Why Two Correct Receivers Can Disagree
Two receivers can occupy the same point, agree on the GNSS position, and still display different elevations. The disagreement may begin only after the position is solved, when each receiver or application applies a different geoid model, datum relationship, or vertical setting.
Begin with What They Agree On
Imagine placing two survey-grade receivers over the same monument. Both obtain an RTK FIX solution. Their latitude and longitude agree, and their ellipsoid heights differ by only a few hundredths of a meter. At this stage, the evidence suggests that both positioning engines have solved essentially the same antenna location.
Then the crews compare the elevations shown on their screens. One receiver reports 70.95 meters. The other reports 70.62 meters. Because the horizontal coordinates agree, the difference seems mysterious. It is tempting to assume that one receiver has a vertical bias or that one manufacturer is less accurate. But the divergence may have occurred after both receivers finished solving the GNSS coordinate.
When two systems agree on ellipsoid height but disagree on orthometric elevation, the GNSS solutions are not the first suspects. The investigation should move downstream to the geoid model, datum, units, and software configuration.
Follow Both Processing Paths
Receiver A solves an ellipsoid height of 42.60 meters. Its software evaluates Geoid Model A at the monument and obtains N = -28.35 meters. Using H = h - N, it reports an orthometric elevation of 70.95 meters.
Receiver B solves the same ellipsoid height of 42.60 meters, but its software evaluates Geoid Model B and obtains N = -28.02 meters. It therefore reports an orthometric elevation of 70.62 meters. The two displayed elevations differ by 0.33 meter, exactly the difference between the two modeled geoid heights.
Nothing in this example requires either receiver to have measured the antenna incorrectly. Both began with the same h. Their numbers separated only when different values of N entered the vertical equation. Once the processing paths are written side by side, the apparent receiver disagreement becomes a model disagreement.
If Receiver A and Receiver B report the same latitude, longitude, and ellipsoid height, at what step did their elevations become different? Not during satellite tracking or RTK ambiguity resolution. The paths diverged when different vertical information was applied to the common GNSS solution.
Why Would Two Models Disagree in the First Place?
It is fair to ask why two respected geoid models would not simply agree with each other. A geoid model is not a direct measurement of gravity alone. National models like these are typically built by combining a gravimetric geoid computed from gravity data with GPS observations taken on the existing benchmark network, then adjusted so the model agrees as closely as possible with that historical network. Because each model is fit to different data at a different time, two models can legitimately produce two different values of N at the exact same point.
A national geoid model is not a pure measurement of gravity. It is a gravimetric surface deliberately adjusted to agree with an existing benchmark network. That is why updating the model, not just the receiver, can change the reported elevation at a fixed point.
Correct Does Not Mean Interchangeable
The fact that both calculations are internally correct does not mean both elevations are equally appropriate for the project. A global geoid model may be suitable for general navigation, while a national model may be designed to produce elevations compatible with a specific vertical datum. A project may also rely on a local calibration tied to nearby control. The correct result is therefore not determined by which number looks more familiar. It is determined by which processing path matches the project reference.
This is why the model name and version must travel with the elevation. Without that information, the final value cannot be reproduced or evaluated properly. Recording only H hides the step that created the difference and encourages future users to treat two unlike elevations as though they were directly comparable.
If two receivers display different elevations, one of them is not automatically wrong. Both may have performed their configured calculations correctly. The real question is which configuration is compatible with the intended vertical datum and project control.
The Same Problem Can Occur Inside One Receiver
A comparison between two instruments makes the issue easy to see, but the same divergence can occur within a single workflow. The receiver may output ellipsoid height while the field application applies a national geoid model. A different application may use a built-in global model. Post-processing software may apply yet another transformation, or it may assume the imported height is already orthometric and leave it unchanged.
Each result may trace back to the same underlying GNSS coordinate. The difference is not necessarily caused by changing observations; it may be caused by changing assumptions about what the incoming height represents and which vertical reference should be used. That is why troubleshooting must follow the coordinate through every device and application rather than stopping at the receiver status screen.
A Practical Comparison Method
When two systems disagree, compare them in the order the coordinate was produced. First compare latitude and longitude. Then compare ellipsoid height before any geoid conversion. If those values agree, record the geoid model and version used by each system, the value of N evaluated at the point, the vertical datum named in the project, and the units of the final elevation.
This sequence separates positioning differences from reference differences. If h disagrees, investigate the GNSS solution, antenna setup, base coordinate, and observation conditions. If h agrees but H does not, the investigation belongs after the position solution. The vertical equation then becomes a tracing tool: the difference in H should be explainable by the difference in N or by another documented vertical transformation.
Two correct receivers can report different orthometric elevations from the same GNSS coordinate because the displayed value depends on more than the RTK solution. Compare ellipsoid height first, then identify the geoid model, vertical datum, units, and any downstream transformation that produced H.
Different geoid models explain one class of elevation disagreement. The next chapter introduces the piece still missing from the comparison: the vertical datum itself, and why a benchmark elevation is incomplete without it.