Infrared thermography: a heat picture speaks only when conditions are right

Everything with a temperature emits infrared. Most drone thermal cameras sense long-wave infrared (LWIR), roughly 8–14 µm, with an uncooled microbolometer. The camera receives radiant energy, not temperature; temperature is a conversion.

To get a correct temperature you must give the camera at least emissivity, reflected temperature, and distance and humidity. In inspection, what usually matters more than the absolute value is the difference from the surroundings (ΔT). Comparing surfaces of the same material and orientation cancels much of the emissivity and reflection error.

Emissivity also depends on viewing angle: the more grazing the view, the lower the emissivity and the stronger the reflection. A facade shot from a rooftop or a panel imaged obliquely may not be comparable with a square-on shot. Also, a clear sky is very cold in the infrared; when glass or metal reflects it, the surface looks colder than it is.

Half the success of thermography is timing. Thermal contrast appears only while heat is flowing: when a facade is heated by the sun or cooling down, when there is an indoor–outdoor difference for insulation, when current flows in electrical equipment.

• Time by orientation: each face gets sun at different hours — east in the morning, south around midday, west in the afternoon. North faces get little sun and show weak contrast.
• Weather: rain the day before, strong wind or weak overcast sun all shrink the contrast.
• Season: for missing insulation, fly on winter nights or early mornings with a large indoor–outdoor difference and no residual solar effect.
• Load: electrical faults heat more at higher current. Declaring “no anomaly” at light load is risky.

For facade surveys, planning infrared facade capture around sunlight, reflection and quality checks (Airffic Flight Institute) sets out the steps.

Under Japan’s periodic building reporting, full-surface hammer sounding of facade tiles is required in some cases. A January 2022 amendment of the relevant notice explicitly allowed infrared surveys with accuracy equal to or better than sounding, including those by drone. The condition is that the survey actually achieves that accuracy. See condominiums for how this is used.

For electrical equipment the basic method is ΔT between phases under the same conditions, not absolute temperature. If one of three phases has a hot joint, that is the clue. Electromagnetic fields near lines are strong, so the aircraft needs precautions too. Strong electromagnetic fields

Thermal cameras are either radiometric, storing temperature data per pixel, or non-radiometric, storing only a coloured picture. For inspection records, radiometric is the only real choice.

Thermal sensors have few pixels (640×512 is common). With a 12 µm pitch, 640 pixels and 13.5 mm focal length, GSD = sensor width × distance ÷ (focal length × pixels) gives about 2.7 cm at 30 m. A rule often quoted is to put several pixels (around 3×3) on anything you want to measure; a small joint imaged from far away is averaged with its surroundings and reads low. The GSD logic is the same as in the visible imaging chapter.

• Through glass or polished metal: LWIR does not pass through glass; you cannot see indoors through windows. Polished metal barely reads.
• Depth: only surface temperature is visible. Deep delaminations and internal voids barely change it.
• Bad conditions: overcast, after rain, strong wind or during the daily crossover, anomalies may not show at all.
• Root cause: you learn that something is hot, not why. Hotspot causes are confirmed with IV-curve tests or on-site checks.
• Proof of absence: not seeing an anomaly does not prove there is none.

By 2035, drones will likely pick the best time to fly from weather forecasts and solar geometry — south faces at midday, west faces in the afternoon — while emissivity and reflected temperature are assigned per surface from visible images and the 3D model. Thermal images line up per member and per year, and only changes in ΔT reach the inspector. This is the editors’ view, but the direction is clear.

The physics will not change: when heat is not flowing, nothing shows. Record the conditions and fly when they are right. That unglamorous discipline will remain at the heart of thermal inspection.