Aerial thermal mapping is a pipeline, not a camera setting: the payload has to record temperature rather than relative contrast, the flight has to be planned around the pixel size your deliverable needs, and the software has to accept the files. Get any one of those wrong and the map looks convincing while being unusable. This guide covers the whole chain for the payload we sell into it — the Sentera 6X Thermal Pro — and the processing path most ag and inspection teams use, PIX4Dfields.
First check: does the software accept your radiometric bands?
Before planning a single mission, confirm how your processing tool treats the files your sensor writes. In PIX4Dfields the distinction is stark:
- Radiometric thermal bands are fully processed, with quantitative pixel values, for the LWIR bands of the MicaSense Altum, Altum-PT and Sentera 6X. Pixel units are stored in centikelvin and convert with standard band math:
Thermal_ir = (LWIR / 100) - 273.15. - JPEG thermal imagery can be stitched, but it is treated as ordinary RGB. The result is a convincing relative colour picture of hot and cold areas with no extractable temperatures, and no histogram analysis.
That single constraint decides whether you are producing a measurement or a photograph of heat. Our Sentera 6X Thermal Pro Sensor Kit for Matrice 300 / Matrice 350 / Matrice 400 writes the radiometric band set; a Sentera 6X Calibrated Reflectance Panel Kit is what makes those values comparable between flights and lighting conditions.
Plan from the deliverable backwards
Decide the output first. An orthomosaic, individual calibrated stills, management zones, or an anomaly list are four different flight plans. For agriculture, the useful output is usually zonation a machine can follow; for asset inspection it is an anomaly log a technician can act on.
Set altitude from the smallest thing you must resolve. Not the height the aircraft can reach. For flat agricultural terrain, 60–100 m is typical, and the thermal sensor should be at least 640×480 to reconstruct fine detail without blurring. For inspection, work back from the defect: if it will not cover enough detector pixels, no amount of processing recovers it.
Raise the overlap. Thermal imagery has lower spatial detail and higher contrast variability than RGB. PIX4Dfields guidance specifies up to 90% frontal and 90% side overlap for thermal projects to avoid stitching failures. In weak-texture scenes photogrammetry tools generally want more overlap than RGB work — DJI Terra documents around 80% frontal and 70% side for thermal, dropping to roughly 60% front and 50% side where a dual-sensor pass uses visible-light matching to place the thermal frames.
Fly a single grid, nadir. For agricultural processing, a single-grid path with the sensor pointing straight down is what the software expects; double-grid and oblique angles are not recommended.
Time the flight to the physics, not the calendar. Solar arrays need sustained irradiance and operating load; moist flat roofing needs the thermal mass trick — after sundown the dry insulation sheds heat while wet insulation keeps radiating — and ag work needs consistent solar conditions between cycles. Record ambient temperature, wind, cloud cover, irradiance or load, soil moisture and irrigation activity alongside each flight.
Pair every thermal capture with visible light. Whether from a dual-sensor payload or a matching RGB pass, the visible frame is how a client (and you) separates a real defect from debris, discolouration or a shadow.
Standstill the aircraft at capture points. Motion blur is the enemy of radiometric detail, and the blur is invisible until you are processing.
Capture discipline
- Let the sensor stabilise before collecting; a radiometric payload warming up produces non-uniformity and calibration drift.
- Use the reflectance panel consistently, before and after the mission, and treat it as part of the payload rather than an accessory.
- Fly the reflectance panel shot the same way every time — a calibration routine that varies between flights is not a calibration.
- Land with reserve power rather than squeezing the last lines out of a pack.
- Do not immediately re-fly a line after landing; a warmed sensor produces a different image from a cooled one.
- Keep the grid, direction, altitude and settings identical between cycles. Data you cannot compare over time is data you cannot bill for twice.
Processing and outputs
Import the radiometric frames and stitch them into a georeferenced orthomosaic. What a client should receive:
- Radiometric dataset — calibrated temperature arrays with absolute values retained per pixel.
- Orthomosaic — a GeoTIFF with an explicit palette, scale and fixed temperature range, so colours mean the same thing on every delivery.
- Zonation or anomaly layer — management zones exported as Shapefile, KML or GeoJSON with calculated acreage for ag work; an anomaly log with coordinates, apparent ΔT, severity and cropped thermal plus visible views for inspection work.
- Paired visible-light dataset at the same coordinates.
- Flight and environmental log — date, time, ambient temperature, wind, irradiance or load, sensor gain, emissivity corrections.
Multiple flights over the same field or asset can sit in one project as separate layers, which is what makes change-over-time reporting possible.
Read the map before you write the report
A thermal map shows surface temperature, not a diagnosis. Anomalies come from load, reflection, emissivity and geometry, and a hot spot is not automatically a defect. Change your viewing angle and shoot again: a reflection moves with your perspective, a real fault stays put. In agriculture the same caution applies in agronomic terms — a temperature difference can come from soil compaction, texture, drainage, canopy structure or moisture, so a scout has to walk the zone before anyone treats it.
Rank findings by consequence, not by temperature. A modest signature on a critical component matters more than a hot auxiliary part.
What we stock
Prices are our live catalog listings; confirm kit and gimbal compatibility on the product page before ordering.
How this guide was built
Processing constraints, supported sensors and flight-parameter figures come from PIX4D's own PIX4Dfields documentation and DJI's Terra manual; the payload and panel come from the live listings in our catalog, checked on November 14, 2026. We have not flown this sensor ourselves. Software guidance changes between releases, so open the source before you rely on a parameter for a paid job.
Sources
- PIX4D — Processing thermal images in PIX4Dfields
- PIX4D — PIX4Dfields acquisition guidelines
- DJI Terra user manual — thermal
- Sentera — manufacturer specifications for the 6X Thermal Pro.
