Field NotesComparisonsNo. 09
Plant sensors compared: what each one measures and how accurate it is
A datasheet tells you how good the chip is on the bench. Placement, calibration and the light source decide how good your reading is. Here are the numbers for each sensor type.

The short answer Most plant sensors measure the plant’s surroundings, not the plant. On datasheets, a good humidity chip is rated about ±1.8% RH (typical), a lux chip about ±10%, a research PAR sensor ±5% (calibration uncertainty), and a research soil probe ±3 points of water content (typical, generic calibration, mineral soil). Placement, calibration and the light source can cost far more accuracy than that.
A monitor’s spec sheet rarely says which part does the measuring, or how well. The part’s datasheet does, with Sensirion’s caveat: its specs “hold for the stand-alone sensing element,” not the finished device built around it.
Air temperature and humidity
A humidity sensor reports relative humidity (% RH) and temperature (°C) at its sensing element. Sensirion’s SHT40 is rated ±1.8% RH typical at 25 °C and ±0.2 °C typical.
- Self-heating. A device’s own electronics warm its sensor. Sensirion’s SCD4x CO2 module ships with a default 4 °C offset on its built-in temperature sensor to correct for that.
- Sun and grow lights. Keep the sensor at canopy height but shaded; outdoors, use a ventilated radiation shield. METER makes the same point about soil probes: temperature readings from a sensor on the soil surface are “influenced by absorbed solar radiation.”
- Airflow. The 4-second humidity response (to 63% of a step change) is measured in air moving at 1 m/s; in a finished device, it depends on the design.
- High humidity. Long spells above 80% RH can offset readings temporarily (+3% RH after 60 hours, in Sensirion’s example), and condensation leaves the sensor unresponsive until it dries.
Light: lux vs. PAR sensors
For lux, PPFD and DLI basics, see the light guide.
Lux sensors are filtered to see like your eye. TI’s OPT3001 “tightly matches the photopic response of the human eye,” peaking at 550 nm. For a 2,000 lux white-LED input, its datasheet output range is 1,800 to 2,200 lux, about ±10%. TI’s 4% light-source variation covers incandescent, halogen and fluorescent light only, for the bare chip. Lux is eye-weighted, so equal lux under two lights can mean very different photon counts.
Quantum sensors count photons from 400 to 700 nm, weighted equally, and report PPFD in µmol/m²/s. Apogee’s SQ-500 has ±5% calibration uncertainty and drift under 2% a year. Mount it level; Apogee suggests recalibration every two years.
Spectral mismatch catches quantum sensors too. Apogee’s older SQ-100 series is off by 72.8% under an 80% red, 20% blue LED mix; the SQ-500 stays within 3.9% under every source Apogee lists. Under LEDs, ask for spectral error data.
Soil moisture
Soil probes estimate volumetric water content (VWC), the volume of water per volume of soil, in m³/m³. METER’s TEROS 12, a research-grade capacitance probe, reads dielectric permittivity (1 for air, 80 for water) at 70 MHz. Its generic calibration is ±0.03 m³/m³ typical in mineral soils with solution EC below 8,000 µS/cm; a medium-specific calibration gets ±0.01 to 0.02 m³/m³.
Potting mix isn’t mineral soil: METER warns that dielectric sensors tend to read low in organic, low-density media, and lists a separate soilless-media calibration for the TEROS 12. Air gaps reduce accuracy too; METER expects a 2 to 3% underestimate with the sensor body on the soil surface.
The capacitive vs. resistive note covers low-cost probes and TDR, and how to calibrate a soil moisture sensor shows how to fit one to your mix.
Soil EC, pH and soil temperature
EC (electrical conductivity) is how readily soil or a solution carries current, in µS/cm (1,000 µS/cm = 1 dS/m). The TEROS 12 reads bulk EC, meaning soil, water and air together, to ±(5% + 10 µS/cm) up to 10,000 µS/cm, and also models pore-water EC. A meter dipped in runoff or nutrient solution reads the EC of that liquid, a third number. Compare like with like.
pH takes its own probe; none of the sensors here read it. The EC and pH note covers both.
Soil temperature is reported by the TEROS 12 to ±0.5 °C from 0 to 60 °C. METER notes that a sensor on the surface picks up absorbed sunlight and is “not necessarily representative of the soil temperature.” Insert it fully, at root depth.
CO2
Sensirion’s SCD4x measures CO2 in parts per million (ppm) using photoacoustic NDIR (non-dispersive infrared), an optical method. At 25 °C, 50% RH and 1013 mbar, the SCD40 is rated ±(50 ppm + 5% of reading) from 400 to 2,000 ppm. Its typical response time is 60 seconds to 63% of a step change.
Example: at 1,200 ppm, the band is ±110 ppm (50 plus 5% of 1,200), or 1,090 to 1,310 ppm.
CO2 matters most in enclosed grow spaces, which is where its assumptions break:
- Self-calibration is on by default and assumes the sensor sees 400 ppm outdoor air for more than 3 minutes at least once a week. An enriched or rarely vented tent may never provide that. The sensor lets you switch it off and recalibrate against a known reference.
- Altitude. It assumes sea level unless you set the altitude or ambient pressure.
Leaf temperature by infrared
An infrared thermometer reads thermal radiation from a surface without touching it. Melexis’s MLX90614 is rated ±0.5 °C from 0 to 50 °C, but only “under settled isothermal conditions” and only if the target fills the field of view. Depending on version, that field runs from 90° down to 5°, and the spot grows with distance:
spot diameter = 2 × d × tan(FOV ÷ 2)
Here d is the distance from sensor to leaf and FOV is the field-of-view angle. Example: at 10 cm, a 90° sensor reads a circle 20 cm across (tan 45° = 1), wider than most houseplant leaves, so soil and pot count too. A 5° version reads about 0.9 cm. Melexis defines field of view at 50% of the signal, so leave margin.
Emissivity is factory-set to 1.0 and adjustable from 0.1 to 1.0; low-emissivity surfaces reflect infrared from their surroundings. Let a handheld unit settle after moving rooms; Melexis warns that thermal gradients, which cause reading errors, “are likely to appear during transient conditions.” The water stress note covers reading leaf temperature against air temperature.
Plant sensors at a glance
| Sensor | Measures | Unit | Datasheet accuracy (example part) | Main pitfall |
|---|---|---|---|---|
| Air temperature and humidity | Temperature and RH | °C, % RH | ±0.2 °C and ±1.8% RH typical, RH at 25 °C (Sensirion SHT40) | Self-heating, sun, condensation |
| Illuminance | Eye-weighted visible light | lux | 1,800 to 2,200 lux (min to max) at a 2,000 lux white-LED input, about ±10% (TI OPT3001) | Misjudges grow LEDs |
| PPFD | Photons, 400 to 700 nm | µmol/m²/s | ±5% calibration uncertainty (Apogee SQ-500) | Spectral error under LEDs |
| Soil moisture | Dielectric permittivity, as VWC | m³/m³ | ±0.03 typical, generic calibration, mineral soil, solution EC below 8,000 µS/cm (METER TEROS 12) | Air gaps; organic mixes read low |
| Bulk soil EC | Conductivity of soil, water and air | µS/cm | ±(5% + 10 µS/cm), 0 to 10,000 µS/cm (TEROS 12) | Bulk, pore-water and runoff EC differ |
| Soil temperature | Temperature at the probe | °C | ±0.5 °C, 0 to 60 °C (TEROS 12) | Sun on a surface probe |
| CO2 | Concentration, photoacoustic NDIR | ppm | ±(50 ppm + 5% of reading), 400 to 2,000 ppm, at 25 °C, 50% RH, 1013 mbar (Sensirion SCD40) | Self-calibration needs weekly fresh air |
| Leaf temperature | Infrared from the surface | °C | ±0.5 °C, 0 to 50 °C (Melexis MLX90614) | Spot wider than the leaf |
What no sensor sees, and why trends win
Apart from leaf temperature, everything above measures the plant’s surroundings. The plant shows its own state in new leaves, droop and color; the sensors tell you what it’s responding to.
The datasheets also make the case for logging, because repeatability is far tighter than accuracy: ±10 ppm against ±(50 ppm + 5% of reading) for the SCD40, under 0.5% against ±5% for the SQ-500. Drift is slow: typically under 0.2% RH a year for the SHT4x in its normal range (5 to 60 °C, 20 to 80% RH). So one sensor compared with itself, before and after watering or week to week, tells you more than one reading against a chart.
Logging much faster than a sensor responds mostly adds rows, not information. Log often enough to catch lights, watering and heating cycles, and leave the sensor in place: a move shows up as a step change unrelated to the plant.
Sources
- Datasheet SHT4x: 4th Gen. Relative Humidity and Temperature SensorSensirion (datasheet version 6.5, April 2024)
- OPT3001 Ambient Light Sensor (ALS)Texas Instruments (datasheet SBOS681C)
- Owner's Manual: Quantum Sensor, Model SQ-500Apogee Instruments
- TEROS 12METER Group
- SCD4x: Breaking the size barrier in optical CO2 sensingSensirion (datasheet version 1.7, April 2025)
- MLX90614 Single and dual zone IR sensor in TO39Melexis (datasheet)
General information, not professional advice for a specific plant or crop. Spot something wrong? Email info@terrathink.io.


