A thermal drone payload lets a search and rescue team detect a human heat signature at up to roughly 1,000 m with a 640×512 sensor, in total darkness, light smoke and partial vegetation cover. Detection is the easy part; the specification that decides whether a search succeeds is the range at which you can tell a person from a deer — and that is four times shorter.
- 640×512Minimum array for standoff search
- 8–14 µmLWIR band
- ±0.01°Stabilisation class
- ~1,000 mDetection, clear conditions
Key takeaways
- Detection is not recognition. A warm blob at 900 m does not dispatch a team; recognition range is typically a quarter of detection range and is the number to plan on.
- Search wide, confirm narrow. Zooming during the sweep is the most common way to waste a sortie.
- Thermal does not see through canopy, rubble or water. It reads the outermost surface, always.
- A detection without a coordinate is not a rescue. Laser ranging converts a sighting into a dispatch.
On this page
Why SAR teams fly thermal
Search windows shrink fast: survivability drops after the first 24 hours, and most ground search happens in poor light. Thermal imaging measures emitted heat in the 8–14 µm band, so a person reads clearly against cooling terrain at night — precisely when ground search is slowest, most dangerous and least effective.
The operational gain is coverage rate. A drone crew sweeping a valley floor covers in twenty minutes what a ground team covers in hours, and does it without exposing searchers to the terrain that caused the incident. That does not replace ground teams; it tells them where to walk.
The second gain is passive operation. Unlike a searchlight, thermal emits nothing — useful in wilderness settings where light discipline matters, and essential where the subject may be avoiding detection.
What the payload must detect — and at what range
Human heat signatures at standoff distance; residual heat trails; distinction between people, animals and warm rocks. That last requirement is what sets the real planning range.
Published detection figures describe the distance at which something warm is visible. Recognition — knowing it is a person — requires roughly four times as many pixels across the target, so it happens at roughly a quarter of the distance. Search leg spacing should be planned on recognition, because a detection you cannot classify does not dispatch anybody. The arithmetic behind this is in how DRI figures are calculated.
| Condition | Detection | Recognition | Notes |
|---|---|---|---|
| Person on cool open ground, clear night | 800–1,200 m | 200–350 m | The datasheet case |
| Person prone or crouched | 400–700 m | 130–250 m | Reduced projected area |
| Person in light scrub | 300–600 m | 120–220 m | Partial occlusion |
| Person under closed canopy | — | — | Thermal does not penetrate foliage |
| Person in water, cold (5–12 °C) | 300–600 m | 100–200 m | Wet head only, reduced signature |
| Person in water, warm (>26 °C) | 100–250 m | < 100 m | Contrast approaches zero |
| Daytime, sunlit rock and soil | 200–500 m | < 150 m | Background may match body temperature |
| Residual heat trail or sit spot | 50–150 m | — | Minutes-scale persistence only |
Plan leg spacing on recognition, not detection. Using the detection figure produces a search pattern that generates contacts nobody can classify, and a crew that spends the sortie re-flying to identify warm rocks.
What defeats thermal search
Being explicit about limits during training prevents crews from losing confidence in the tool on scene.
Canopy. Long-wave infrared does not penetrate foliage. A subject under closed forest canopy is not detectable from above, at any resolution. Search the gaps, trails, watercourses and edges.
Water. A person in water exposes a wet head only, and wet skin approaches water temperature within minutes. In warm water the contrast can fall below one kelvin — see maritime SAR for the full treatment.
Thermal crossover. At dawn and dusk, background temperature passes through body temperature and contrast briefly disappears entirely. It is not equipment failure; it is a twenty-minute window to plan around.
Daytime sun. Sunlit rock and soil can exceed skin temperature, inverting or erasing the signature. Daytime thermal search over open ground is far weaker than night search, which surprises crews trained only at night.
Specification priorities for SAR
Rank in this order, because each one gates the next.
Thermal resolution — 640×512 over 256×192 for standoff search. This directly sets both detection and recognition range and cannot be recovered downstream. Gimbal stability (±0.01° class) — search is flown at speed and any residual motion smears a small target into the background. Zoom for identification once a contact is held. Laser ranging for position hand-off to ground teams. AI tracking to hold a moving subject while the aircraft repositions.
Note what is not on the list: palette choice, storage capacity and hybrid zoom figures. They matter operationally but they do not determine whether the subject is found.
From contact to coordinate
The most common way a successful detection fails to become a rescue is location hand-off. A contact reported as “in the gully, north of the treeline” cannot be actioned by a ground team moving in darkness.
A payload with integrated laser ranging computes a target coordinate directly, which turns a sighting into a dispatch. Report it with an accuracy radius rather than as a point — coordinate error grows with slant range and viewing obliquity, and on a search flown at altitude the geometry is rarely ideal. The terms are broken down in the geolocation error budget.
Where ranging is unavailable, the fallback is to hold the aircraft over the contact and vector the team in visually. It works, but it commits the aircraft and ends the search until the team arrives.
Recommended UAVThermal payloads
MV-2P — 640×512 thermal at 130 g for sub-250 g and lightweight platforms. The unit that is already in the vehicle and airborne first.
MV-4X — quad-sensor micro flagship combining starlight, 640×512 thermal and NIR illumination for detection and confirmation without visible light.
OP-90A — thermal with zoom and 1,200 m laser ranging for wide-area search where coordinate hand-off matters.
OP-125A — long-range EO/IR with 2,000 m ranging and AI target tracking for long-range coordination and multi-team incidents.
Related reading
Technology: from NETD to DRI and starlight and low-light imaging.
Field practice: search patterns for thermal SAR, the SAR team equipment checklist and real detection data from six payloads.
Adjacent missions: disaster response and maritime SAR and coastguard.
FAQ
How far can a thermal drone detect a person?
Roughly 800–1,200 m for detection with a 640×512 sensor on a clear night over cool open ground. Recognition — knowing it is a person rather than an animal — happens at roughly a quarter of that, 200–350 m, and that is the figure search patterns should be planned on.
Can thermal find someone under trees?
No. Long-wave infrared does not penetrate foliage, so a subject under closed canopy is not detectable from above at any resolution. Search the gaps, trails, watercourses and forest edges, and use ground teams for closed canopy.
Does thermal search work in daylight?
Far less well than at night. Sunlit rock and soil can exceed skin temperature, which inverts or erases the signature. Crews trained only on night operations are routinely surprised by how weak daytime thermal search is over open ground.
What is thermal crossover?
The period at dawn and dusk when background temperature passes through body temperature and contrast briefly disappears. It typically lasts around twenty minutes and is a planning consideration, not an equipment fault.