What Did the Thermal Survey Actually Find?

At the confirmed breach the method worked, and it worked clearly. The low-altitude pre-dawn dataset showed a well-defined localised warm anomaly at the breach location, exceeding background surface temperature by approximately two to three degrees under near-equilibrium conditions. The anomaly was spatially coherent, it appeared in multiple independent frames, and its shape followed the downslope direction in which surface water was visibly running. Line profiles taken across it showed a peak against a stable surrounding baseline rather than a general elevation of the whole road surface, which is the distinction that separates a discrete source from a warm section of carriageway.

The more valuable finding was what the survey correctly rejected. Linear features crossing the carriageway, consistent with buried service ducts or trench backfill, produced their own repeatable thermal separation from the surrounding asphalt of roughly 0.7 to 0.8 degrees. They were measurable, consistent and entirely real, and they were not leaks. They were classified as infrastructure because they showed no plume and no migration, presenting instead as a straight bounded line consistent with a construction feature. Any method that flags those as candidates will drown a network operator in false positives, and the ability to set them aside on stated evidence is what makes the rest of the findings worth reading.

Adjacent open volcanic ground, with sparse vegetation and exposed lapilli, produced no pattern that could confidently be attributed to leakage. Rather than report that as a clear result, it was recorded as non-diagnostic for this application, because uneven natural ground of this kind carries too many competing thermal explanations to support a confident negative.

The Real Deliverable Was the Protocol, Not the Leak

The commissioning authority already knew where that leak was. What it did not have, and what the exercise was actually built to produce, was a written, proprietary survey protocol calibrated to this class of ground and climate, with its performance characterised and its failure modes documented.

That document fixes everything a repeatable survey depends on and leaves nothing to the judgement of whoever happens to be holding the controller: flight altitudes and the ground sample distance each one yields, overlap and gimbal angle, the emissivity and reflected apparent temperature to be applied and the method for establishing them on site, the palette and thermal span used during triage, the environmental limits at which capture is postponed, and the two-time-of-day strategy with the reasoning behind each window. The altitude comparison went into it as a finding: low altitude is the analytical dataset, medium altitude is context only, and the highest altitude was reported as not diagnostic for leak detection, which is exactly the sort of conclusion a proof of concept exists to produce.

It was written from the outset to be transportable. The structure separates data collection from interpretation, so a competent local operator captures to the protocol in their own territory while the certified Level 3 analysis and reporting is carried out against that dataset. The value of that split is practical rather than theoretical: it means a survey can be delivered in a country the thermographer has never stood in, without the certification integrity of the report depending on their physical presence, and it means the marginal cost of extending into a new territory is a trained operator rather than a travelling specialist.

The territories it was designed for are the ones where the conditions are hardest and the problem is largest. Arid and volcanic ground of this kind is common across southern Europe, the Atlantic islands and much of Africa, and those are also the regions where water scarcity makes losses from a distribution network economically and politically significant rather than merely inconvenient. A protocol proven in high-porosity volcanic soil under a strong solar load transfers to those conditions far better than anything written for a temperate climate. Adoption in further European and African markets is the intended path for this work, and the calibration exercise described here is what makes that a credible proposition rather than an ambition.

What the exercise established, in practical terms:

• That a leak at this depth in this ground produces a surface signature the method can resolve, and roughly how large that signature is.
• That low altitude is not optional, and that coverage gained by flying higher costs more than it returns.
• That the pre-dawn window carries the cleanest signal, with the afternoon pass valuable as a cross-check rather than a substitute.
• That buried services will be seen, will look convincing, and must be discriminated on shape and behaviour rather than temperature alone.
• That open, vegetated and uneven ground should be scoped out at the enquiry stage rather than surveyed and then caveated.
• That the conditions rules have to be enforced rather than noted, because in this climate the ground itself will produce a convincing false result.

Why Test a Method on a Leak You Have Already Found?

It is a fair question, and the answer is the whole point of the exercise. A survey flown over unknown ground that returns nothing tells you either that there is no leak or that the method cannot see it, and there is no way to tell those apart. Flown over a leak that is already confirmed, a null result would have been decisive, and a positive result establishes the size and shape of the signature to look for everywhere else. Characterising an instrument against a known answer before trusting it on an unknown one is ordinary measurement discipline, and on a technique as easy to over-read as thermography it is the difference between a screening tool and an expensive source of false confidence. The protocol written here has since governed leak detection work on ground where nothing had been confirmed in advance, which is the only real test of whether the exercise was worth carrying out.