A ten-cell matrix. One input per cell: what to measure, the alarms that matter, the sensor, where to put it, and how to keep it honest.
You met the ten inputs in Growing as science — what each one does to the plant. This page is the next layer: turning each input into a number you can trust, an alarm that fires while there's still time, and a probe that doesn't lie. It sits between two things you already have on this site. It does not re-explain why an input matters — that's the input's own page in Growing. It does not teach how a sensor works — that's the Sensors lesson in the Fundamentals. It specifies what to instrument, and how.
One rule sits above all the others. An uncalibrated probe is worse than no probe, because it lies with authority. A ten-thousand-dollar nutrient-dosing system run by a hundred-dollar pH probe that hasn't been calibrated in three months is a ten-thousand-dollar system making decisions on bad data. Calibration discipline — the keep it honest line in every cell below — is the spine of the whole system. A reading you can't trust isn't data; it's noise wearing a number's clothes.
The cells follow the site's two-zone framework: the five aerial inputs first, then the five root-zone inputs — the same order the operator screen uses, five across the top, five across the bottom.
Aerial zone
1 · Light — PPFD, DLI, photoperiod
Measure: photosynthetic light at canopy height — PPFD (the instantaneous intensity, µmol/m²/s) and DLI (the daily total, mol/m²/day). For short-day crops, the integrity of the dark period. Alarm when: the schedule says lights-on but PPFD reads near zero (fixture or sensor failure — immediate); any light leak above ~0.5 µmol during a cannabis dark period (critical — it disrupts flowering); the daily total falling short with most of the photoperiod gone (early enough to react). A sensor reading under 75% of its zone's average flags a fouled or aging lamp. Sensor: a quantum (PAR) sensor with a cosine-corrected diffuser; ±5% is the working accuracy. A periodic spectrometer check verifies spectrum — not a continuous job. Place it: level, at canopy height, above the canopy not below, raised as the plants grow. One per lighting zone minimum; a grid in high-value rooms catches uniformity loss early. Keep it honest: wipe the diffuser weekly (dust and spray residue read as "less light"); factory recalibration yearly. Couples to: leaf temperature (light heats the leaf), CO₂ (no light, no injection), and VPD (leaf temperature feeds the leaf-VPD estimate). → Why it matters: the Light cluster. · How sensors work: Fundamentals · Sensors. · Products: the Catalog.
2 · Air temperature
Measure: dry-bulb air temperature at canopy level; leaf-surface temperature where you can (an infrared spot reading); the day–night difference (DIF); and the rate of change during light transitions. Alarm when: temperature leaves the crop's band (for cannabis flower, roughly below 16 °C or above 32 °C is critical; lettuce and strawberry run cooler); the rate of change exceeds ~2 °C per minute at a transition (it collapses VPD); temperature falls toward the dew point (condensation risk). Sensor: a platinum RTD (PT100/PT1000) for accuracy and long-term stability (±0.1 °C). In any high-light room the sensor must sit in an aspirated — fan-ventilated — radiation shield, or the lamps heat the sensor body and inflate the reading. Place it: at canopy level, shielded and aspirated; a second point high in tall rooms catches the floor-to-ceiling gradient. Keep it honest: quarterly cross-check against a reference thermometer. Couples to: VPD (temperature is half the equation — every change moves it), CO₂ (photosynthetic efficiency shifts with heat), and disease risk (Botrytis near the dew point; powdery mildew at 20–25 °C with high humidity). → Why it matters: the Air Temperature cluster. · How sensors work: Fundamentals · Sensors. · Products: the Catalog.
3 · Humidity & VPD
Measure: relative humidity as the raw input — but VPD (vapor pressure deficit, kPa) as the number you steer by. Also the dew point, and the leaf-level VPD estimate. Alarm when: VPD leaves the band (for cannabis flower, below ~0.6 kPa is the disease zone, above ~1.6 kPa is the stress zone, above ~2.0 is severe); VPD falls faster than ~0.15 kPa/min at lights-off (the dehumidifier can't keep pace with the temperature drop); cumulative leaf-wetness passes two hours (Botrytis germinates around four). Sensor: a capacitive RH/temperature probe (±2% RH), co-located so one device gives both readings — again, aspirated in high light. Place it: at canopy height alongside the temperature sensor. Keep it honest: quarterly cross-check (saturated-salt solutions give a known RH reference). The rule that matters: relative humidity alone is almost useless for plant management. Any screen that shows only %RH is failing the operator. VPD is the primary number; RH is the secondary diagnostic. Couples to: everything aerial — it is temperature and humidity combined — and, through transpiration, calcium delivery down in the root zone. → Why it matters: the VPD cluster. · Calculate it: the Library (VPD). · How sensors work: Fundamentals · Sensors. · Products: the Catalog.
4 · Carbon dioxide
Measure: CO₂ concentration (ppm) at canopy height — plus a separate safety sensor at worker breathing height. Alarm when: concentration drifts off target (for cannabis flower, target ~1,000–1,200 ppm; low below 800, wasteful above 1,400); and the safety line that overrides everything: 5,000 ppm at breathing height trips a worker alarm and shuts injection off. The system must be physically incapable of reaching 40,000 ppm (the immediately-dangerous threshold) through any single fault. Sensor: an NDIR sensor — non-dispersive infrared — at ±30 ppm or better. Budget units at ±75 ppm are unusable when you're targeting 1,200, because the error band is wider than the response you're chasing. Place it: the cultivation sensor at canopy height, centered, away from injection points and doors; the safety sensor at head height. Keep it honest: semi-annual calibration against a known span gas; quarterly cleaning of the optical path. Couples to: light and temperature (the photosynthetic triangle — CO₂ only pays off with matching light and warmth), airflow (without it, leaf-level CO₂ runs hundreds of ppm below the room reading), and stomata/VPD (enrichment partly closes stomata, lowering transpiration). → Why it matters: the CO₂ cluster. · How sensors work: Fundamentals · Sensors. · Products: the Catalog.
5 · Airflow
Measure: air velocity at canopy level (m/s); fan status (running, speed, drive frequency); pressure drop across filters (it rises as they load with dust). Alarm when: any fan that should be running reads zero (immediate — a stalled fan in a sealed room is a fast emergency); canopy velocity under ~0.3 m/s (the boundary-layer warning — leaf conditions are diverging from your sensors); sustained velocity above ~1.0–1.5 m/s (mechanical stress, stomatal closure). Sensor: a hot-wire anemometer for accurate low-speed readings (used at commissioning and for spot checks); fan-drive feedback or a current sensor as a continuous proxy; a differential-pressure transmitter across filter banks. Place it: the commissioning grid measures every 2 m × 2 m cell at canopy height; a permanent canopy anemometer is worth it in high-value rooms. Keep it honest: re-run the airflow grid quarterly; for airflow, calibration is largely a commissioning exercise. Couples to: all the aerial inputs — airflow is what collapses the boundary layer and keeps the leaf living in the room's conditions instead of its own. Below 0.3 m/s, your temperature, CO₂, and VPD sensors are quietly lying. → Why it matters: the Airflow cluster. · How sensors work: Fundamentals · Sensors. · Products: the Catalog.
Root zone
6 · Water quality
Measure: at the source — EC (electrical conductivity, mS/cm), pH, alkalinity (ppm CaCO₃), sodium, chloride, iron, boron, and chloramine if you're on municipal water. After treatment — RO permeate EC and rejection rate. Alarm when: source EC climbs (above ~0.5 mS/cm needs action; above ~1.0 points to reverse osmosis); alkalinity above ~150 ppm (it will fight every pH correction you make); sodium above ~50 ppm; RO permeate EC rising past ~5% of the feed (the membrane is failing); filter pressure drop up more than ~15% (fouling). Sensor: inline EC on the RO permeate (continuous); periodic lab analysis for the full ion picture; an inline chlorine sensor if you use municipal water without RO. Place it: on the permeate line and the feed line; lab samples drawn from the source. Keep it honest: verify rejection rate on a schedule and trend it — don't spot-check it. Couples to: pH (alkalinity sets your acid demand), nutrition (source calcium, magnesium, and sodium come straight off your fertilizer targets), and your EC budget (source EC eats into the EC you have left for nutrients). → Why it matters: the Water cluster. · How sensors work: Fundamentals · Sensors. · Products: the Catalog.
7 · pH
Measure: solution pH — continuously inline in recirculating systems, by pour-through in substrate. In substrate, feed pH and root-zone pH are not the same number; track both. Alarm when: pH leaves the window (5.5–6.5 is the operating range; below 5.0 risks root damage, above 7.0 makes iron chemically unavailable even though it's physically present in solution); the calibration slope drifts more than ~15% from theoretical (the probe is dying); root-zone pH and feed pH diverge by more than ~0.5 (the substrate is doing something worth investigating). Sensor: a glass-electrode pH probe with a transmitter; ±0.1 is the floor, ±0.05 if it's driving automated dosing. Place it: inline in the recirculating stream; store the probe in KCl solution between uses — never dry, never in distilled water. Keep it honest: this is the make-or-break. Two-point calibration (pH 4.0 and 7.0 buffers) weekly for inline probes, before every session for handhelds. Single-point calibration is not enough — it checks offset but not slope. Replace inline probes every 6–12 months. Log the slope and offset every time. Couples to: the Clean Intervention idea — most acids add a nutrient you never meant to add (the common pH-down is phosphoric acid, which is phosphorus). A SCADA worth its name tracks the nutrient load your acid is quietly contributing; a non-mineral pH adjuster adds none, so there's nothing to track. → Why it matters: the pH cluster. · The principle: Clean Intervention. · How sensors work: Fundamentals · Sensors. · Products: the Catalog.
8 · Nutrition (EC)
Measure: EC continuously, on the feed line and the drain/return; individual ions by periodic lab analysis. EC reads total dissolved salts — it tells you how much, never which. Steering by EC alone is like managing a budget by total spending without ever categorizing a single expense. Alarm when: EC leaves the crop's band (cannabis flower ~1.4–2.2 mS/cm; lettuce ~0.8–1.4; tomato fruit much higher); the drain-to-feed ratio climbs (sodium or unused salts accumulating); EC or pH lands outside the expected range right after a recipe change (a mixing error caught immediately). Sensor: inline EC (conductivity) on feed and drain; lab analysis weekly in recirculating systems for the ion breakdown. Place it: feed line and return line both — the difference between them is your uptake. Keep it honest: monthly calibration against a known standard; weekly cleaning (salt builds up on the probe). Couples to: calcium delivery and VPD — calcium rides to the plant on transpiration, so when VPD drops and transpiration falls, calcium delivery falls even though the solution hasn't changed. A good system estimates delivered calcium from VPD and airflow, not just from its concentration in the tank. → Why it matters: the Nutrition cluster. · The principle: Clean Intervention (decoupled calcium). · Products: the Catalog.
9 · Dissolved oxygen
Measure: dissolved oxygen in the root zone (mg/L). This is the least-monitored root-zone variable in CEA, and the 4–6 mg/L band — where the plant quietly underperforms without visibly failing — is where a lot of undiagnosed trouble lives. Alarm when: DO drops into 4–6 mg/L (increase aeration); below 4 (critical — fermentation, Pythium pressure); below 2 (root death, hydrogen sulfide). In deep-water culture a power failure is roughly 2–4 hours from hypoxia — so aeration belongs on backup power, treated as life-safety, not as optional kit. Sensor: an optical (luminescent) DO sensor — no membrane to replace, minimal drift; preferred for CEA over the older electrochemical type. Place it: at the return line, after the solution has passed the roots (worst case) — not in the reservoir by the air stone (best case, and it flatters the number). Keep it honest: monthly verification against air-saturated water at a known temperature; monthly cleaning of biofilm. Couples to: root-zone temperature above all — warm water holds less oxygen and the roots consume it faster, so DO and temperature collapse together. Combined with days-since-clean, the two drive the Pythium risk. → Why it matters: the Dissolved Oxygen cluster. · How sensors work: Fundamentals · Sensors. · Products: the Catalog.
10 · Root zone temperature
Measure: solution or substrate temperature at the roots (°C) — independently from air temperature, on its own control loop. Alarm when: RZT leaves the band (cannabis ~18–20 °C optimal, critical above 28; strawberry runs cool, and a small substrate swings fast); RZT sits more than ~2 °C off setpoint for over 30 minutes (the chiller or heater is undersized). One advisory worth wiring in: if the air-temperature setpoint and the RZT setpoint are identical, flag it — the root zone probably isn't being managed independently. Sensor: a waterproof temperature probe (a DS18B20 is cheap and reliable; several can share a single wire). Place it: in the solution stream closest to the roots — not in the reservoir, where the delivery path adds heat before it reaches them. In substrate, mid-depth and mid-canopy. Keep it honest: verify against a reference; these are stable, low-maintenance sensors. Couples to: dissolved oxygen and uptake together — the triple effect: cooling the solution from 28 °C to 20 °C raises oxygen saturation by ~15%, cuts the roots' oxygen consumption by ~40–50%, and slows Pythium sharply, all from one intervention. → Why it matters: the Root Zone Temperature cluster. · How sensors work: Fundamentals · Sensors. · Products: the Catalog.