The fastest way to make a monitoring system irrelevant is to send too many alerts. When every small threshold crossing generates a notification, growers learn to swipe them away. Better alerts are selective: they explain what changed, why it matters and how confident the system is.
Thresholds are useful — but incomplete
A threshold is easy to understand and can be appropriate for hard limits. The problem begins when every variable uses rigid thresholds without considering duration, rate of change or crop context. A moisture value that briefly touches a limit is different from a sustained decline across several readings.
Persistence reduces noise
Before escalating an alert, a monitoring system can ask whether the condition persists. This simple idea helps filter communication glitches, transient irrigation effects and one-off spikes. Persistence should be tuned to the variable: pH drift, moisture stress and temperature excursions operate on different time scales.
Rate of change can be more important than position
Two zones can have the same current moisture but very different trajectories. A rapidly drying zone may deserve attention even before it reaches a fixed threshold. Likewise, a steadily rising EC baseline can be more informative than a single high point immediately after fertigation.
Context makes alerts explainable
An alert should answer three questions: what changed, what evidence supports it and what the next check or action is. Combining moisture, EC, pH, temperature and history can make a notification more meaningful than “sensor 3 out of range.”
Severity should reflect operational risk
Not every deviation is an emergency. Useful systems distinguish informational changes, review-needed conditions and genuinely urgent events. That hierarchy helps protect attention, which is one of the most limited resources in a working greenhouse.
Djonix philosophy
Djonix is designed to surface fewer, more explainable recommendations. The AI Agronomist can recommend, explain or abstain. The target is not maximum notification volume; it is a better signal-to-noise ratio for the grower.
An alert should earn the grower’s attention
- State what changed in plain language.
- Show the evidence and trend behind the alert.
- Use persistence so one noisy sample does not become an emergency.
- Distinguish information, review-needed and urgent conditions.
- Give a practical next check or action.
- Stop repeating the same alert unless severity or evidence changes.
Frequently asked questions
Why are threshold alerts noisy?
A threshold ignores duration and trajectory. A brief crossing caused by irrigation or sensor noise can trigger the same alarm as a sustained problem.
What is persistence in alerting?
Requiring a condition to remain present for an appropriate period or number of samples before escalation.
Should AI decide alert severity?
AI can help rank context, but severity rules for operationally critical conditions should remain transparent and testable.
Related reading
Sources & further reading
- Horticulturae 2026: supervised greenhouse decision-support review
- Agriculture 2026: smart greenhouses and intelligent systems review
External sources are provided for technical context. Crop targets and management decisions should be adapted to the crop, substrate, water source and local agronomic guidance.
See the root zone. Understand the change.
Djonix combines root-zone sensing, live monitoring, mobile alerts and 12 months of AI Agronomist in the Founding Edition.
Reserve Djonix · €400