Measure
A sensor at the base of a living plant records the electrical potential between the plant and the surrounding soil, continuously and on site. The sensor design is our own and is being refined in field trials.
Technology
Reading the electrical signal of a living plant
A plant maintains a measurable electrical potential against the soil it grows in. That potential has been documented and its mechanism published. VoltreeAI develops sensors that record it continuously, in the field, under the canopy, and proprietary, AI-driven algorithms that interpret it. Whether that interpretation adds useful information on plant water status or fuel moisture is being evaluated.
Forest floor, sensor at the base of a tree
The physical basis
A concentration cell, driven by pH
In 2008, researchers at MIT reported sustained differences of 50 to 200 millivolts between an intact tree and its soil, and ruled out a simple dissimilar-metal reaction as the cause.
They attributed the potential to a biological concentration cell driven by the pH difference between the tree and its soil. The research was funded by Voltree's founders.
Sustained voltage differences of 50 to 200 mV were recorded between the xylem of intact trees and the surrounding soil, with identical electrodes at both terminals.
PLoS ONE, 2008
Measure, transmit, interpret
Three steps, from plant to data
Transmit
Low-power nodes store readings on site and forward them to our data platform.
Network design for large sites: in development
Interpret
Proprietary, AI-driven algorithms interpret the signal in the context of local conditions, and their results are checked against established reference methods. Whether they add useful information on plant water status or fuel moisture is being evaluated. We do not present an interpretation as validated before it is.
Interpretation: in evaluation
Siting
Under canopy against open clearing
| Weather station in a clearing | VoltreeAI node | |
|---|---|---|
| Siting | Open ground, chosen for exposure and access | Under the canopy, at the base of a living tree |
| Measured quantity | Air temperature, humidity, wind, precipitation | Electrical potential between plant and soil |
| Relationship to fuel | Fuel moisture is estimated from a weather model | Measured at the living plant; relationship to fuel moisture under evaluation |
| Cadence | Interval reporting, typically hourly | Continuous sampling |
Comparison
Against the alternatives
| Method | What it measures | Under the canopy? | Character |
|---|---|---|---|
| Weather stations | Air temperature, humidity, wind, in open clearings | No | The reference network fire danger is modelled from |
| Manual fuel sampling | Gravimetric fuel moisture, the reference standard | Yes | Point measurements, labour-bound, not continuous |
| Satellite indices (NDVI and similar) | Canopy greenness, after it has changed | No | Broad coverage, coarse resolution, reports the consequence |
| Satellite fuel-moisture products | Canopy-top moisture index | No | Broad coverage, coarse resolution, multi-day timing |
| In-forest sensor networks | Air temperature, humidity, smoke and gas | Yes | Continuous, at node level |
| Voltree.ai | Electrical plant signal plus local conditions | Yes | Continuous at the plant; added value under evaluation |
We are not the only sensor under the canopy, and we do not claim a general advantage over satellite or NDVI monitoring. Any comparison depends on the crop, the signal, the observation frequency, the conditions and the reference method. We are testing whether the plant itself is an earlier instrument than the air around it.
Continuous potential trace, one node, twenty-four hours
Open questions
What we are still establishing
The existence of the potential is settled. Its behaviour as an operational signal is not. These are the questions our validation programme is built around.
That is an open question. Company-sponsored university work in 2016 and 2017 observed voltage changes before visible stress in a controlled study. Whether the signal adds information over existing methods in the field, whether any lead time appears, and how reliably across species, sites and seasons, are outcomes the validation programme will report, not figures we set in advance.
Quantitatively, across species and seasons, this is unresolved. It is a primary subject of our validation programme, and we do not present a correlation we have not demonstrated.
Disease, mechanical damage and drought all change the physiology behind the measurement. A reading can only be interpreted with knowledge of the plant it came from. A missing or unusual signal is not, on its own, evidence that a plant is dead or diseased.
We publish what we find, including results that do not support the hypothesis.