinates at approx. 35d nlat & 100d wlong, approx. +2090 ft. above msl. On this assumed plane I recorded a USC&GS BM approx 25k ft. from the base.
My question is..what is the predictable error between the difference in elevations of the two points (base and BM) as observed by the difference in "Z" values of the gps observation and the 'true' difference if I were to run a level loop?
Paden
I hate to say it, but your post makes no sense!
Narrow the "field" a little, and define your terms better.
🙂
Loyal
More details on the specific GPS techniques would be helpful. I would also suggest the Ngs webinars on GPS derived ortho heights. Part one lays the ground work for large network projects and somewhere in part two they touch on accuracies expected from rtk or rtn measurements. Lots of variables in this scenario...
Some of the things that will affect the response:
1. Equipment used, including antenna types.
2. Good GPS obstruction conditions at both points.
3. Types of observation recorded (static or RTK? L1 only, L1+L2?).
4. Duration of observations at each point.
5. Duration of overlap between observations at the points.
6. Level of confidence in your antenna heights.
7. Geoid model accuracy at each point.
8. Degree of divergence between tropospheric conditions at the points.
If you're in the continental US, have at least 60 minutes of clean L1/L2 data at sites with no obstructions or multipath sources, use correctly-modeled antenna offsets with reliable antenna heights, an accurate geoid model, have similar weather at each station, and use dependable processing software, I'd say you stand a pretty good chance of getting within 2 cm of the NAVD88 height difference between the stations. I'd guesstimate the 95% uncertainty at 5 cm, and the 99% uncertainty at 8 cm.
What does OPUS say?
You are going to need to have some actual data to come up with a prediction. You will need to be able make some comparisons, and not rely only on "predicted" accuracy reported by post processing software.
More data is better. Six hours should give you a good variety of satellites. More than 24 hours would be a waste as you will just be repeating the previous day's constellation.
Over 25k feet a geoid model might or might not make much difference in fairly level country. It would be good to be able to compare a geoid modeled differential to a differential based only on height above ellipsoid.
With enough data you should be able to compute a good ionospheric model and have a reliable float solution to compare to a fixed ambiguity solution.
Generally, the closer the agreement of the results of different solutions, the better the reliability of the results.
Howdy,
Mr Frame provides a good list of factors to consider in the determination of a GPS-derived height. When dealing with short baselines (<10 KM) and small differences in elevation (tens of meters) between ends of the baseline, we expect the impact of the troposphere to be correlated between sites. The choice of tropo model and its parameters has large impacts on height determinations. A solution other than (correctly) fixed integer should not be used.
A correctly-derived NAVD88 height as indicated requires not only a good NAD83 ellipsoid height but also an accurate model of the ellipsoid-geoid separation (GEOID 09 is the best available). The national accuracy of GEOID 09 is no better than 2 cm. In mountainous areas it is much less accurate.
Remember as well that hybrid geoid models use GPS on BM data. This data set is based on the assumption that the NAVD88 BM was correctly determined and was still valid and that the ellipsoid height was also correctly determined.
It is easy to calculate the difference between the published NAVD88 height and the sum of the ellipsoid height and geoid model.
If there is a large difference understanding *why* is the problem? Is the published value still valid? Has something happened to the physical monument since its height was determined? How was its height determined? Are there any notes on the data sheet indicating problems with it? These would include "NO CHECK", "VERTCON", "VERT ANG", "N HEIGHT", or "RESET."
Cheers,
DMM
Is this a problem between NGVD 29 and NAVD 88?
Improper application of geoid model?
Check manufacture's spec.
For the error in the vector itself consult your instrument specs. Typical spec is around 10mm + 5ppm, and that should be considered optimum. So that would account for around 0.05 feet. Perhaps twice that in practice, perhaps not. It does very much depend on the method used and the site conditions. A few seconds of RTK? Multipath? Or hours of Static in the clear?
Then there is the error in measure up. At both ends. Add at least a couple of hundreths for that, perhaps more.
Then, the geoid model. As mentioned by others, allow 2cm , which roughly equals 0.06'. Can be more.
So you are looking at 0.13' error in the GPS elevation at best, and doubling that is easy. If you run multiple OPUS-S sessions on a single point you will see elevation spreads in the vicinity of this figure.
As for the levelling, assuming digital levels and solid turning points on a cool day -suppose maybe 0.001 per turning point and maybe 50 turning pts on the outbound leg and 50 back. These sorts of errors do tend to cancel out so you might well close this loop to a few hundreths.
In NC, GPS is only to be used in Class C Surveys (Vertical Control)
"For Class C vertical control surveys in North Carolina, the vertical error in feet shall not exceed 0.30 times the square root of the number of miles run from the reference datum. The vertical error in global position surveys shall not exceed five centimeters relative to the referenced benchmark(s) at the 95 percent confidence level (2 sigma) accuracy as defined in Federal Geographic Data Committee Standards."
