Plant phenotyping · Updated November 2026
Plant phenotyping systems fall into two families. Physiological (gravimetric) systems, such as the Plant-DiTech PlantArray, weigh every pot continuously and calculate how much water each plant uses, minute by minute. Image-based systems photograph plants with colour (RGB), fluorescence, thermal or hyperspectral cameras and calculate traits from the pictures. Both are called “high-throughput phenotyping”, but they measure different things, answer different questions and run very differently in a lab or greenhouse.
This article compares them on the science (what each actually measures) and on day-to-day operation (space, set-up, data, staffing and cost drivers), so that research groups and purchase committees can choose the right tool, or the right combination.
The short answer
- Choose a gravimetric system such as PlantArray when the question is about water: transpiration, water-use efficiency, stomatal behaviour, drought, salinity or the point at which a plant starts to suffer from drying soil. It measures water flux directly and continuously, plant by plant.
- Choose imaging when the question is about form and colour: growth, leaf area, architecture, greenness, senescence, photosynthetic efficiency or disease symptoms, especially across many plants or in the field.
- Combine them for stress physiology programmes. Imaging shows what the plant looks like; gravimetry shows what the plant is doing with water at the same moment.
On this page
1. How each system works
Gravimetric physiological phenotyping (PlantArray)
Each pot sits on its own temperature-compensated load cell, a precision weighing scale. With the soil surface covered so that water leaves the pot only through the plant, any loss of weight over a short interval is water transpired by the plant; any gain over days is growth. Soil-moisture and atmospheric sensors are logged alongside, and each unit has its own valves, so irrigation and stress treatments can be applied pot by pot.
In the system described by Halperin and colleagues at the Hebrew University of Jerusalem (The Plant Journal, 2016), weights were read every 15 seconds and averaged every 3 minutes. From these readings the software derives whole-plant transpiration, biomass gain, stomatal conductance and root water flux, and identifies the critical soil water content: the point in a drying cycle at which transpiration begins to fall sharply.
Image-based phenotyping
Plants are imaged on a schedule, either moved to cameras on a conveyor or imaged in place by cameras on a gantry. Each camera type gives different traits:
| Imaging type | What it measures | Typical uses |
|---|---|---|
| Visible (RGB) | Projected leaf area, height, colour, canopy cover, architecture | Growth curves, greenness, senescence, seedling vigour |
| Chlorophyll fluorescence | Efficiency of photosystem II (Fv/Fm and related parameters) | Photosynthetic performance, early disease and herbicide effects |
| Thermal infrared | Leaf and canopy temperature | A proxy for transpiration and stomatal closure under water stress |
| Hyperspectral | Reflectance across many narrow wavebands | Pigments, water content, vegetation indices, disease severity |
| 3D (stereo, laser, time-of-flight) | Canopy structure, leaf angle, height, leaf area index | Architecture and canopy development |
Categories follow the review by Li, Zhang and Huang (Sensors, 2014).
2. Theoretical comparison: what is really being measured
The core difference is direct versus inferred measurement. A load cell measures water leaving the plant in grams. Most imaging traits are inferred from light reflected or emitted by the visible surface of the plant, and are then related to physiology through calibration.
| Question | Gravimetric (PlantArray) | Image-based systems |
|---|---|---|
| Nature of the measurement | Direct: change in pot weight, i.e. water lost or biomass gained | Indirect: pixels, temperature or spectra interpreted as plant traits |
| Main traits | Transpiration, transpiration rate, water-use efficiency, stomatal conductance (calculated), root water flux, biomass gain, critical soil water content | Leaf area, height, architecture, colour, Fv/Fm, canopy temperature, spectral indices |
| Time resolution | Continuous: readings every few minutes, day and night, so diurnal patterns are captured | Snapshots at each imaging time (often once or a few times a day); changes between images are not seen |
| Whole plant or surface | Whole plant, including the root zone’s water uptake | The visible surface; overlapping leaves and the root system are hidden from most cameras |
| Early stress detection | Responds as soon as transpiration changes, often before any visible symptom | Visible symptoms (wilting, colour) lag behind; thermal imaging responds earlier but depends on the environment |
| Sensitivity to light and weather | Weight is unaffected by light; temperature effect on the load cell is small and compensated (vendor figure: about 2 g per 10 °C) | Image quality depends on lighting, background contrast and calibration; thermal readings depend on air temperature, humidity and wind |
| Biomass | Measured as weight gain, normalised to plant size | Estimated from projected area or 3D volume; accuracy falls as canopies grow and leaves overlap |
| Physiological meaning | Links water use directly to soil water and atmospheric demand at each moment | Strong on structure and photosynthesis; water status is inferred |
Two examples from the literature show the difference in practice:
- Whole-plant water relations. Halperin et al. screened 65 tomato introgression lines on 96 weighing lysimeters and checked the results against conventional gas-exchange instruments, which gave similar absolute values on control plants. Gas exchange measures one leaf at a time and is, in the authors’ words, tedious and limited in the number of plants.
- Imaging under drought. Briglia et al. (Frontiers in Plant Science, 2020) imaged 45 grapevines under three water treatments, grown on a gravimetric platform of precision scales (±1 g) that controlled irrigation. RGB pixel counts predicted leaf area well (R² = 0.78), and 3D imaging tracked leaf angle (R² = 0.73 against water potential, versus 0.86 for manual measurement). Chlorophyll fluorescence (Fv/Fm) did not change under water restriction, showing that some imaging traits are insensitive to drought.
3. Operational comparison: running the system
| Factor | Gravimetric (PlantArray) | Image-based systems |
|---|---|---|
| Where it runs | Greenhouse, net house or growth room, on level benches or floor | Growth chamber, greenhouse or field, depending on the platform |
| Plant handling | Plants stay in place on their scales for the whole experiment | Plants move to the cameras (conveyor) or cameras move over plants (gantry); handheld and field options exist |
| Pot and plant format | One pot per unit; vendor sizes: small units for 1.5–8 L pots (up to 6 per m²) and large units for 10–40 L pots (up to 2 per m²); very small plants may need several per pot | Pots, trays or plots; small plants such as Arabidopsis suit top-view imaging well |
| Scale and throughput | Vendor minimum order is 12 units; arrays scale to hundreds or thousands of units, each running its own treatment | Hundreds to thousands of plants per day on automated platforms; throughput is limited by imaging time per plant |
| Site requirements | Level benches, drainage, 110/220 V power and a water supply (vendor figure: about 4 atm); covered pots to stop evaporation | Controlled lighting or shading for consistent images, space for conveyors or gantries, imaging cabinets |
| Treatments | Per-plant irrigation, drought, salinity and nutrient scenarios run automatically by soil water content | Treatments applied separately; the system observes their effects |
| Calibration | Load cells are factory calibrated and temperature compensated; reference units track evaporative demand | Colour, thermal and spectral calibration for each camera; trait models often calibrated against destructive harvests for each species |
| Data | Hundreds of numeric readings per plant per day, analysed in the vendor’s software (data export by CSV or API) | Large image files; image processing and segmentation pipelines; hyperspectral data volumes are especially high |
| Skills needed | Experiment design and physiology; little image-analysis expertise | Image analysis, segmentation and, increasingly, machine learning |
| Main cost drivers | Number of units, sensors and greenhouse preparation | Camera types (hyperspectral and 3D are the most expensive), automation and computing |
Pot sizes, unit density, minimum order and site requirements are Plant-DiTech’s published figures.
4. Advantages and limits of PlantArray
Advantages
- Direct, continuous water data for every plant, day and night, without touching the plant.
- Physiology, not just appearance: transpiration, water-use efficiency, stomatal conductance and the critical soil water content are calculated from the same readings.
- Early stress response: changes in transpiration are seen before visible symptoms.
- Treatments built in: each pot can follow its own drought, salinity or recovery scenario, so plants of different sizes can be stressed to the same soil water level.
- Precise: vendor figure of ±2 g on a 12 kg load cell, with the temperature effect compensated.
- Few moving parts: plants stay on their scales; one failed unit does not stop the rest.
Limits
- Pot-based: it measures plants in pots in a greenhouse or room, not field plots, so results need field validation. Halperin et al. position it for early-stage selection before field trials.
- One plant per unit: throughput scales with the number of load cells, and each needs bench space.
- No picture of the plant: it does not record shape, colour, leaf angle or disease symptoms unless imaging or other sensors are added.
- Interpretation: soil-drying experiments need careful design; the authors note they can be among the most difficult experiments to interpret.
- Site work: level benches, drainage and water pressure must be prepared before installation.
5. Advantages and limits of imaging
Advantages
- Rich structural and colour data: growth, leaf area, architecture, greenness and senescence from the same images.
- Photosynthesis and pigments: fluorescence and hyperspectral cameras see processes that weight cannot.
- Non-contact and scalable: image sensors have become cheaper and more robust, and the same methods extend to field plots, drones and handheld devices.
- A visual record: images can be re-analysed later with new algorithms.
Limits
- Indirect traits: water status, stomatal behaviour and biomass are inferred and need calibration for each species and set-up.
- Occlusion: overlapping leaves are hard to separate in images, and the root system is hidden.
- Snapshots: changes between imaging times are missed, including most of the daily course of transpiration.
- Environment and calibration: illumination, background and, for thermal imaging, air temperature and humidity affect the readings.
- Data handling: image processing needs specialist skills, and spectral images generate large data volumes.
- Cost: hyperspectral and 3D sensors are expensive, and automated conveyors or gantries add to the cost.
6. Using both together
The two families are complementary. Reviews of imaging recommend combining several sensors to tell different stresses apart, and Plant-DiTech’s software can take in imaging and other sensor data alongside the weight readings. A practical combination for a drought or salinity programme:
- PlantArray controls the stress per plant and records transpiration and water-use efficiency continuously.
- Periodic RGB or 3D imaging records growth and architecture on the same plants.
- Chlorophyll fluorescence (for example PSI FluorCam) or a leaf spectral meter checks photosynthetic performance at key stages.
- Results are matched by plant and time, giving both what the plant looks like and what it is doing with water.
JKG BioScience supplies both families: Plant-DiTech PlantArray for physiological phenotyping and PSI instruments for chlorophyll fluorescence and spectral measurement, together with plant growth chambers for controlled-environment work.
7. Which to choose
| Your main question | Best first choice |
|---|---|
| How much water does each line use, and how efficiently? | Gravimetric (PlantArray) |
| When and how fast do lines close their stomata as soil dries? | Gravimetric (PlantArray) |
| Screening for drought or salinity tolerance before field trials | Gravimetric, with imaging for growth |
| Growth curves, leaf area and architecture over time | RGB or 3D imaging |
| Photosynthetic efficiency, early disease or herbicide effects | Chlorophyll fluorescence imaging |
| Pigments, nutrient status or field plots | Hyperspectral or multispectral imaging |
| A full stress-physiology programme | Both, matched by plant and time |
Frequently asked questions
What does PlantArray measure that imaging cannot?
Continuous whole-plant water use: transpiration, transpiration rate, water-use efficiency, calculated stomatal conductance, root water flux and the critical soil water content at which transpiration starts to fall. Imaging can only infer water status, for example from canopy temperature.
Can PlantArray replace imaging?
Not for structural or colour traits. It does not record leaf area, architecture, colour or disease symptoms. For those, add imaging; PlantArray’s software can take in imaging and other sensor data.
Is gravimetric phenotyping suitable for field trials?
No. It works with plants in pots on load cells in a greenhouse, net house or growth room. It is designed for early-stage selection before larger field trials.
What pot sizes does PlantArray use?
According to Plant-DiTech, small units take pots of about 1.5–8 L and large units about 10–40 L. Very small plants may need several plants per pot.
Is imaging more affordable?
It depends on the cameras. Basic RGB imaging is relatively low cost; hyperspectral, 3D and fully automated conveyor systems are expensive. PlantArray cost scales mainly with the number of units.
Plan your phenotyping set-up
Tell us your crops, stresses and space, and we will recommend gravimetric, imaging or a combination, with installation and support in India.
Related pages
- Plant physiology and phenotyping equipment
- Plant-DiTech PlantArray in India
- PSI chlorophyll fluorescence and spectral instruments
- Plant phenotyping applications
- AI-ready plant research
- How to choose a plant growth chamber
References
- Halperin O, Gebremedhin A, Wallach R, Moshelion M. High-throughput physiological phenotyping and screening system for the characterization of plant–environment interactions. The Plant Journal (2016). doi:10.1111/tpj.13425
- Li L, Zhang Q, Huang D. A review of imaging techniques for plant phenotyping. Sensors 14(11): 20078–20111 (2014). doi:10.3390/s141120078
- Briglia N, Williams K, Wu D, et al. Image-based assessment of drought response in grapevines. Frontiers in Plant Science 11: 595 (2020). doi:10.3389/fpls.2020.00595
- Plant-DiTech. PlantArray product information and Q&A. plant-ditech.com