The deliverable is a list of candidate locations along the surveyed corridor, each one positioned, described, and graded by how much confidence the evidence supports. That grading is the point of the exercise, because a survey that returns thirty equally weighted anomalies has not actually narrowed anything.

Grading is done against stated temperature thresholds, fixed before the analysis begins. Rather than comparing everything with the road surface, the reference is built around the pipe itself: a local baseline is established on a stable section of the buried run in each zone of the corridor, because pipe temperature changes along a route as hydraulic load changes. An anomaly must exceed that local baseline by 0.3°C to be entered for review at all, and by 0.5°C to be classified as a confirmed candidate. Because the rule is fixed and published, a strong candidate on your survey means the same thing as a strong candidate on anybody else’s, and the classification can be checked by a third party rather than taken on trust.

Passing the temperature test is only the first stage. Every flagged location is then assessed against a further set of criteria in a defined order: the magnitude of the departure, the shape of the anomaly across the pipe and along it, whether the warm zone has broadened beyond the normal pipe corridor, whether it runs downslope in the way escaping water would, what the visible imagery shows at the same spot, and what infrastructure is known to be there. A candidate supported by three or more of those independently is a different proposition from one that is merely warm.

Corroboration in the visible imagery raises confidence further. Cracking in a road surface, ground that is damp or unusually green, or vegetation growing where it should not, all strengthen a thermal finding. So does the reverse: on one corridor, road reinstatement patches were measured and found to read slightly cooler than the aged asphalt around them, which eliminated the patch material itself as a heat source and meant any warmth over a patch required an explanation from underneath it. That is the kind of reasoning a Level 3 analysis is for, and it is what separates a defensible candidate list from a list of warm spots.

This is not the right tool for every leak. On a short run in an accessible location, acoustic survey and correlation are quicker, cheaper and more precise, and any competent leak detection contractor will find it faster than a drone will. The aerial survey earns its place in a narrower set of situations, and in one of them it has very little competition at all.

District heating and hot water networks are where it works best by a considerable margin. The escaping fluid is hot, so the contrast at the surface is large and unambiguous rather than marginal, and it does not depend on the subtle differences in thermal behaviour a cold water leak relies on. On a campus, hospital, housing scheme or industrial site network, one visit covers the whole distribution layout.

Long runs across open ground are the second case. Where a main crosses fields, verges or estate land for a kilometre or more, walking it with acoustic equipment is slow and repeated speculative excavation is expensive. Flying the corridor and returning a handful of ranked candidates changes the economics of the investigation rather than just the method. The third case is ground you cannot open speculatively: a live yard, a service road, a car park, a listed setting, or anywhere reinstatement costs more than the excavation.

The fourth is international, and it is the one this service was actually built for. The protocol behind these surveys was developed and proven overseas, in arid volcanic ground that drains away the very moisture the method depends on, which is close to the hardest condition thermographic leak detection faces. It was written from the outset to travel: the document fixes the altitudes, the radiometric parameters, the environmental limits at which capture is postponed and the two survey windows, so that a trained local operator can capture to it in their own territory while the certified Level 3 analysis and reporting is carried out against their dataset. The practical consequence is that a survey can be delivered in a country without flying a specialist to the site, and the cost of working in a new territory is a trained operator rather than an air fare.

That matters most where water is scarce and losses are political rather than merely expensive. Across southern Europe, the Atlantic islands and much of Africa, distribution networks run through exactly the difficult ground this method was calibrated against, and the case for finding losses without excavating is correspondingly stronger. If you are a network operator, a utility or an aid or development programme working in those conditions, the protocol already exists and has been tested against a known answer.