We processed some data and noticed we had a bad ellipsoid ht. on a fixed point. We constrained the survey to that point. The survey processed and adjusted fine according to our reports.
We proceeded to falsify ellipsoid "busts" on other projects and found that the ellispoid height changes didn't change our X,Y coords. Why?

Hi:

Interesting question.

Can you supply a couple more details:

(1) How large were the ellipsoid height "busts"

(2) Did you transform to a Local Grid system, and if so how many local points did you use to calculate the transformation

(3) How many points did you hold fixed when you did the Adjustment? Were 3 coordinate values held fixed for each control point -- example: (Northing, Easting, and Ellipsoidal Height held fixed?)


Regards,

Richard
Magellan tech. Support

The amount of relative error in the Northings and Eastings of your ROVER points will depend largely on the magnitude of the BASE ellipsoid height error. In other words, the "absolute" error of the BASE position in WGS '84 leads to relative errors in the Northings and Eastings of ROVER points.

At the vector processing stage, a general rule of thumb is that for every 20 meters of radial error in the fixed base station position, an error of approximately 1 part per million (ppm) occurs in the length of the calculated baselines. This is assuming that the BASE coordinates have been fixed with a 0 standard deviation.

So, an ellipsoid height "bust" of 20 meters would then cause on average about 0.01 meters of error in a baseline length of 10000.000 meters.

This baseline length error would then cause a relative error of about 0.01 meters in the processed Northings and Easting of any ROVER positions that use the BASE position as a known point. The actual magnitude of the Northing and Easting error will depend on how the baseline length error is distributed in the DeltaX, DeltaY, and DeltaZ components of the baseline.


Then, at the minimally constrained network adjustment stage, the baseline errors will also cause errors in the Northings and Eastings of the ROVER points. The Tau test will probably not detect the baseline errors because they are systematic. The error detection will be most difficult when the satellite geometry is similar for each baseline.

Then, at the constrained adjustment stage you would have better success in detecting these relative errors (assuming that you have accurate control point coordinates).

So, for the best relative accuracy, the "absolute" errors of the BASE coordinates need to be as small as possible.

The "absolute" accuracy of the BASE is of course even more critical when the survey needs to be tied to an accurate coordinate system such as the State Plane Coordinate System of 1983. Also, if accurate mean Sea Level Heights are to be produced from the Ellipsoidal heights, the "absolute" ellipoid height error will translate directly into the Mean Sea level heights.


Richard

Magellan Technical Support

Sorry for not getting back to you sooner. Thanks for the reply.
This was a Fast Static Survey only.
No base or Rover. We used 4 Receivers.

The bust we falsed in was about 10 meters.

The knowns were from a well established GPS network.

Thank You

Ok, a 10 meter height ellipsoid height bust will cause about 0.5 ppm error in your baselines. Remember, the 0.5 ppm amount is an average value.

I did a test on some static data and I saw a 0.6 ppm change in the baseline length when I changed the BASE station ellipsoid height by 10 meters. This is close to the expected error of 0.5 ppm.


Maybe your baselines are very short? This would explain why you don't see any change in your Northings and eastings at the millimeter level.


Why do your heights have to be modifed by 10 meters if you are using a well established GPS control network? Is it because you have to try and match a local height system?







Richard