Uptake Index
methodology.

Fertilizer applied outside an active root uptake window achieves nothing. It runs through the rhizosphere, leaches past the root zone, and ends up either in groundwater or bound to soil colloids that the tree cannot access. Decades of agronomic literature establish this, but the literature offers no practical method to detect when a specific tree is actively pulling nutrient from solution. Operators have historically scheduled fertigation by calendar, by canopy stage, or by composite weekly programs — none of which align reliably with the per-tree uptake reality.

The Uptake Index is a per-tree, per-night score computed from continuous in-soil sensor data. It identifies the time window during which a tree is actively drawing solution out of the root zone, and it scores the strength of that drawdown relative to the tree's own recent history. A high Uptake Index in the 22:00–04:00 window says the tree is hungry and is actively absorbing; a low Uptake Index says fertigation now will be wasted regardless of how textbook the recipe is.

Uptake Index is computed from three continuous sensor signals captured at the root zone of each instrumented tree:

• Electrical conductivity (EC) — measured in µS/cm. EC of the soil solution drops when ions are removed from solution, whether by root uptake, by precipitation, or by dilution with irrigation water. The drop signature differs by mechanism, which is why EC alone is not sufficient.
• Volumetric water content (moisture) — measured as a percentage. Required as a gating signal: if moisture is rising or unstable, the EC drop is most likely dilution from irrigation rather than uptake. If moisture is stable or falling slowly, the EC drop is more likely real uptake.
• pH — measured in pH units. Required as a second gating signal: an EC drop accompanied by a sharp pH excursion is usually a chemical artefact (fertigation event, soil amendment dissolution, microbial bloom) rather than uptake.

All three signals come from the same RS-485 Modbus in-soil sensor, sampled once per hour. The Uptake Index is therefore computed at one-hour resolution from 24 measurements per day per tree.

The Uptake Index is computed specifically over the nocturnal window — 21:00 to 05:59 local time. Three reasons:

During the day, transpiration dominates root-zone water dynamics. The tree pulls water out of the soil through the stomata at a rate that swamps any signal from active nutrient uptake. EC will drop because the soil is drying, not because the tree is selectively absorbing solute. The signal-to-noise ratio is poor.

At night, stomata close and transpiration collapses to near zero. Any continued reduction in root-zone EC is much more likely to reflect active solute uptake by the root system, sometimes called nocturnal phloem loading. The signal-to-noise ratio is much better in this window.

Empirically, the Uptake Index measured nocturnally correlates with downstream agronomic outcomes — flush vigour, fruit set, recovery from stress — far more cleanly than the same metric measured over a full 24-hour period. This is a result, not an assumption; it was validated by running the calculation both ways across multiple tree-seasons and observing which version predicted observable tree behaviour.

The computation has three stages: EC drawdown, moisture gating, and pH gating. Each stage produces a sub-score; the final Uptake Index is a function of the three, weighted to penalise gating failures more heavily than weak EC signals.

EC drawdown rate

Compute the slope of EC over the 21:00–05:59 window. A monotonically declining EC across the window produces a positive drawdown rate; a flat or rising EC produces zero or negative. The drawdown is normalised against the tree's own trailing-fourteen-day median nocturnal drawdown — this is the core of the per-tree-relative scoring. A tree pulling 200 µS/cm out of solution overnight is a strong absolute signal, but the more relevant question is whether this tree is drawing harder or softer than itself recently. A tree that has historically drawn down by 80 µS/cm per night and now draws down by 30 is reporting weakness, even though 30 is not weak in absolute terms.

Moisture stability gate

Compute the Pearson correlation between time and moisture across the nocturnal window. A correlation near zero indicates moisture is flat — the gate is open and the EC signal can be trusted as real uptake. A correlation with large magnitude in either direction indicates a dilution or drainage event during the window — the gate closes and the EC drawdown is treated as suspect. This catches the case where a delayed irrigation event in a neighbouring zone causes capillary spread into the root zone of the tree being scored.

pH gate

Compute the range of pH across the nocturnal window. A range under about 0.3 pH units indicates stable solution chemistry — the gate is open. A range above about 0.6 pH units indicates a chemical event — likely a residual fertigation pulse or buffer collapse — and the gate closes. Strong fertigation events typically swing pH by 0.8–1.5 units; this filter cleanly excludes the most common false-positive class.

The final Uptake Index is the EC drawdown sub-score multiplied by the product of the two gate values, where each gate value is 1.0 (open), 0.5 (partial), or 0.0 (closed).

The Uptake Index never compares a tree against an absolute threshold. It always compares the tree against itself over time. This design decision was deliberate and is foundational.

In-soil sensors of the class used here have meaningful absolute calibration error. A capacitive moisture sensor placed two centimetres further from a fine root will read several percent lower forever. A pH probe drifts over months. EC absolute readings are sensitive to the specific soil chemistry and to electrode aging. Any system that compares the absolute reading of one sensor against an absolute threshold inherits all of this error.

A delta-based system inherits almost none of it. Calibration error is constant within a given sensor; subtraction cancels it out. Sensor aging is slow enough that a fourteen-day trailing median substantially absorbs the drift before it becomes signal. The result is a per-tree score that remains comparable across seasons, across sensor replacements, and across orchards with different soil character — none of which is true of a threshold-based system.

The cost of this design is that the index requires a minimum history before it can be computed reliably. A tree with fewer than five to seven nights of data has no usable trailing median; the system reports the index as experimental and weights the operator-facing recommendation accordingly.

The primary use of the Uptake Index is timing — answering the question when should the next fertigation event for this tree begin. A fertigation event scheduled to coincide with the peak of the tree's recent nocturnal uptake window is meaningfully more effective than the same recipe delivered at noon. Empirically, the same nutrient mass delivered during the active window produces visibly stronger above-ground response than the same mass delivered outside it.

The secondary use is anomaly detection. A tree whose Uptake Index collapses by more than about 40% relative to its own fourteen-day median, while neighbouring trees on the same irrigation lateral continue to show their normal index, is reporting that something has changed at the tree level — root disease, lateral or fine-root damage, an emitter blockage, nematode pressure, or early vascular obstruction. The signal appears in the index typically days to weeks before any above-canopy symptom becomes visible.

The tertiary use is comparative agronomy. Tree-level Uptake Index values can be aggregated to zone level, variety level, or cultivar level. Over a full season, this exposes which varieties consistently show stronger uptake under the same input program — a piece of information that has obvious implications for replanting decisions.

The closest commercial systems on the market measure soil moisture at field level (one or two sensors representing several hectares) and surface either a single moisture trace or a binary irrigation recommendation. EC is sometimes measured but typically at the fertigation bench, in the supply line — that is, EC of the water going into the irrigation system, not EC of the soil solution at the root zone of a specific tree. This is a fundamentally different measurement.

Per-tree continuous in-soil EC sensing, combined with moisture and pH gating, computed nocturnally against a tree's own trailing median, is not a configuration any commercial precision-ag platform currently surfaces as a primary signal. The Uptake Index methodology is the result of running this configuration in production and discovering, empirically, which combinations of gating and normalisation produce a per-tree signal stable enough to act on.

The methodology is being developed under provisional patent protection. Public disclosure of the core approach is intentional; the substantive moat is the longitudinal per-tree dataset that the methodology has been validated against — not the formulation of the index itself.

The Uptake Index operates at per-tree resolution and answers the timing question — when is this tree absorbing. The complementary curve-shape irrigation logic operates at zone resolution and answers the volume question — how much should the next event deliver. The two layers are designed to work together; uptake timing without volume sizing is half the answer.

The sensor hardware that generates the underlying EC, moisture, and pH streams is documented at /open-hardware, with companion pages on sensor wiring and power design. For the broader system architecture, see /technical.