Skip to content

Capability

Yield prediction

Phoeniks runs a process-based crop-model ensemble — APSIM, DSSAT and ORYZA — over a property's soil, weather and remote-sensing history to produce a per-property yield estimate with an explicit confidence level. Wheat and rice process routes are implemented first; the crop catalogue defines what can be modelled where.

01

How an estimate is produced#

  • Crop catalogue and can-grow matrix determine which crops are physically plausible on the parcel.
  • Per-listing crop-fit signal ranks those candidates against the local climate and soil envelope.
  • The simulation ensemble runs the selected process routes over the property's history.
  • The scoring engine converts model output into a comparable score with a confidence band.
  • A scenario configuration manager records the assumptions used, so a run can be replayed.
02

Ensemble, not a single model#

Single-model yield figures are hard to defend in front of a model-risk function. Running three independent process models over the same inputs exposes disagreement between them, and that disagreement is reported rather than averaged away.

Frequently asked

Which crops are covered?
Wheat and rice process routes are the implemented core, with the crop catalogue and can-grow matrix defining the wider set of crops that can be assessed for suitability. Coverage varies by country.
Do you publish accuracy figures?
Validation metrics are computed under a documented holdout and calibration protocol and delivered in the validation pack. Phoeniks does not publish headline accuracy claims that a reviewer cannot reproduce.

Related pages

See also

Request access

Institutional access. No self-serve trial.

Request access