Thermal vs Visual Imagery on a Roof Inspection

A building owner asking for a drone roof inspection is usually offered two things — a visual survey and a thermal survey — and often assumes the thermal one is the better version of the visual one. It is not. They are different instruments answering different questions, and the most common way to waste money on a roof inspection is to buy the wrong one for the question you actually have.

Here is what each sensor genuinely shows on a commercial roof, and how to tell which you need.

What the visual camera shows

A high-resolution visual survey documents the condition of the roof surface. Flown properly — a nadir mapping pass for the overall record, plus close-range obliques on anything of interest — it gives you:

• Membrane condition: splits, tears, punctures, open laps and seams, blistering, ridging, surface degradation.
• Flashings, terminations and penetrations — the details where most roofs actually fail.
• Ponding water and the low areas where it collects, which is a drainage finding as much as a membrane one.
• Debris, gravel displacement, and mechanical damage from foot traffic or servicing.
• Rooftop equipment: units, curbs, condensate lines, leaking or corroded fixtures, and anything that has been added since the roof was built.
• Hail evidence — spatter marks, granule loss, bruising and impact damage. In Alberta this is a routine reason for the flight rather than an incidental finding.
• A dated, geo-referenced record of the whole roof, which is worth having independently of any defect found.

If the visual pass is flown photogrammetrically, you also get a measurable orthomosaic of the roof — useful for quantifying areas, laying out a repair scope, and comparing against next year's flight. Our post on how to read an orthomosaic covers what that product can and cannot tell you.

What the visual camera cannot do is see through the membrane. A roof can photograph well and still be soaked underneath.

What the thermal camera shows — and what it does not

The first thing to be clear about: a thermal camera does not detect water. It measures the infrared energy leaving a surface, which it renders as an apparent temperature. That is all it does.

A roof moisture survey works because of what wet insulation does with heat. Water has a high heat capacity, so insulation saturated with water absorbs more of the day's solar energy and releases it more slowly than the dry insulation around it. During the evening, as the roof cools, the wet areas stay warmer for longer. Flown in that cool-down window, those areas show as warm anomalies against a cooler background, usually in irregular patches that spread out from a defect in the membrane.

Everything about that mechanism implies conditions:

• The roof needs to have been loaded with heat — a reasonably sunny day beforehand.
• The surface must be dry. Rain, snow, dew or standing water on the membrane masks what is underneath completely.
• Wind steals the signal. Convective cooling evens out the differences the survey depends on.
• The scan happens during the cool-down, not in the middle of the day when solar loading swamps everything.

A thermal roof scan flown in the wrong conditions is not a weaker result. It is an uninterpretable one, and a clean-looking scan flown badly is worse than no scan at all because it gets filed as evidence of a dry roof. Our post on the conditions that make or break a thermal inspection goes through this in detail.

There are also roofs where the technique simply does not apply well — assemblies where the insulation is not positioned to hold and reveal moisture in this way, ballasted and protected-membrane roofs where the thermal signal is buried, and some metal systems. An operator who tells you a thermal scan is not the right tool for your roof is doing the job properly.

Why radiometric is the part that matters

This is the distinction that separates a usable thermal deliverable from a set of orange-and-purple pictures, and it is rarely explained to buyers.

Non-radiometric thermal imagery is a colourised image. The camera maps the temperature range it currently sees onto a palette, and it rescales as it moves — so the same anomaly can look dramatic in one frame and invisible in the next, and there is no temperature value stored anywhere. You cannot measure on it, you cannot set a consistent scale across the roof, and you cannot re-examine it later. It looks like a thermal inspection and it is not one.

Radiometric imagery stores a calibrated temperature value for every pixel. That single difference gives you:

• Measurable differences. An anomaly can be quantified against the surrounding roof rather than described as "warmer".
• Consistent scaling. The whole survey can be rendered on one temperature scale, so a mild anomaly on the north end and a severe one on the south are comparable on the same image.
• Re-analysis. The data can be revisited months later with a different scale or threshold without re-flying.
• Defensible findings. When the survey goes into a capital plan, a warranty conversation or an insurance file, "this area was measurably warmer than the surrounding roof, at this scale, on this date" is a statement that survives scrutiny. A screenshot is not.

Ask directly whether the deliverable is radiometric, and whether you receive the radiometric files or only rendered images. It is the single most useful question a roof buyer can ask, and the same distinction drives our solar thermal inspection work, where every finding has to be severity-ranked against measured values rather than eyeballed.

A thermal anomaly is a candidate, not a verdict

Thermal imagery locates areas that behave like wet insulation. Other things also behave like wet insulation: mechanical equipment and its heat, recent foot traffic, differences in the roof build-up, patches and repairs, shading, standing debris, and a stray blanket of cloud passing during the scan.

So the anomaly is where you go and check. The standard sequence is: thermal survey identifies and maps candidate areas, then those areas are verified on the roof — a moisture meter reading, a test cut or a core sample — and only then does a repair scope get written. Any operator or contractor who moves straight from a thermal image to a tear-off quantity is skipping the step that makes the number real.

A related caution: thermal does not locate leaks. It maps where moisture has accumulated in the assembly. Water enters at one point and travels, sometimes a long way, so the wet area and the entry point are frequently not in the same place. That is exactly why the visual survey is not optional — the entry point is usually a visible defect.

When to buy both — which is most of the time

The two datasets are strongest together, and it costs very little more to capture both on one mobilisation than to capture either alone. The pairing works like this:

• Both flag the same area — a visible split with a wet zone spreading from it. That is a high-confidence finding with a cause and an extent.
• Visual only — surface damage with no moisture behind it yet. This is the best possible finding, because it is the cheap repair made before it becomes an expensive one.
• Thermal only — moisture with no obvious surface defect above it. That points at a detail, a penetration or a termination somewhere else, and it is where the close-range imagery earns its place.
• Neither — genuine, documented reassurance, provided the thermal was flown in valid conditions.

Cases for a visual survey alone: post-storm and post-hail documentation, an insurance or warranty claim needing a dated record of physical damage, routine annual condition monitoring, and any roof where the assembly makes a moisture scan unreliable. Aerial evidence for claims is its own workflow — see our insurance and loss adjusting page.

Cases for thermal in the mix: any roof approaching a repair-or-replace decision, a roof with a leak nobody can locate, verifying a recent repair or a new installation, and scoping a replacement where knowing how much of the insulation is wet changes the budget substantially.

What the deliverable should contain

Whichever you buy, the report should have:

• A roof plan or orthomosaic that every finding is located on, so a repair crew can find it.
• Findings individually numbered, located and described — not a gallery of images with a summary paragraph.
• For thermal: the capture date and time, the conditions, the temperature scale used, and a statement of the verification status of each anomaly.
• An explicit statement of the survey's limits — areas not covered, conditions that constrained the work, and anything the method cannot determine.
• The underlying data, not only the PDF, so next year's survey can be compared against this one.

A report that will not state its own limits is telling you something. Our post on how to vet a commercial drone contractor in Alberta covers what else to ask for before the flight is booked.

The short version

Visual tells you what the roof looks like. Thermal tells you where heat is behaving oddly, which — flown in the right conditions, captured radiometrically, and verified on the roof — tells you where the moisture is. Buy the visual survey always. Add thermal when a decision depends on what is underneath, and insist that it is radiometric and flown in a window that makes the result mean something.