An isotherm turns judgment into a threshold: paint every pixel above 80 °C red, and the overheated connector announces itself — no operator vigilance required. It is the simplest automation in thermography and the most underused.
- 1 thresholdReplaces continuous vigilance
- Delta-TWhat to set it from
- RadiometricThe hard prerequisite
- Post-flightWhen it can be re-run
Key takeaways
- Radiometric payloads know each pixel’s temperature, so display logic can colour only the exceptions and leave the rest greyscale.
- Alarm-driven patrol shifts operator attention from “scan everything” to “verify flags”, which survives fatigue on long corridors.
- Set thresholds from delta-T above reference components, not from absolute numbers — absolutes inherit every measurement uncertainty you have.
- Radiometric recordings can be re-thresholded afterwards, so yesterday’s flight can answer tomorrow’s question.
On this page
How Isotherms Work
Radiometric payloads store a temperature value for every pixel, which means display logic can do something a plain thermal camera cannot: colour-code by absolute value rather than by relative brightness. Above-threshold, below-threshold, or between two limits — each can be painted while everything else stays neutral greyscale.
The perceptual effect is the point. A conventional palette maps the whole scene across a colour ramp, so the operator has to judge which warm region is warm enough to matter. An isotherm removes that judgement from the moment of observation and moves it to the moment of threshold-setting, where it can be made carefully and documented.
The scene stays greyscale; only exceptions shout. That is the entire design, and it is why isotherms outperform prettier display modes for routine inspection work.
Alarm-Driven Patrol
Set the threshold from your severity standard — say 25 °C above ambient for electrical hardware — fly the route, and let the payload flag candidates. Attention shifts from “scan everything” to “verify flags”, and that shift is what survives operator fatigue on kilometre-long corridors.
This matters most exactly where human attention fails. An operator watching a power line corridor for forty minutes is not watching it as well in minute thirty-eight as in minute three, and the defect does not care which minute it appears in. A threshold does not get tired.
The related benefit is consistency between operators and between flights. Two people scanning the same corridor apply different implicit thresholds; the same configured isotherm applies the same one. For a monitoring programme where this quarter has to be comparable with last quarter, that consistency is worth more than the fatigue resistance.
It pairs naturally with AI detection, which solves the adjacent problem of noticing objects rather than temperatures.
Choosing Thresholds Honestly
Thresholds inherit every measurement uncertainty you have. Emissivity, distance, viewing angle, wind, solar loading — each shifts the apparent temperature, and an absolute threshold treats all of that as if it were signal.
Set them from delta-T above reference components rather than absolute numbers. “25 °C above the coolest comparable phase” is robust because the reference experiences the same conditions as the target; “above 80 °C” is not, because 80 °C means something different on a shiny aluminium clamp than on a painted steel one — emissivity in drone thermography covers why shiny metal in particular misleads.
| Threshold basis | Robustness | Failure mode |
|---|---|---|
| Absolute temperature | Low | Emissivity and distance errors read as exceedances |
| Delta-T above ambient | Moderate | Solar loading and wind shift the ambient reference |
| Delta-T above reference component | High | Requires a comparable peer in frame |
| Delta-T versus same asset last survey | Highest for trending | Requires logged conditions on both flights |
And log the conditions with each run. A threshold-based finding without recorded conditions cannot be defended, and cannot be compared with the next survey — the same discipline covered in substation workflow, where load is the variable that invalidates everything.
A threshold encodes a decision about significance. Set it from the standard or equipment specification you are inspecting against, not from what makes the screen look tidy. Too low and everything paints; too high and you have built a display that never triggers.
Where It Fits in the UAVThermal Range
Thermometry versions across the range support spot and region measurement with configurable high and low alarms. The prerequisite is calibrated per-pixel data, so this is a capability of the radiometric configuration rather than of the payload model.
| Payload | Published temperature measurement entry |
|---|---|
| LX-9A | ±3 °C or ±3% of the reading |
| LX-9B | Optional — thermometry type |
| OP-125A | Optional — thermometry type |
| OP-90P | Optional — thermometry type |
| MV-2P | Supported via subsequent firmware updates |
Specify radiometric when you request a quote, and describe your alarm workflow so the configuration matches what you intend to do. Radiometric vs non-radiometric covers why the decision is irreversible — a non-radiometric payload cannot run isotherms at all, because it has no per-pixel temperature to threshold against.
What Isotherms Do Not Solve
An isotherm finds pixels above a number. It does not know whether those pixels are a fault, a reflection or a piece of equipment that is supposed to be hot.
Reflections are the classic false trigger: a shiny surface reflecting the sun or a warm object nearby reads above threshold and paints exactly like a real hotspot. The standard discriminator still applies — change viewing angle and see whether it moves — but the isotherm will not apply it for you. Thermal reflections and false hot spots covers the misreads.
Nor does thresholding replace severity grading. A component 30 °C above its peers at 20% load is a more serious finding than one 30 °C above at full load, and no threshold captures that. The isotherm narrows the frame to candidates; a thermographer decides what they mean — see thermography training and certification.
Related reading
- Radiometric Temperature Measurement on UAV Payloads
- The Thermal Image Processing Pipeline
- Radiometric vs Non-Radiometric Thermal
- Thermal Palettes: White Hot, Black Hot and Ironbow
- Emissivity in Drone Thermography
- Thermal Reflections and False Hot Spots
- Substation Thermal Inspection by Drone
- AI Target Tracking on UAV Payloads
- OP-125A — thermometry optional
- LX-9B — thermometry optional
- MV-2P — 640×512 thermal at 130 g
- Power Line Inspection Drone Payloads
FAQ
Do isotherms work in recorded data?
Yes — radiometric recordings store per-pixel temperature, so they can be re-thresholded in post-processing. Yesterday’s flight can answer tomorrow’s question, and a survey captured at one threshold can be re-examined at another without re-flying. This is a strong argument for capturing radiometric even when the live workflow does not need it, provided the payload is a thermometry configuration.
Isotherm or AI detection?
They solve different problems and work well together. An isotherm finds pixels above a temperature; AI detection finds objects of a class regardless of temperature. On an electrical corridor the isotherm is what you want, because the anomaly is defined thermally. On a search or security patrol, AI detection is, because the target is defined by what it is rather than how hot it is.
Why set thresholds from delta-T rather than absolute temperature?
Because an absolute threshold inherits every measurement uncertainty in the chain — emissivity, distance, viewing angle, wind, solar loading. A reference component in the same frame experiences the same conditions, so comparing against it cancels most of that out in a single step. “25 °C above the coolest comparable phase” survives scrutiny; “above 80 °C” invites an argument about whether the emissivity setting was right.
Questions about the technology? Talk to our engineers — we reply within 2 business days.