Roof leak detection answers two different questions, and most wasted money on a leaking roof comes from confusing them. The first question is where water is entering the roof. The second is how far it has already spread inside the assembly. A method that answers one of those rarely answers the other, and a survey that mixes them up produces a report nobody can price work against.
This guide sets out the methods in normal use on flat and low slope commercial roofs, how each one actually works, and, more usefully, what each one can and cannot establish. Our guide to finding and fixing roof leaks covers what to do while the water is still coming in. This one is about how the search is done.
Entry point or extent
An entry point is a defect: an open seam, a split boot, a failed termination, a puncture. It is small, it is usually at a detail rather than in the open field, and it is found by looking in the right place.
Extent is the wet area inside the roof. It is always larger than the defect, it spreads sideways through the insulation, and it is invisible from the surface. It cannot be found by looking. It has to be measured.
You need both before anyone quotes. The entry point tells you what to fix. The extent tells you how much material has to come out, which is the line that moves a repair price more than any other.
Visual inspection, working uphill
The search starts at the stain and moves against the fall, checking every detail on the way: the nearest penetrations, the seams above them, the flashing at any wall or curb upslope, the drain surround, the edge.
What it establishes. A candidate entry point, and usually the right one. Most leaks on most roofs are found this way, by somebody who knows that laps are probed by hand rather than looked at, and that a seam which is sound to the eye can be open to a blunt probe.
What it cannot establish. How far the water has traveled, whether the insulation is wet, or whether the defect found is the only one. A roof that has one open seam frequently has several, and the one found first is simply the one nearest the ceiling stain.
Moisture meters
Two instruments get called a moisture meter and they are not the same.
A capacitance or impedance meter is placed on the surface and reads the electrical properties of what sits beneath it. It does not measure water directly. It reports a relative value, and the value only means something compared with readings taken on the same roof, on the same assembly, on the same day. Readings are taken on a grid, and the map is built from the pattern rather than from any single number.
A pin or probe meter takes a reading through a small hole, which makes it invasive and makes it specific.
What they establish. Where one part of a roof reads consistently wetter than the rest of it. On a uniform assembly that is a reliable map of the extent.
What they cannot establish. The entry point, the depth at which the water sits, or an absolute wetness. They are defeated by ballast, gravel, a wet surface, a thick or reinforced coating, or a roof built from several different assemblies, because then the comparison that makes the readings meaningful no longer holds.
Infrared thermography
A wet area of insulation holds heat longer than a dry one. A thermal camera scanned across the roof after sunset, on a clear evening following a sunny day, shows those areas as warm anomalies while the rest of the roof has already cooled.
What it establishes. A shape. It is the fastest way to cover a large roof and the best way to produce a map that a repair can be drawn on.
What it cannot establish. That the anomaly is water. Heat is held by other things: a change of insulation, an area of different thickness, a patch, debris, a surface repair, even a section walked on that afternoon. Infrared finds anomalies, and anomalies are confirmed with a meter or a core, not assumed. It also needs conditions it cannot be given on demand, so it is the method most often scheduled and then rescheduled.
Electronic leak detection
These methods pass a low or high voltage across the membrane and look for the path that current takes through a breach. They need a membrane that does not conduct over a deck or a layer that does.
What it establishes. Breaches in the membrane itself, precisely, and it can cover a whole roof systematically rather than only the area somebody suspects. On a new roof it is used to test the installation before anything is loaded on top of it.
What it cannot establish. Anything above the membrane. Water entering over a coping, through a wall above the upstand, around a window or from a failed condensate line is not a breach in the roof, and this will not find it. It also does not work on every assembly, and a wet surface, a conductive membrane or a ballasted roof can rule it out before it starts.
Water testing
One area is isolated and flooded at a time, starting low and moving uphill, with somebody watching inside. Done properly it is the most conclusive method there is, because the answer is the water coming through while a known area is wet.
What it establishes. Which specific detail leaks. Not a candidate, not an anomaly: the one that failed.
What it cannot establish. Extent, and nothing at all if it is done badly. Hosing a whole roof at once wets everything and proves nothing. It is also the method with the most ways to go wrong: water into a building that was dry, drains overloaded, a test area not isolated, weight of water on a deck of unknown condition. It should be the step taken when the cheaper methods have run out, not the first move.
Core cuts and test cuts
A small opening is cut through the assembly, the layers are examined and photographed, the material is checked for moisture, and the hole is closed permanently the same day.
What it establishes. What the roof is actually made of, in what order, and whether it is wet at that point. It is the only method that shows you the assembly rather than inferring it, which matters because the drawings are often wrong and roofs get covered over.
What it cannot establish. Anything beyond the immediate area around the hole. Cores are confirmation, not search. Their number should be small, their positions should be chosen by the survey that came before them, and each one should be recorded and made good.
Sequencing, and what no method will tell you
The cheap order is the right order. Look first, measure second, confirm third, and only flood the roof when the first three have not settled it.
| Step | Method | Question it answers |
|---|---|---|
| One | Visual, uphill from the stain | Where might it be entering |
| Two | Meter grid or infrared scan | How far has the water spread |
| Three | Cores in the marked areas | Is the material actually wet, and what is it |
| Four | Water test or electronic detection | Which detail is it, exactly |
No method on that list tells you how long the roof has been wet, who caused it, or whether an insurer will treat it as sudden damage or as wear. Those are questions answered by records, which is the argument for dated photographs and a maintenance history rather than for a better instrument.
A survey should come back with the methods used, the conditions on the day, the wet areas marked on a roof plan, the cores photographed and located, and a plain statement of what was established and what is still uncertain. A report that hides the uncertainty is less useful than one that names it, because the next decision is what to open up, and that decision needs to know what is a fact and what is a shape on a screen.
Water coming in now? Start with roof leaks. Deciding what to do once it is traced? Read repair or replace.
Common questions
What is the difference between an infrared scan and a moisture survey?
An infrared scan is one way of doing a moisture survey. A moisture survey is the job: mapping where water has got into the roof assembly and how far it has spread.
- Infrared reads surface temperature, not water
- A capacitance meter reads electrical properties at a point, not an image
- Core cuts show the actual layers and whether the material is wet
- A survey normally uses one to find suspects and another to confirm them
Ask which methods were used, in what order, and how the findings were verified. A survey that scanned and never confirmed anything has produced a map of anomalies rather than a map of wet insulation, and the two are not the same thing to price a repair against.
Can a leak be found without cutting into the roof?
Often, yes. Most leaks are found by a visual search working uphill from the stain, and several of the instrument methods are entirely non destructive.
- Visual inspection, infrared, meters and electronic detection all leave the roof closed
- A water test wets the roof but does not open it
- What no non destructive method can do is confirm what the layers are
- Only a small core shows the actual assembly and whether it is wet
A reasonable sequence is to search without cutting, then cut a small number of cores in the places the search pointed at, and close them properly the same day. Cutting first and searching afterwards is the version to refuse, because it makes holes in a roof on the strength of a guess.
Why did the survey find nothing when the roof clearly leaks?
Usually because the leak needs a condition that was not present on the day. A roof only leaks when water reaches the defect, and several methods need the roof to be dry to work at all.
- Driven rain from one direction enters details that are sound in still weather
- A backup leak needs enough rain to overload the drainage first
- Infrared needs a clear evening after a sunny day, and wet weather defeats it
- A meter cannot read through ballast, gravel or a soaked surface
Tell the surveyor what the weather was doing every time it leaked, including the wind direction and how long the rain had been falling. An intermittent leak is not a small leak. It is one that needs a particular condition, and the survey has to reproduce that condition or wait for it.
Is a drone survey any use for finding leaks?
It is useful for seeing a roof you cannot safely walk and for recording its layout. It is not a substitute for a survey, because the things that leak are not visible from above.
- Open seams, soft spots and lifted edges are found by hand and by foot
- An aerial photograph shows standing water, debris and obvious storm damage
- A thermal camera flown over a roof produces anomalies that still need checking
- Nothing flown can tell you whether the insulation under the membrane is wet
Treat aerial images as a map to plan the walk with, not as the walk. The one thing they do very well is give you a dated overhead record of the whole roof, which is worth having before storm season and worth repeating after it.
Who should be in the building during a water test?
Someone who knows the interior and can watch the suspected area continuously while the crew works outside, in contact with them the whole time.
- The observer reports the moment water appears, not afterwards
- Ceiling tiles come out first so the deck above can be seen
- Stock, equipment and electrics under the test area are covered or moved
- The test stops the moment water shows, so the area being flooded is the answer
Agree before it starts who can call a halt and what happens if water appears somewhere unexpected. A test with nobody watching inside tells you only that the roof still leaks, which is what you knew, and it has put water into the building to prove it.