Inspect IQ

Drone Thermal Inspection: How Thermography Works and What Heat Patterns Reveal

By Kush Agarwal, President and CEO, YelloSKYE. Published 2026-10-19.

Executive Summary

• The Question: How does drone thermal inspection work, and when can you trust what a thermal image is telling you?
• The Answer: A thermal camera records the infrared energy a surface gives off and converts it into apparent temperature. Faults appear as heat in the wrong place, but only when the survey is flown in conditions that create heat flow, the readings are corrected for the surface material, and a visible image confirms what the warm patch really is.
• The YelloSKYE Angle: Inspect IQ flies radiometric thermal and high-resolution RGB sensors together, and every anomaly becomes a located record in the DEX platform that can be compared survey to survey.

Drone thermal inspection uses an infrared camera on a drone to map surface temperatures across an asset, so that faults which change how heat moves become visible: a wet patch of roof insulation, an overheating electrical joint, a failing solar cell. Nobody has to climb up and nothing has to be switched off. The drone provides fast coverage of large or awkward assets; thermography is what turns the flight into a diagnosis.

At YelloSKYE, thermal capture is part of Inspect IQ, our asset inspection solution. This guide is about the method rather than one asset type. It explains how a thermal camera forms an image, which conditions make a survey reliable or worthless, what common heat patterns mean on roofs, solar arrays, electrical equipment and industrial plant, and why a thermal image should always be read beside an ordinary photograph. Roofs get their own treatment, from planning to repairs, in our guide to drone roof inspection.

How a thermal camera turns heat into an image

Everything warmer than absolute zero gives off infrared radiation, and the warmer an object is, the more it emits. Thermal cameras used for inspection are tuned to the long-wave infrared band, roughly 8 to 14 micrometres, where objects at everyday temperatures emit most strongly. Each pixel measures how much of that energy arrives, and software converts the values into a colour scale.

Two consequences shape every inspection.

The camera sees surfaces only. It cannot look through a membrane, a wall or a glass panel. What it records is the surface effect of something beneath. Water in roof insulation holds heat longer than dry insulation, so the membrane above it cools more slowly after dark. Thermography is therefore an inference, and a good analyst always asks which physical process would produce the pattern on screen.

The value the camera reports is an apparent temperature. It assumes the surface emits energy in the expected way and that nothing else is adding to the signal. When those assumptions fail, the number is wrong, however sharp the image looks.

Radiometric and non-radiometric cameras

A non-radiometric camera gives a picture of relative warmth. It can tell you that one component is hotter than its neighbours, but its values cannot be relied on. A radiometric camera stores a calibrated temperature for every pixel, and those values can be corrected later for the material, the stand-off distance and the air conditions. When findings feed maintenance budgets, radiometric data is the norm, because the analyst can compare one anomaly with another and follow the same component across several surveys. Inspect IQ uses radiometric thermal sensors for this reason.

Emissivity and reflection

Emissivity describes how efficiently a surface radiates compared with a perfect emitter. Most non-metallic materials on buildings and plant, including concrete, brick, roofing membranes, painted steel and glass viewed square on, have high emissivity, so their readings are dependable. Bare and polished metals have low emissivity. They give off little energy of their own and reflect a great deal of their surroundings, including the sky and the sun.

So an unpainted busbar can look cool while running hot, or look hot because it mirrors a warm flue nearby. Analysts set emissivity for each material, read temperatures from painted or oxidised areas where they can, and check whether a hot spot moves when the viewing angle changes. A spot that moves is a reflection; one that stays put is a real heat source.

Resolution and stand-off distance

Thermal sensors have far fewer pixels than visible cameras; 640 by 512 is a common resolution for drone thermal payloads. If a defect spans fewer than a few pixels, the camera blends it with cooler material around it and under-reports its temperature. Set flight height and stand-off distance by the smallest fault you need to measure: a single solar cell or a cable lug needs a much closer pass than a roof moisture survey.

Conditions that make or break a thermal survey

A thermal flight in the wrong conditions produces convincing images that mean very little. Faults must be creating a temperature difference, and the environment must not hide it.

The right time of day depends on the asset

There is no universal best hour for thermography. Each asset needs its own source of heat flow.

• Low-slope roofs: fly after a clear, sunny day, once the sun has gone down. The roof stores heat during the day. Dry areas shed it quickly at dusk, while wet insulation keeps releasing it, so moisture shows as warm patches.
• Solar arrays: fly in strong, steady sunshine with the modules producing close to full output. Faults appear only while current flows through them, which is why the IEC technical specification for outdoor photovoltaic thermography (IEC TS 62446-3) sets a minimum irradiance for valid surveys.
• Building envelopes: fly when the gap between indoor and outdoor temperature is largest, often early in the morning before sunlight reaches the walls, so heat escaping from inside is not swamped by solar gain.
• Electrical and process equipment: fly while it runs under normal or peak load, and keep direct sun off the measured components where possible.

Weather before the flight matters too

An asset carries a thermal memory of the hours before the survey. A roof flown after an overcast day may show no moisture at all, because it never stored enough heat to give back, and an afternoon shower erases patterns by cooling everything evenly. Check the previous day's weather as well as the flight forecast, and reschedule rather than fly a survey that cannot be read.

What heat patterns reveal, asset by asset

Thermography is mostly pattern reading. The shape, size and position of a warm or cool area tell you more than its peak value.

Roofs and building envelopes

On flat roofs, wet insulation shows after sunset as irregular warm areas, often spreading out from a seam, a penetration or an outlet. Sharp-edged rectangles are more likely to be a change of material or an old patch repair. On walls, regular warm lines that follow the structure, such as slab edges, columns and lintels, point to thermal bridging. Irregular warm zones suggest missing or slumped insulation, or air leaking through gaps. YelloSKYE's automated asset inspection use case describes radiometric sensors being used to assess insulation health, HVAC performance and moisture hidden behind facades.

Solar arrays

Drone thermography is now routine on photovoltaic modules, because their faults have recognisable signatures.

In YelloSKYE's solar farm thermal inspection work, hot spots, diode failures and string disconnections are flagged automatically and each is tied to its GPS position, so maintenance crews go straight to the right module instead of searching rows. The same use case reports a 75 per cent reduction in inspection costs compared with manual ground patrols.

Electrical assets

In electrical systems, resistance turns into heat. A loose, corroded or undersized connection forces the same current through a smaller contact area, so it runs hotter than the conductor on either side. On overhead lines, towers and outdoor substations, drone thermography looks for warm joints, clamps and terminations, and for one phase running warmer than the other two under the same load, which can indicate imbalance or a developing fault. For grid assets, Inspect IQ also checks for cracked insulators, missing hardware and vegetation encroachment, without line outages. Indoor switchgear and panels are beyond a drone's reach and remain a job for handheld thermography.

Industrial structures

Stacks, kilns, ducts, tanks and lagged pipework all keep heat in or out. A warm band on the outside of a stack can mean the refractory lining inside has worn away. Warm spots along insulated pipework show damaged or soaked lagging. In storage tanks, the boundary between liquid and vapour often appears as a temperature line, giving a quick check of fill level or sludge build-up. Inspect IQ's vertical stack scans look for thermal anomalies alongside concrete spalling and surface corrosion, which is where flying more than one sensor earns its keep.

Why thermal and RGB belong in the same flight

A thermal image on its own invites wrong conclusions. A warm patch on a roof might be wet insulation, a dark heap of debris that soaked up sunshine, or a reflected flue. The visible image settles most of these questions in seconds.

Flying both sensors together gives each finding what a thermal image cannot supply alone: a clear view of what the object is, a sharp record of any visible damage and a location precise enough to find it again. In practice, analysts orient themselves on the RGB map, detect on the thermal layer and return to the RGB to explain what they found.

For a wider view of which payload suits which job on a project, see our guide to drones in construction.

From thermal image to maintenance decision

A survey earns its cost only when its findings are acted on.

1. Correct the radiometric data. Apply emissivity by material, distance and air conditions so temperatures can be compared fairly.
2. Classify each anomaly. Record its type, its temperature difference against a similar component nearby, and the probable cause. The difference is usually more telling than the absolute value, because it cancels out the day's weather.
3. Locate it. Tie every finding to coordinates and to the owner's own reference: module and string, tower and phase, roof zone.
4. Verify the serious ones. A thermal finding is a reason to look closer, so confirm it with an electrical test, a moisture probe or a visual check before major spending.
5. Track it over time. A warm joint that stays stable from year to year is a different risk from one that gets hotter each survey.

Inspect IQ hands findings over as a structured defect database: annotated images, GPS coordinates, severity ratings and recommended remediation, exportable to PDF, Excel and BIM-linked formats. Inside the DEX platform, defects are geo-tagged on the captured asset, dashboards show open issues and severity across a portfolio, and the historical comparison lets you slide between survey dates to see when a defect started to form.

What drone thermography cannot tell you

Knowing the limits keeps the method honest.

• It cannot see inside. Some hidden problems, such as dry voids or small cracks beneath a membrane, give no thermal signal at all.
• It depends on conditions. Some days and seasons simply do not offer enough temperature difference.
• It is not equally accurate on every surface. Shiny metal, steep angles and very small targets distort readings.
• It does not replace testing. Electrical faults need electrical confirmation, and moisture findings need cores or probes before a large repair.
• It records one moment. Trends need repeat surveys in comparable conditions.

Thermal survey checklist

Use this before commissioning or flying a thermal inspection:

• Write down the question: moisture, electrical faults, solar performance, insulation or process equipment.
• Pick the time window that creates heat flow for that asset, and check the previous day's weather.
• Confirm the asset will be energised, generating or loaded during the flight.
• Set flight height from the smallest defect you need to measure.
• Capture radiometric thermal and high-resolution RGB together, and log wind, cloud and air temperature.
• Apply emissivity by material and treat readings from bare metal with caution.
• Classify anomalies by pattern and temperature difference, not by colour alone.
• Verify serious findings on the ground, and repeat surveys in comparable conditions on one timeline.

Explore Inspect IQ