Livestock work is the most forgiving thermal application there is — a 38 °C animal on 8 °C pasture is an unmissable target. The difficulty is never detection; it is counting accurately, disturbing the herd as little as possible, and covering ground fast enough that the flight replaces a truck rather than adding to it.
Key takeaways
- Four distinct jobs pay for the aircraft — head counts, birthing checks, predator patrol and finding animals in cover.
- Counting works because animals are consistently warmer than ground; it stops working when the ground is not.
- Contrast degrades in daytime sun, on wet coats and over recently sun-heated surfaces — early morning is the reliable window.
- The honest payback comes from labour hours replaced, not from animals found; count that first.
On this page
Four jobs that pay for the aircraft
Night counts. Thermal counts stock in darkness, in scrub and in terrain where a ground count is slow and inaccurate. On open pasture a well-flown count is routinely within 1–2 % of the true figure.
Calving and lambing watch. Checking a calving paddock at 02:00 without driving through it is the single most valued application we hear from ranchers. A cow separated from the herd and stationary for hours is a visible pattern in thermal, and finding a newborn before dawn changes survival odds.
Predator detection. Coyote, fox and feral dog movement at the paddock edge shows clearly, and unlike a spotlight the drone does not announce itself.
Welfare screening. Lameness shows as an asymmetric hoof or joint temperature; mastitis as udder asymmetry; fever as an elevated ocular or muzzle reading. This is screening, not diagnosis — it tells you which animal to bring in.
Detection and counting parameters
| Task | Altitude AGL | Thermal FOV | Detection range | Accuracy note |
|---|---|---|---|---|
| Cattle count, open pasture | 80–120 m | Wide | 1,000–1,500 m | 1–2 % error achievable |
| Cattle count, light scrub | 50–80 m | Wide | 500–800 m | 5–10 % error; expect occlusion |
| Sheep count, flocked | 60–90 m | Wide | 600–900 m | Clustering causes undercount |
| Calving paddock check | 40–70 m | Medium | 400–700 m | Look for isolation, not just presence |
| Newborn calf / lamb | 30–50 m | Medium | 200–350 m | Small mass, cools fast, fly low |
| Predator (coyote-size) | 50–80 m | Wide | 400–700 m | Gait pattern aids identification |
| Welfare / lameness screen | 15–30 m | Narrow / zoom | 30–80 m | Needs radiometric and calm animals |
What degrades thermal contrast on livestock
- Wet coat. Rain-soaked animals lose several kelvin of apparent surface temperature. Counts after rain undercount.
- Hot ground. On a summer afternoon, pasture can exceed animal surface temperature and contrast inverts or vanishes. Fly at night or in early morning.
- Heavy fleece. A full-fleece sheep is thermally well insulated; the readable signal is largely head and legs. Post-shearing counts are much easier.
- Dense canopy. Thermal does not see through foliage. Timbered paddocks will always undercount from the air.
- Herd clustering. Animals in contact merge into one blob. Counting software and human counters both fail here; a second pass from a different angle usually resolves it.
Recommended UAVThermal payloads
MV-2M — dual-sensor micro pod. Light, quiet and cheap enough to be the everyday tool; thermal for the count, visible for the tag read once you are close.
MV-4M — quad-sensor micro pod for larger holdings where one sortie has to cover many paddocks.
OP-80N — where predator identification and boundary security matter as much as stock, the black-light night-vision channel adds colour identification without visible illumination.
Related reading
Technology: uncooled VOx microbolometers and starlight and low-light imaging.
Field practice: livestock monitoring with thermal drones, thermal drone deer recovery and cold-weather thermal operations.
Adjacent missions: wildlife monitoring and conservation and precision agriculture.
What it replaces, and the honest payback
The comparison is against vehicle time. A large paddock check by truck costs fuel, an hour or more, and disturbance to the stock; the same check by drone costs fifteen minutes and does not push cattle around. On extensive holdings the drone often pays for itself in fuel and labour inside a season.
Calving and lambing is where the return is largest but hardest to quantify. Finding a calf in difficulty at 03:00 instead of at first light changes survival odds materially, and a handful of saved animals a season covers a micro payload outright. Ranchers who run the numbers usually do it on this alone.
Be realistic about the ceiling. A drone does not replace stockmanship, does not muster, and in timbered or broken country it will always undercount. The programmes that work treat it as a fast triage layer that tells you which paddock to drive to — not as a replacement for driving.
Where this goes wrong
Livestock thermal is easy to fly and easy to over-trust.
- Counting after rain. Wet coats shed several kelvin of apparent temperature. Counts flown on a wet night undercount and, worse, undercount unevenly.
- Counting in the afternoon. Sun-heated pasture can exceed animal surface temperature, inverting or erasing contrast. Night and early morning are the working windows.
- Trusting the count in timber. Thermal does not see through canopy. A treed paddock will always read low; report it as a minimum, not a count.
- Descending too far during calving checks. Hovering low over a cow about to calve is the one manoeuvre most likely to cause the problem you flew out to prevent.
- Reading welfare from a distance. Lameness and mastitis screening needs radiometric capability, close standoff and calm animals. From 80 m it is guesswork.
FAQ
How accurate is a thermal drone livestock count?
On open pasture at night, 1–2 % error is achievable and better than most ground counts. Accuracy falls quickly with scrub cover, herd clustering and wet coats — in timbered paddocks expect to undercount regardless of payload quality.
Will the drone disturb the herd?
Above roughly 60 m most cattle ignore a small multirotor; sheep are more reactive and horses considerably more so. Fly higher on the first passes, descend only when needed, and avoid hovering directly overhead during calving checks.
Can thermal detect illness in livestock?
It screens rather than diagnoses. Asymmetric limb or udder temperature and elevated ocular temperature are useful flags that tell you which animal to examine. Reliable readings need a radiometric payload, close standoff and reasonably calm animals.