Products · Intelligent Payloads and Accessories

Choosing a Thermal Imaging Payload

Thermal payload selection is driven by what you are trying to see, how far away, and whether you need a temperature value or just a temperature difference.

2 of 10Method note

Overview

Decide radiometric or non-radiometric first. A temperature value you cannot calibrate is not a measurement.
Detector resolution and lens together set the instantaneous field of view; both matter for small targets at distance.
Thermal contrast depends on the scene, not only on the camera — a hot object on a hot background is not detectable regardless of resolution.
Check that the recorded format retains the original radiometric data, or the images cannot be re-analysed.

Thermal shows temperature, and that is both the value and the limit

A thermal imaging payload produces an image of apparent temperature rather than of reflected light. That is the whole basis of its value: it works in darkness, it sees through some obscuration, and it reveals thermal behaviour that precedes visible failure in electrical connections, mechanical components, building envelopes and process equipment. It is also the whole basis of its limitation: it only shows what has a thermal signature, and many failure modes do not have one until they are advanced.

So the first selection question is not resolution or sensitivity — it is whether the target condition is thermally observable at all. A loose bolt that has not yet begun to heat has no signature. A delaminated roof panel may only show a signature during a particular part of the day. If the signature is not there, no specification will help, and the honest answer is that thermal is the wrong tool for that failure mode.

The second question is when the signature is observable. Thermal contrast depends on the thermal state of the object and its surroundings, which changes with time of day, weather, season and recent operating history. Electrical load, solar loading and wind all move the contrast. Selecting a thermal payload without deciding when the survey will fly is specifying a sensor without specifying the measurement.

Resolution, sensitivity and the specification that matters

Thermal resolution is usually much lower than visible-light resolution, and it is specified in detector pixels rather than in image pixels. What matters operationally is not the detector count but the spatial resolution achieved at the working distance — how small an object can be at that distance and still produce a reliable temperature indication rather than a blended average with its surroundings.

That distinction is the most important and most often missed point in thermal selection. An object smaller than the instantaneous field of view does not produce its own temperature; its reading is diluted by whatever else fills that spot. Inspection specifications should therefore be written as a minimum object size at a working distance, and the payload selected against that, rather than against a detector count.

Thermal sensitivity — the smallest temperature difference the sensor can resolve — is the other axis, and it trades against spatial resolution for a given detector and optics. Higher sensitivity is valuable where the signature is subtle; spatial resolution is more valuable where the target is small. Decide which your failure mode needs before comparing, because the two pull against each other.

Radiometry, calibration and whether the numbers mean anything

There is a real difference between a thermal camera that produces an image and one that produces measurements. A radiometric payload, properly calibrated, gives temperature values that can be compared across time and between assets. A non-radiometric one gives relative contrast, which is sufficient for finding a hot spot and insufficient for trending one.

If the programme will compare readings across epochs — which is where thermal inspection produces most of its value — radiometry and calibration discipline are requirements, not upgrades. That includes a documented calibration state, consistent emissivity settings, consistent distance and a recorded ambient reference. Without those, two readings taken months apart are not comparable and the trend is an artefact.

Emissivity deserves particular attention because it is a property of the surface, not the sensor. Different materials radiate differently, and shiny or reflective surfaces produce misleading readings that include reflected radiation from elsewhere. Where the asset mix includes reflective materials, expect to manage this explicitly rather than to rely on a default setting.

Operating conditions and integration

Thermal payloads are used in the conditions where visible-light payloads struggle — night, smoke, light obscuration, low sun — and that is precisely the point. It also means the flight planning assumptions differ: surveys may be scheduled for night or for a specific part of the day when contrast is best, and crew and airspace arrangements need to accommodate that.

Integration follows the same four checks as any payload: mounting provisions, available power, control and data interface, and mass and centre-of-gravity envelope. Thermal payloads are typically compact and low-power relative to their value, but the interface question is unchanged and should be confirmed rather than assumed. Our payload interface and platform compatibility note covers this.

Where thermal is combined with a visible-light payload on the same aircraft, the value is in registration — being able to see the thermal anomaly and the visible context together. That requires the two streams to be geometrically consistent, which is a payload rig and processing question rather than a sensor one, and it should be specified as such.

Selecting in practice

The practical sequence is: confirm the failure mode is thermally observable; establish when the contrast is best; write the requirement as a minimum object size at working distance; decide whether radiometric measurement is needed; then confirm the four interface properties. That order prevents the common failure of buying a capable thermal sensor that cannot answer the question the programme actually has.

Typical applications where this works well include substation thermal anomaly patrol, solar PV thermal anomaly detection, storage tank anomaly recording and night thermal search. Each has a different signature and a different best time to fly.

Send your asset types, the failure mode you are looking for, the working distance, whether you need measured temperatures or only contrast, and your platform to our engineering team. We will confirm whether thermal is the right tool and, if it is, propose the payload and the survey conditions it needs.