How a GPR survey works on a UK construction plot
Ground penetrating radar is a geophysical method, not a magic camera. An antenna transmits a short electromagnetic pulse. Interfaces in the ground — the crown of a pipe, a void, a change from made ground to natural clay — reflect energy back to a receiver. The surveyor then interprets those reflections against a grid, a site reconnaissance and the desktop utility study that should already exist. That interpretation is what becomes a quality-levelled line on a PAS 128 drawing.
In the United Kingdom the method earns its keep because so much buried apparatus is non-metallic. Plastic water mains, polyethylene gas insertions, fibre ducts and abandoned brick culverts do not announce themselves to an electromagnetic locator. PAS 128:2022 therefore treats GPR as a required companion to EML on Type B work, not an optional extra for “difficult” sites. The CICES survey-types article states that pairing in plain language.
Grid spacing is a specification choice. A wide transect on an open field may be enough to confirm that a recorded crossing is where the statutory plan said it was. A city-centre courtyard with a reinforced slab usually needs a dense grid and later processing. Clients who ask for “a quick GPR sweep” are often asking for a methodology type that cannot support the quality level they will later claim in a design report.
Limits you should expect us to write down
Radar does not like electrically conductive ground. Heavy clay after rain, saline fill, and some industrial made-ground will shorten the useful depth. Steel mesh in a slab creates a cluttered first metre. Those are not reasons to skip GPR; they are reasons to say so in the residual-risk note and to consider a trial hole on the crossings that actually control the design.
Depth estimates from GPR are interpretations. They depend on an assumed velocity in the ground. A PAS 128 quality level that quotes a vertical accuracy is making a statement about that interpretation, not promising a surveyed invert as if the pipe had been exposed. If you need a measured depth for a critical crossing, Type A is the honest next step.
UK places where radar is doing real work
Thames Tideway and the later connections around it are a fair example of mixed materials: large concrete structures, older metallic plant, and newer plastic connections in the same streets. A GPR survey there is not a substitute for the statutory record; it is how you look for the things the record never held. Birmingham city-centre regeneration — tight plots, podium slabs, and services diverted around successive retail and civic schemes — is another environment where processed radar is the difference between a blank square on the drawing and a qualified unknown.
We name those places as application settings. We do not claim we surveyed them. The same methods apply on a Leeds campus spine, a Glasgow waterfront plot, or a highway widening in the Midlands. The specification does not change because the postcode does.
From radar screen to a drawing someone can use
A useful GPR product is a set of interpreted features in the same coordinate system as the topographic survey, attributed with a quality level and a note on confidence. Pretty hyperbolas in a slide deck do not help an excavator. That is why we treat GPR as an input to utility mapping rather than a standalone performance.
If you already hold a Type D plot, send it with the enquiry. If you do not, start there. Then tell us whether the working area is a slab, a carriageway or a soft verge. Those three sentences decide more about cost and quality than a request for “the most accurate GPR in the UK”. Read the pairing explainer on why Type B needs EML and GPR before you strip one method out of a specification to save a day.
Grid, antennas and the temptation of a single transect
A single GPR transect across a proposed trench can be a useful reconnaissance. It is not a Type B survey of a working area. PAS 128 methodology types exist because the distance between lines changes what you can defensibly claim. On an open field you may accept a wider spacing and still support a modest quality level. On a Birmingham city-centre slab, or around a Thames Tideway connection, a wide spacing is how unknown ducts hide between the lines and then appear in a trench photograph nobody wanted.
Antenna choice is another specification fact, not a brand preference. Higher frequencies resolve shallow detail and die earlier. Lower frequencies look deeper and smear the first metre. Dual or multiple antennas, and later processing, are how urban plots stay honest. We will say which antennas we intend to use. We will not pretend a single real-time screen is a processed cube.
Traffic management belongs in the same paragraph as the grid. If you can only occupy a lane for four hours, the grid will be incomplete unless you buy another shift. It is better to write that in the proposal than to issue a drawing that silently omits a bay. Residual risk is a feature of PAS 128, not an apology.
Interpretation should be done by someone who has also read the Type D plot and walked the Type C evidence. Hyperbolas that “look like a pipe” next to a recorded water main are one story. Hyperbolas in a blank square with no owner are another. Both can be mapped. Only one should be labelled as if the statutory return had already proved it.
When the radar is finished, the product is still a PAS 128 segment in a coordinated model. If you only want pictures of screens, say so — and accept that those pictures will not help an excavator at 06:30 on a wet Tuesday in Leeds.