Multispectral and thermal cameras compared

Choose bands and calibration before choosing megapixels

An agricultural camera should capture the quantities needed for the decision. Multispectral instruments measure reflectance in selected bands; thermal instruments observe emitted infrared radiation. RGB cameras add visible detail for inspection and reconstruction. More channels or more pixels can expand analysis, but they also add integration, storage, calibration and processing requirements.

The five options below include standalone payloads and DJI's integrated Mavic 3 Multispectral camera system. They represent different routes to crop imagery, rather than five interchangeable accessories. Confirm the aircraft mount and power arrangement before comparing a payload quotation with the price of an aircraft that already contains its optics.

Understand what each instrument supplies

MicaSense RedEdge-P combines five narrow multispectral bands with a panchromatic channel. The panchromatic data support finer-detail outputs through appropriate processing; they do not change the native information recorded by each multispectral detector. MicaSense Altum-PT adds a thermal channel alongside multispectral and panchromatic observations, making it relevant when temperature and reflectance need to be collected together.

Sentera 6X is a sensor family with multispectral and thermal configurations. Select the exact model: the presence and thermal characteristics of one version must not be transferred to another. MAPIR Survey3W is offered with several filter combinations, including red, green and near infrared. Choose the spectral variant deliberately; a visible-light Survey3W is not the same instrument as its RGN configuration.

DJI Mavic 3 Multispectral integrates an RGB camera and four multispectral cameras on a dedicated aircraft. Its compact acquisition package avoids a separate camera mount, but binds the sensing workflow to that aircraft. Evaluate the raw band images, sunlight information and processing compatibility needed by your software.

Resolution is band-specific

Check the ground sampling distance of each band at the proposed flight height. A high-resolution RGB or panchromatic channel can show structures that a lower-resolution spectral band cannot independently resolve. Pan-sharpening combines information to create a finer-grid product, but that is a processing operation with assumptions. For plot measurements, inspect the unsharpened bands and alignment as well as the attractive composite image.

Different sensors use different band centres and bandwidths. Two products labelled near infrared may sample somewhat different wavelengths, so identical index formulas can yield different values. Maintain the same instrument and calibration approach for repeat trials, or develop a documented cross-sensor comparison. Do not interpret a camera change as a crop change.

Thermal pixels also differ from visible-image pixels. Canopy temperature analysis needs enough pixels on the crop and an approach to separating foliage from exposed soil. Emissivity settings, reflected radiation, air conditions and sensor calibration affect interpretation. Thermal imagery can inform water-stress assessment, but a hot pixel alone does not measure root-zone deficit.

Calibration and installation

Use the manufacturer's reflectance panel and irradiance procedure. Keep the panel clean, level and free of operator shadow, and capture it under the conditions specified for the system. Record panel identifiers and retain the raw images. An upward-looking sunlight sensor should have the correct mounting orientation and an unobstructed view. Check aircraft vibration, trigger synchronisation, cable retention and clearance through the full gimbal movement.

Verify timestamps, camera positions and lens settings in the processing software. Spectral channels must align well enough for the size of features being analysed. A small offset can create false index edges along rows or plot boundaries. Inspect several raw bands and a processed sample before committing to a season-long workflow.

Compact multispectral cameras brought band-specific aerial analysis to small unmanned aircraft after agricultural remote sensing had already developed through airborne and satellite instruments. Modern payloads add synchronisation, irradiance measurement and integrated thermal channels. Historical camera names remain common in research papers, so match a study to its exact sensor rather than assuming a current model has identical bands.

Terms on a payload datasheet

Reflectance

The fraction of incident radiation reflected by a surface under defined conditions.

Radiometry

The measurement and calibration of electromagnetic radiation.

Panchromatic

A broad spectral channel used for spatial detail.

Red edge

The region of rapid change between red absorption and near-infrared reflectance in vegetation.

Irradiance

Incoming radiant power per unit area.

Questions about camera selection

Do I need thermal for NDVI?

No. NDVI uses red and near-infrared reflectance. Thermal supplies different information and a separate interpretation workflow.

Can I mount any payload on my drone?

Check supported mounts, power, triggering, mass and software. A physical fit does not establish a safe or calibrated integration.

Can I compare maps from different days?

Use consistent geometry, illumination handling and calibration. Retain raw data and metadata so that an apparent change can be investigated.