The job RTK can't do
On most of our road and earthworks projects, RTK GNSS does the heavy lifting — it's fast, it's accurate enough horizontally, and one rover covers ground a level crew never could. But every season there's a moment where a project manager asks for a height we'd be embarrassed to give them from a GNSS receiver: the deflection of a bridge bearing, the cant on a rail alignment, the settlement of a tank foundation across six monthly visits. That is when the digital level comes off the rack.
The reason is simple and physical. GNSS height is the weakest component of a satellite fix — it's the dimension most polluted by atmosphere, multipath, and geometry. A good RTK vertical sits in the ±15–25 mm band (typical/illustrative). A digital level running a tied loop sits in the ±0.3–1 mm per kilometre band (typical/illustrative). For deformation and precise control, that order-of-magnitude gap is the whole game.
The vertical control behind our numbers
- 800,000+
- feddans levelled
- Vertical control delivered across agricultural and land projects
- 3,000+
- km of roads
- Grade and profile control on highway corridors
- 600+
- km of railways
- Where cant and level tolerance are unforgiving
What a benchmark network actually is
A benchmark is a permanent, stable mark whose height is known relative to a national or project datum. A benchmark network is a set of those marks tied together by levelling so that any one of them can be recovered and trusted years later. The discipline isn't the instrument — it's the closure. We never accept a project height that hangs off a single mark. Every levelling run starts on one known benchmark and must close back onto another (or back onto itself), and the misclosure tells us whether the work is good.
In practice we build a small local network at the start of a long project: two or three protected marks, cast or driven where machinery won't touch them, levelled into the national framework and into each other. Everything downstream — staking, as-builts, monitoring epochs — references those marks. If the network is honest, a height we shoot today and a height we shoot in eight months are directly comparable, and that comparability is the entire point of monitoring.
Choosing the height tool for the job
| Criterion | RTK GNSS | Total Station | Digital Level |
|---|---|---|---|
| Typical vertical accuracy | ±15–25 mm | ±2–5 mm | ±0.3–1 mm/km |
| Speed over open ground | Fastest | Moderate | Slowest |
| Works without line of sight to marks | Yes | No | No |
| Right for deformation / rail / structural | No | Sometimes | Yes |
| Removes operator reading blunders | Yes | Partly | Yes (bar-coded staff) |
Vertical accuracy figures are typical/illustrative ranges for planning, not instrument guarantees; field results depend on procedure and conditions.
How we run a precise levelling loop
- 1
Field-test the instrument first: run a two-peg / collimation check so we know the line of sight is true before any data is logged.
- 2
Set out from a known benchmark, and plan the run so it closes back onto a second known mark — never a dangling end.
- 3
Keep backsight and foresight distances balanced at each setup; equal sight lengths cancel collimation and most refraction error.
- 4
Read forward through the loop, logging staff and distance to the data collector, watching the running distance balance.
- 5
Run the return leg back to the start (or onward to the second benchmark) to give an independent check and cancel systematic error.
- 6
Compute the misclosure against the allowable tolerance for the loop length; if it fails, re-run the suspect section rather than adjusting blindly.
- 7
Adjust and distribute the accepted misclosure, then publish the heights with the closure figure attached so the next crew can trust them.
Vertical accuracy by method (typical/illustrative)
Test before you trust
Per ISO 17123, a level's performance should be verified by a defined field procedure before precise work, not assumed from the spec sheet. We treat the two-peg collimation check as non-negotiable at the start of every campaign and after any rough transport — a few tenths of a millimetre of collimation error accumulates silently over a long loop and is invisible until your closure fails.
Stability starts at the tripod
The levelling instruments in our fleet
Levels
Automatic and digital levels for high-precision elevation and benchmark networks.
automatic & digital levels
Automatic and digital levels from our 90-instrument fleet, kept field-tested and ready for precise loops.
Where this fits in our work
References
- ISO 17123 series — Field procedures for testing geodetic and surveying instruments — International Organization for Standardization (ISO)
- Guidelines for Real-Time Kinematic (RTK) GNSS surveying and geodetic control — US National Geodetic Survey (NGS/NOAA)