Terra Intelligence
We don't farm by feel. We farm by data.
Predictive phenotyping across 50,000 individually numbered trees — soil chemistry, canopy health and harvest windows modelled six weeks ahead.
50,000
Trees Monitored
95–98%
Forecast Accuracy
6 wks
Forecast Horizon
3
Sensing Depths
How this module works
Terra Intelligence is the sensing layer of JUDE. Every block at Somanya is instrumented with soil probes reading pH, NPK and volumetric moisture at three depths, while a DJI Agras drone fleet flies multispectral passes on a fixed cadence.
Those readings feed a digital twin — a virtual replica of the plantation where each tree carries its own history: rootstock, irrigation events, nutrition programme, pest pressure and every harvest it has ever produced.
The yield model turns that history into a forecast. Buyers get committed volumes and harvest windows weeks before a single mango is picked, and agronomists get an alert the moment a block drifts outside its optimal envelope.
Data Flow
- 01CaptureSoil probes, weather stations and drone passes stream telemetry every 15 minutes.
- 02ModelReadings are fused with agronomic history in the digital twin and scored against optimal envelopes.
- 03DecideIrrigation, nutrition and pick-date recommendations are issued per block.
- 04CommitVerified forecasts release exportable volume into the trade portal.
Connects To
What it does, end to end
Soil & Nutrition Analytics
Continuous pH, NPK and moisture telemetry at 20cm, 50cm and 100cm, with automated fertigation recommendations per block.
Multispectral Drone Fleet
NDVI and thermal passes detect stress, disease and water deficit days before symptoms are visible to the eye.
Digital Farm Twin
A virtual replica of every tree — clone, age, canopy volume, yield history and live sensor state in one model.
Yield Forecasting
95–98% accuracy at six weeks out, feeding directly into export allocation and forward contracting.
Disease & Pest Prediction
Weather-coupled models flag anthracnose and fruit-fly risk windows with recommended intervention timing.
Harvest Window Optimiser
Brix, dry matter and firmness trajectories set the exact pick date for each destination market.
Yield & Harvest Window Simulator
Model a block's conditions and see how JUDE adjusts forecast yield, quality grade and pick date.
No active agronomic alerts
Forecast Yield
25.3 t/ha
Brix at Pick
18.9°
Pick Window
-3 days
Allocation Recommendation
Grade A — GOLD export
Model confidence 97% at six-week horizon.