GPR utility survey accuracy: depth, uncertainty, PAS 128 and ASCE 38

What GPR really measures, where the uncertainty comes from, and how to back up the accuracy you report under PAS 128 in the UK and ASCE 38 quality levels in the US.

The standards, in short

In the UK, PAS 128 sets out how underground utility surveys are specified and delivered, and ties every utility on the plan to a quality level and a stated positional accuracy. Detection with geophysical techniques such as GPR and electromagnetic location sits between desktop records and verification by trial holes.

In the US, ASCE 38 defines four quality levels: QL-D from existing records, QL-C adding surface features, QL-B where utilities are designated with geophysical methods such as GPR, and QL-A where they are exposed by test holes or potholing. Most GPR work on a utility survey is QL-B.

Under both, an accuracy figure is only worth something if it can be justified, utility by utility.

Where plan-position uncertainty comes from

Antenna positioning: a fixed RTK GNSS solution gives a few centimetres; a wheel-based survey without GNSS depends on setting out and on how straight the lines are.
Picking the hyperbola apex: a few centimetres on a clean echo, more on a weak or noisy one.
Line spacing: between two lines the utility position is interpolated, and a bend or a branch can fall between them.
Utilities crossed at an angle, if their direction is not taken into account.

A simple, defensible way to quantify uncertainty: measure how far the detected points sit from the fitted axis of the utility, line by line. A utility whose detections line up within 5 cm is easy to stand behind; one whose points scatter by 40 cm should be reported as such.

Depth: the weak spot when the velocity is wrong

Depth depends directly on the wave velocity in the ground. A 20 % velocity error is a 20 % depth error: 20 cm on a utility at 1 m. The velocity should be estimated from the site's own hyperbolas and, where possible, confirmed at a trial hole.

GPR measures the top of the utility, not its centre line: on a 200 mm duct that is a 10 cm difference. The deliverable should say which depth is reported.

What a solid deliverable contains

A plan in the project's coordinate system (in Great Britain, British National Grid; in the US, the state or project datum, often NAD83).
For each utility: depth to top, plan-position uncertainty, confidence of the detection.
The ground velocity used and how it was obtained.
Areas where the survey was inconclusive, flagged as such: a utility not detected is not a utility that is not there.

How Radar Plan 3D helps

For each utility Radar Plan 3D gives the depth to top at every point, a plan-position uncertainty computed from the detections, and a confidence level (confirmed, probable, to verify) based on the probability that its alignment is down to chance. The ground velocity estimated from the data is in the report, next to the one entered by the operator.

These figures support the quality level you report; they do not replace the surveyor's judgement, nor verification by trial holes or potholing where it is needed.

Updated 23 Oct 2026