Satellite, drone or field sensors: choosing a crop monitoring layer

Crop monitoring is rarely the goal in itself. It feeds a few recurring decisions: where to send a scout this week, which zones get more or less nitrogen, when to irrigate a block, and what evidence to file after hail or flooding. Each has a different tolerance for delay, a different spatial unit and a different person acting on it.

Scouting needs frequent whole-farm coverage that points people at problems. Variable-rate inputs need zone-level detail at a few growth stages. Irrigation needs daily readings regardless of cloud. Insurance and compliance files need dated records a third party will accept. Decide which you are buying for first; it changes the ranking below.

Most satellite crop monitoring starts with vegetation indices computed from red and near-infrared reflectance. NDVI saturates in dense canopies, so a field at full cover can look uniformly healthy while nitrogen runs short. Red-edge indices such as NDRE stay sensitive for longer, which is why the narrow red-edge bands on Sentinel-2 matter for mid- and late-season decisions¹.

The binding constraint is cloud: in a wet spring, clear images can be missing for exactly the weeks when a nitrogen or fungicide decision is due. Synthetic aperture radar images through cloud, day and night², but its backscatter responds to canopy structure and moisture rather than chlorophyll, so it complements optical indices rather than replacing them.

Pixel size sets the smallest unit you can manage. With public pixels of ten meters at best¹, headlands and narrow strips blend into mixed pixels, so small or irregular fields, including many smallholder plots, need commercial imagery, drones or ground checks.

Drones resolve what satellites cannot: emergence gaps, individual tree canopies, weed patches and storm damage. The price is labor. Someone has to plan flights, manage batteries, lay out reflectance panels so multispectral values compare between days, and stitch overlapping images into an orthomosaic before analysis begins.

Aviation rules shape what is practical. In the US, small drones fly under 14 CFR Part 107, which requires the remote pilot or a visual observer to keep the aircraft in visual line of sight throughout the flight³; dispensing crop-protection products from the air also brings in 14 CFR Part 137 on agricultural aircraft operations⁴. In the UK, work outside the open category needs a CAA operational authorisation, for example under PDRA01 or a UK SORA-based application⁵.

Rules for flying beyond visual line of sight are changing in several jurisdictions, so confirm the current position with your regulator before planning routine whole-farm flights.

Mixing layers usually means mixing vendors. Before signing, check who owns raw and derived data, whether you can export it in bulk, and whether licenses let you share derived maps with agronomists, landlords or buyers.

Interoperability standards help where machinery is involved. ISOBUS (ISO 11783) defines communication between agricultural machinery and data transfer to farm software⁷, which is how a variable-rate prescription reaches a spreader or sprayer. AgGateway's ADAPT framework provides plugin libraries that convert farm management data to and from a common object model⁶. ColdAI's agriculture stack is sensor-agnostic by design⁸, which in practice means building around open formats and keeping the farm's data exportable.