Real-Time Kinematic (RTK) GNSS technology is now standard practice on construction sites. Surveyors use it for control layout, grade checking, machine guidance, and as-built verification.
When an RTK receiver reports a fixed solution, the displayed horizontal accuracy, often 1–2 centimeters, feels reassuring. But the vertical component of that same solution tells a more complicated story.
Vertical accuracy in RTK surveys is consistently weaker than horizontal accuracy under ideal conditions, and it degrades significantly in the active, cluttered environment of a construction site.
Understanding why this happens is important not just for surveyors, but for the architects, engineers, and project managers who rely on that data to make consequential decisions about grades, drainage, and structural elevation.
The Structural Gap Between Horizontal and Vertical RTK Precision
The difference starts with geometry. Because receivers can only track satellites in the upper hemisphere, the vertical component of the position solution has fundamentally weaker geometry than the horizontal.
This is measured as Vertical Dilution of Precision (VDOP), a unitless value that quantifies how satellite geometry amplifies positioning errors in the height dimension.
NOAA’s National Geodetic Survey (NGS) Technical Memorandum NGS-58, the federal standard for GPS-derived ellipsoidal heights, recommends collecting observations only when VDOP is below 6, and staggering repeat observations by three to four hours to capture meaningfully different satellite geometry.
Peer-reviewed research published in Applied Geomatics found that the absolute error of ellipsoidal heights from RTK and Network RTK measurements typically ranges from 3 to 5 centimeters and that receiver-reported RMS precision values can only be treated as a measure of precision, not absolute accuracy.
Multipath: The Dominant Error Source on Active Job Sites
On a construction site, the most damaging threat to vertical accuracy is multipath interference. Multipath occurs when GNSS satellite signals bounce off nearby reflective surfaces, steel frames, metal decking, construction equipment, and temporary structures, before reaching the receiver antenna. The receiver interprets the delayed, reflected signal as part of its positioning solution, introducing errors that standard differential corrections cannot eliminate.
Research published in the American Journal of Geographic Information System examined how different construction-relevant surface materials affect multipath. The study found that aluminum and other metals caused the greatest signal distortion, followed by glass and wood. Critically, the research confirmed that changes in the vertical component were approximately twice those in the horizontal directions, which is a direct consequence of the satellite geometry described above.
Work in GNSS Multipath Errors and Mitigation Techniques notes that multipath is typically not modeled in standard GNSS data processing pipelines, and can contribute errors of several centimeters to the carrier-phase measurements that RTK relies on.
On a job site where metallic structures are constantly being erected, moved, and reconfigured, the multipath environment changes continuously, making it impossible to model or predict in real time.
The Ellipsoidal Height Problem: What RTK Actually Measures
There is also a fundamental conceptual issue in how RTK vertical values are used on construction sites. RTK receivers output ellipsoidal heights, distances measured from an idealized mathematical model of the Earth’s surface.
But construction grades, drainage calculations, and structural elevations are referenced to orthometric heights, which measure vertical distance above the geoid. As established in Converting GPS Height into NAVD 88 Elevation by NGS researchers Milbert and Smith, these two surfaces differ across the conterminous United States by roughly 8 to 53 meters.
Bridging that gap requires a geoid model, and as documented in NOAA Technical Report NOS NGS 72 — GEOID18, the uncertainties in that conversion do not include errors inherent in NAVD 88 itself; meaning the true vertical error budget is always larger than the receiver display suggests.
What This Means for the Built Environment
For engineers specifying finished floor elevations, architects coordinating with civil consultants, and contractors setting grade stakes, RTK vertical limitations are both a quality control issue and a liability issue.
Construction defects traceable to elevation errors are a recurring source of professional liability claims.
Best practices include anchoring RTK surveys to benchmarks with established NAVD 88 orthometric heights, verifying elevations with check shots before and after each session, monitoring VDOP and avoiding height-critical work when it exceeds recommended thresholds, and positioning rover antennas away from metallic structures to reduce multipath exposure.
RTK technology is powerful and, when used correctly, well suited to most construction layout tasks. But vertical accuracy on active job sites requires discipline, verification protocols, and a clear-eyed understanding of the factors that cause it to degrade.
