Most bad thermal findings are not sensor failures — they are interpretation failures. These seven mistakes account for the majority of embarrassing reports, and every one is preventable with technique.
- 7Recurring mistakes
- 0That need better hardware
- TechniqueWhat fixes all of them
- TrainingWhere it comes from
Key takeaways
- A hotspot that moves as you orbit is a reflection; a real anomaly stays put on the component.
- Bare metal emits poorly and can read tens of degrees wrong. Grade shiny hardware by comparison with identical neighbours.
- Electrical and moisture work belongs in early morning or after sunset — solar loading masks the deltas you seek.
- Digital zoom adds no information. Close the distance, use optical zoom, or report at detection confidence rather than identification confidence.
On this page
1 & 2: Reflections and Low-Emissivity Surfaces
1. Reading reflections as hotspots. Polished metal reflects the sky, the sun and you. The discriminator is movement: a “hotspot” that moves as you orbit is a reflection; a real anomaly stays put on the component. This test takes ten seconds and prevents the single most common false finding — thermal reflections and false hot spots covers the variants.
2. Trusting temperatures on low-emissivity surfaces. Bare metal emits poorly, so the number on screen can be tens of degrees wrong — a cool-looking bare busbar can be dangerously hot. Grade shiny hardware by comparison with identical neighbours rather than absolute values, which is exactly how substation inspectors work. Emissivity in drone thermography covers the correction.
These two share a root cause: a thermal camera measures radiation and infers temperature, and both reflection and low emissivity break the inference. Neither is a hardware limitation and neither is fixed by a better sensor.
3 & 4: Timing and Palette
3. Flying at solar noon for anomaly detection. Sun-soaked surfaces mask the deltas you seek. Electrical and moisture work belongs in early morning or after sunset, when stored solar energy has dissipated and the remaining differences are the ones you came for.
The exception is solar PV, which needs irradiance — a panel fault only shows when the panel is generating. That inversion catches people who learned the “avoid the sun” rule and applied it universally. Timing rules are specific to each application, and they are in each guide: roof moisture, envelope, solar, turbine blades.
4. Wrong palette for the task. Rainbow palettes hide subtle search targets in colour noise; white-hot hides fine inspection gradients in a narrow greyscale range. Match palette to mission, and switch deliberately rather than habitually — the palette guide has the mapping.
5, 6 & 7: Zoom, Load and Conditions
5. Digital zoom instead of getting closer. Enlarged pixels add no information. Close the distance, use optical zoom, or accept the standoff and report at detection confidence rather than identification confidence — DRI explained covers the difference, and hybrid zoom covers why the big multiplier is not reach.
6. Grading severity without load context. Resistance heating scales with the square of current, so the same defect looks unremarkable at 25% load and alarming at 90%. A finding without a recorded load figure cannot be graded or trended — substation workflow covers why this invalidates more programmes than any other single omission.
7. Not recording the conditions. Ambient temperature, wind, sky state and time bound what the survey could see. A report without them implies complete coverage it cannot support, and a survey flown in conditions outside the valid window is not a weaker survey — it is an invalid one.
| Mistake | The tell | The fix |
|---|---|---|
| Reflection as hotspot | Bright spot moves as you orbit | Change viewing angle before recording |
| Low-emissivity reading | Confident number on shiny metal | Compare with identical neighbours |
| Solar noon timing | Everything warm, no clear deltas | Early morning or after sunset — except solar PV |
| Wrong palette | Target hard to pick out | White hot to search, ironbow to inspect |
| Digital zoom abuse | Big soft image | Optical zoom or report at detection confidence |
| No load context | Severity graded on a mild morning | Record load; grade against phase peers |
| No conditions logged | Report implies full coverage | Log ambient, wind, sky, time with every survey |
The Common Root
None of these seven is a sensor problem, and none is fixed by buying a better payload. They are all failures of technique, and they all come from the same gap: a thermal image is easy to produce and hard to interpret correctly.
That gap is exactly what thermography training addresses — emissivity, reflected temperature, delta-T grading and measurement discipline are the syllabus, and they map one to one onto this list. Thermography training and certification covers what a Level 1-style course contains and why clients increasingly require it.
The second remedy is process. Consistent altitude and angle, logged conditions, comparison against peers, and verification of critical findings at close range — the workflow in the complete guide to drone inspections catches most of these before they reach a report.
Every one of these is preventable in the field. Not one of them requires equipment you do not have. They require ten seconds of angle change, a note of the load figure, and a survey flown in the right hour.
Related reading
- Radiometric Temperature Measurement
- Shutters, Blackbodies and Two-Point Correction
- Thermal Reflections and False Hot Spots
- Emissivity in Drone Thermography
- Thermal Palettes: White Hot, Black Hot and Ironbow
- Drone Thermography Training and Certification
- Substation Thermal Inspection by Drone
- The Complete Guide to Drone Inspections
- MV-2P — 640×512 thermal at 130 g
- OP-90A — zoom, thermal and ranging
- OP-125A — flagship EO/IR pod
- Solar Farm Thermal Inspection
FAQ
How do I tell a reflection from a real hotspot?
Change your viewing angle and watch what happens. A reflection moves across the surface as you orbit, because it is an image of something else; a genuine thermal anomaly stays fixed on the component. This ten-second test prevents the most common false finding in thermal inspection, and it costs nothing but a slight reposition before you record.
Why is solar PV an exception to the timing rule?
Because a photovoltaic fault only reveals itself when the panel is generating. Most thermal anomaly work avoids solar loading because it masks the differences you are hunting; solar inspection needs irradiance because the defect is a failure to convert it. Operators who learn the general rule and apply it universally fly solar surveys at dawn and find nothing.
Which of the seven causes the most damage?
Grading severity without load context, because it invalidates entire programmes rather than individual findings. Resistance heating scales with the square of current, so surveys flown at low load systematically under-report every resistive defect — and a programme that reports “no findings” after a low-load survey has drawn a conclusion its data does not support. Reflections produce embarrassing single findings; missing load context produces false confidence at scale.
Questions about the technology? Talk to our engineers — we reply within 2 business days.