Drone NDVI analysis uses a multispectral camera to compare how much red light a plant absorbs with how much near-infrared light it reflects, producing a value between −1 and +1 for every pixel of a field. Higher values generally mean denser, healthier vegetation.
The resulting crop health map shows which parts of a field are under stress before the eye can see it, but it only creates value once the weak zones are checked on the ground and tied to a fertilising, irrigation or spraying decision.
What the eye misses in the field
The traditional way to spot crop stress across large acreage is to walk it. The trouble is that by the time stress becomes visible, with yellowing leaves and stalled growth, part of the yield loss has already happened. Someone standing at the field edge cannot see a weak patch in the middle, and walking hundreds of hectares every week is not realistic.
So decisions tend to follow the average: the same fertiliser rate, the same irrigation time and the same spray programme across the whole block. Yet soil type, slope, drainage and cropping history vary within a single field. An average decision wastes inputs on the strong zones and under-serves the weak ones.
The same imagery can be used not only for input decisions but also to work out how much crop there will be before harvest. We explain that work, built on tree counts and field sampling, in drone crop yield estimation. The surface model built by photogrammetry is also used on sites and quarries for drone stockpile volume measurement.
The cost of stress you cannot see
Water is one of the most expensive farm inputs, and agriculture accounts for the largest share of water use worldwide.
According to FAO AQUASTAT's water use methodology, global water withdrawals split 69% agricultural, 12% municipal and 19% industrial.
With a line item that large, giving every block the same irrigation time means a significant share of water and pumping energy goes where it is not needed. The losses are not only on the input side either. FAO's SDG 12.3.1 global food loss indicator estimates that 13.3% of food was lost in 2023 between harvest and retail (on farm, in transport, storage, wholesale and processing), with fruits and vegetables highest at 25.4%. NDVI does not solve post-harvest loss directly, but by showing early in the season where the crop is struggling, it is one of the most practical tools for reducing in-field yield loss and misdirected inputs.
How drone NDVI analysis works: calculating and reading the map
Healthy leaves absorb red light strongly for photosynthesis and, because of their internal structure, reflect near-infrared (NIR) light strongly. In a stressed plant, that contrast shrinks. NDVI turns it into one number:
NDVI = (NIR − Red) / (NIR + Red)
As a general guide, water and some surfaces give negative values, bare soil gives low values close to zero, and values rise as vegetation becomes denser. Absolute thresholds, however, vary with crop, growth stage and camera calibration. That is why an NDVI map is best read relatively: the area that is clearly lower than its neighbours, in the same field on the same date, is the area to investigate. It is the same idea behind anomaly detection: flag what deviates from normal, then look for the cause on the ground.
Is NDVI enough on its own?
No. A few complementary indices and data sources reduce the risk of misreading it:
| Indicator | When it helps | Watch out for |
|---|---|---|
| NDVI | Overall canopy density and stress | Can saturate in dense, closed canopies |
| NDRE (red edge) | Nitrogen status later in the season | Needs a camera with a red-edge band |
| Orthophoto (RGB) | Visual checks, boundaries, channels | Shows what is visible, not stress |
| Soil moisture sensor | Telling drought stress from other causes | A point measurement; read alongside NDVI |
| Satellite imagery | Regular, low-cost monitoring of large areas | Limited by resolution and cloud cover |
We compare where satellites and drones complement each other, by scale and by question, in satellite imagery for agriculture.
Drone NDVI analysis, step by step
- Define the question. Rather than "where is the problem?", ask something tied to a decision, such as "find nitrogen-deficient zones and apply variable-rate fertiliser". Flight timing and index choice follow from that.
- Plan the flight. Multispectral camera, flight altitude, image overlap and resolution are planned together. Fly when shadows are short and light is steady.
- Calibrate radiometrically. Without calibration panel shots before and after the flight and data from a sunlight sensor, maps from different days cannot be compared.
- Process. Images are stitched with photogrammetry into a georeferenced orthophoto, NDVI is calculated and the field is split into zones (low, medium, high). Zoning is the field equivalent of the pixel classification used in computer vision quality inspection on the factory floor.
- Ground-truth. Visit the low zones and establish whether the cause is disease, pests, drought, drainage or nutrient deficiency, taking soil and leaf samples where needed.
- Turn it into a decision. The zone map becomes a prescription map for variable-rate fertilising or spraying, or feeds into the irrigation plan.
- Repeat and compare. Several flights in a season show whether the intervention worked; maps of the same zone from different dates are reviewed side by side. Making that change visible to management is the job of a map-based management dashboard that produces KPIs field by field.
Who gets the most out of NDVI analysis?
Drone NDVI does not deliver the same value on every field. It pays back fastest in these situations:
- Fields with large in-field variation. On big parcels where soil, slope or drainage vary, zone-based treatment makes far more difference than a single rate. On a small, uniform plot, walking it is often enough.
- High-input crops. In orchards, vegetables and industrial crops, where spending on fertiliser, water and crop protection is high, every unit of input sent to the wrong place costs more.
- Contract farming and multi-grower operations. Cooperatives, food companies with contract growers and agronomists can use NDVI to compare many parcels on the same basis and prioritise field visits.
- Damage and insurance. After hail, flood or fire, documenting the boundary and size of the affected area with measurements is stronger evidence than an estimate by eye.
The most common mistakes
Treating the colourful map as the answer. A red zone says "there is a problem", not "what the problem is". Spraying without ground-truthing can mean treating drought stress as disease and wasting inputs.
Comparing without calibration. Without calibration, a map shot on a cloudy day and one shot in full sun will show the same field differently. If you are monitoring through the season, this step cannot be skipped.
Leaving the output in a folder. If the NDVI map sits on a hard drive, decisions are still made by eye. The map belongs in the same system as field boundaries, planting records and sensor data, which is why delivery in GIS-ready formats such as GeoTIFF matters. We describe the same principle for infrastructure in our GIS for municipalities guide.
Relying on a single data source. NDVI shows the weak zone; a soil moisture sensor tells you whether that zone is dry. Read together, they make irrigation decisions far more reliable. We cover the sensor side in detail in our article on precision agriculture and soil sensors, and how those sensors send data across large fields in our LoRa vs NB-IoT vs cellular comparison.
On grazing farms, an NDVI map can be combined with herd location data from GPS collars; we cover that side in our article on the livestock monitoring system.
NDVI shows crop health in open fields; under cover, climate and irrigation are measured and controlled directly. We cover the greenhouse sensor and control layer in our article on smart greenhouse automation.
An NDVI map does not make a decision on its own; it pays off when joined with plot, irrigation and input records. We set out the order of those layers in our digital agriculture guide for farms.
How we handle drone NDVI analysis at Digital Bridge
Our aim is to deliver an output you can act on, not a folder of raw imagery. This is how we work:
- We pin down the decision with you. Crop, growth stage and decision (fertiliser, irrigation, spraying, damage assessment) are agreed first, and flight timing and index choice follow.
- We process the imagery. Through our drone data processing service we produce NDVI maps, georeferenced orthophotos and, where useful, 3D surface models from multispectral imagery. We can also process imagery from your own drone.
- We deliver GIS-ready files. NDVI maps come as raster and vector layers and orthophotos as GeoTIFF, so they open directly in geographic information system software.
- We combine it with sensor data. We show the NDVI map on the same panel as soil moisture, pH and weather station readings from a precision agriculture IoT system, which makes it easier to tell drought stress from other causes. We explain how field sensors and devices are monitored from one centre in our guide to the remote monitoring IoT platform.
- We connect it to decisions. If you wish, we make the outputs trackable by parcel in a BI dashboard and report maps from different dates side by side.
Because the same team develops the sensors, the data transmission devices and the panel through an end-to-end IoT device development process, there is no gap in responsibility between the device in the field and the software at the centre. We work across all provinces of Türkiye, remotely and on site. For neighbouring topics from GIS to management dashboards, see our full Data & Analytics guide.
Next step
Pick one parcel first: ideally one whose yield fell short last season or where you already see clear differences within the field. Write down which decision you want to make better for it. Then get in touch and we will agree the flight timing, index and ground-truthing plan together for that crop and growth stage.