There is a specific kind of frustration that comes from delivering a dataset that looks clean, held fix throughout the session, reported centimetre-level precision in the controller, and still doesn’t close against control.
The receiver did exactly what it was supposed to do. The error was introduced before the first point was ever collected, in a setup decision that took ten seconds and felt completely routine. That’s what makes GNSS mistakes genuinely difficult to manage: the bad data is indistinguishable from the good data until something downstream exposes it.
Antenna Height: Small Entry, Large Consequence
Multipath: The Fix Light Stays Green
Baseline Length and the Limits of CORS
Coordinate System and Unit Mismatch
Ellipsoid Height Delivered as Orthometric Height
Level Bubble Calibration
Setup Documentation That Holds Up Later
Preparation Is the Only Reliable Error Management
Antenna Height: Small Entry, Large Consequence
Antenna height error sits at the top of this list because it produces a systematic vertical offset across every single observation in the session, and it leaves no trace in the receiver output. The instrument reports what you told it. If you told it wrong, every coordinate you collect is wrong by exactly the same amount in the vertical, which means your data looks internally consistent and passes all the checks that don’t involve an independent control point.
The source of the error is almost always one of three things: measuring from the wrong reference point on the instrument, entering the value in the wrong unit, or carrying a number incorrectly under field pressure. Different manufacturers define antenna height differently, some to the bottom of the receiver housing, some to the antenna reference point, some to a physical shoulder on the mount. These are not interchangeable, and assuming one convention when your instrument uses another introduces a fixed error equal to the offset between those two reference points.
Know your instrument’s measurement convention precisely. Verify that your field software is applying it correctly at the start of each session. Check the entered value before you collect your first point. This takes two minutes and eliminates one of the most common sources of systematic error in GNSS fieldwork.
Multipath: The Fix Light Stays Green
Multipath is the error that catches experienced operators as often as new ones, because the receiver gives no obvious indication it’s happening. When satellite signals reflect off buildings, vehicles, water, steel equipment, or compacted soil before reaching the antenna, the receiver processes a signal that has travelled a longer path than the direct line from satellite to antenna. The computed position is displaced by an amount proportional to the reflection geometry, and because the error is geometrically consistent, it doesn’t average out with repeated observations the way random noise does.
Antenna placement discipline is the primary defence. Keep your base away from large metal structures, position it high enough to reduce low-angle incoming reflections, and use an antenna with a proper ground plane that attenuates signals arriving from below the horizon. For rover work, develop the habit of noting the environment at each setup, a shot taken two metres from a steel culvert is a meaningfully different observation than one taken in the open, regardless of what the satellite count shows.
Baseline Length and the Limits of CORS
The RTK error model is explicit: accuracy degrades at a rate of approximately 1 ppm of baseline length on top of the base system noise. At 10 kilometres that’s a 10-millimetre addition. At 40 kilometres it’s 40 millimetres, before accounting for the fact that atmospheric and ionospheric conditions between the reference station and your rover’s location have also diverged, which they always do, and increasingly so with distance.
In Canada, CORS infrastructure is dense in southern urban corridors and thin across most of the country north of that band. Surveyors working in northern British Columbia, the boreal regions of Alberta and Saskatchewan, or anywhere across the territories regularly face baselines that are longer than optimal, and sometimes longer than defensible for the accuracy class they’re working to. Accepting a 60-kilometre baseline on a cadastral survey because the nearest CORS station is 60 kilometres away is not a workflow. It’s a risk.
When baseline length exceeds 20 to 30 kilometres, setting up a local base over a known control point and running your own correction infrastructure is the correct approach. It eliminates baseline-dependent error within your working area entirely and gives you correction geometry that a distant CORS station cannot provide.
Coordinate System and Unit Mismatch
This error typically isn’t caught in the field. It surfaces when data is brought into CAD or GIS software and fails to overlay correctly with existing work, sometimes days or weeks after the session. The two most common versions are a metres-versus-feet unit mismatch in the projected coordinate system, and selection of a coordinate system that’s adjacent to but not identical to the one used by previous survey work on the same project.
In Canadian practice, NAD83 and its provincial realizations are both in active use, and the differences between them are not negligible at survey accuracy levels. Before any project, explicitly confirm the coordinate system and linear unit used by existing data you need to connect to. Verify it in your field software settings directly, don’t assume it carried over from a previous job or template. Sites like epsg.io let you inspect coordinate system metadata including linear unit before you leave the office.
Ellipsoid Height Delivered as Orthometric Height
GNSS measures height above the ellipsoid. Engineering design, drainage analysis, and topographic mapping require orthometric height referenced to the geoid. The separation between these two surfaces varies across Canada from roughly 10 to 40 metres depending on location. Delivering a dataset in ellipsoidal heights when the client needed orthometric heights isn’t a minor discrepancy, it makes the elevation data unusable for most practical applications.
The correct vertical reference for Canadian survey work is CGVD2013 with the HT2_2010 transformation grid. Confirm that your field software is applying the appropriate geoid model for your region, and label every dataset you deliver with the vertical datum explicitly. Never assume the receiving party will know to ask.
Level Bubble Calibration
A pole bubble that’s out of adjustment introduces tilt error into every measurement without triggering any warning in the receiver. The instrument has no way of knowing the pole isn’t plumb, it trusts the operator. Bubbles drift out of calibration through vibration, thermal cycling, and impact, and they rarely drift far enough to be obvious without a deliberate check.
The field verification takes under a minute: hold the pole against a vertical surface, centre the bubble, rotate 180 degrees, and check whether it stays centred. If it moves, the bubble needs adjustment before the session continues. Check it regularly in the office and verify it any time the pole has been dropped or transported roughly.
Setup Documentation That Holds Up Later
The failure to document setups adequately is the mistake that causes the most damage long after the fieldwork is done. When a project is revisited, extended, or disputed, the field notes are often the only record of what actually happened on site. Missing antenna heights, undocumented coordinate systems, ambiguous mark descriptions, and unrecorded correction sources are gaps that become unresolvable problems when someone needs to reconstruct the session months later.
Every setup should record the following before observations begin:
- Instrument model and serial number
- Antenna height, the reference point measured to, and the value entered in the controller
- Coordinate system, datum, and geoid transformation applied
- Correction source, CORS station or local base, with approximate baseline distance
- Fix status and any notable signal conditions during the session
At Bench-Mark, the approach we recommend to customers is straightforward: treat documentation with the same discipline as the measurement itself. The coordinates are what you deliver. The notes are what you defend them with.
Preparation Is the Only Reliable Error Management
Every mistake covered here shares a common characteristic, none of them require unusual circumstances. They happen on routine jobs with functioning equipment operated by competent people. What they require is a skipped step, an unchecked assumption, or a setting that carried over incorrectly from a previous session. The fix for all of them is the same: methodical setup verification before the first point, not diagnostic work after the last one.
