How it decides

Every alert traces back to a published model and a dated infection event. This page names the science, so you — or your agronomist — can check it.

/01

Infection events, not scores

  • · There is no "risk score 0–100" anywhere in the product. Published epidemiological kernels — temperature × leaf-wetness response surfaces, of the kind published by Magarey and colleagues — produce dated, checkable infection events.
  • · Powdery mildew runs on the Gubler-Thomas risk index framework, adapted to the vineyard's own conditions.
  • · An event has a date, an evidence trail and a deadline — a sentence you can act on, or check against what actually happened.

/02

Phenology, not the calendar

  • · Growth stage comes from degree-day accumulation on the BBCH scale, not from the date.
  • · Season phase is derived from phenology, so both hemispheres work correctly.
  • · Every timestamp is read in the vineyard's own time zone.

/03

Biogeography

  • · Pests that don't occur at the vineyard's location are not assessed — and the app says so, and says why.
  • · "Not present here" is never rendered as "no risk".
  • · The grape moth model selects among four species by geography.

/04

Site adjustment

  • · Elevation, slope, aspect, cold-air drainage and soil drying all modify the assessment — two parcels a kilometre apart can get different sentences on the same night.

/05

Frost

  • · Severity comes from FAO critical-temperature tables, by tissue and growth stage — a frost that kills a shoot in April is a non-event in January.
  • · Radiative and advective frost are distinguished; a dew-point gate and cold-air drainage refine the night-by-night call.

/06

Winter chill

  • · Chill is accumulated in Dynamic Model chill portions — the standard the horticultural literature has converged on — and compared against the requirement of the varieties actually planted.
  • · Pruning-wound susceptibility decays on a published healing curve; rain forecast onto open wounds raises an alert.

/07

One engine, shared

  • · The identical model code runs in the app and on the server that sends push notifications — so a push can never disagree with what the screen shows.

/08

Community weighting

  • · Nearby confirmed reports feed the risk model, weighted by two factors: severity and distance — a severe report 2 km away matters more than a mild one 20 km away.
  • · All community data is anonymised.

The ten threats we model

Each has its own published model, run only where biogeography says it can occur.

Downy mildew

Plasmopara viticola

Favoured by warm, wet conditions. Oily leaf spots and white downy growth; the classic wetness–temperature infection event.

Powdery mildew

Erysiphe necator

Thrives warm and dry with moderate humidity. Modelled on the Gubler-Thomas risk index framework.

Botrytis

Botrytis cinerea

Grey mould of ripening berries. Favoured by prolonged wetness and tight clusters.

Black rot

Guignardia bidwellii

Brown leaf lesions and fruit mummification, driven by warm rain events.

Phomopsis

Phomopsis viticola

Cane and leaf spot. Early-season rain onto young shoots is the danger window.

Anthracnose

Elsinoë ampelina

Sunken lesions on shoots and berries after warm, heavy rain.

Sour rot

Yeast–bacteria complex

Late-season berry breakdown by yeasts and acetic bacteria, vectored by fruit flies near harvest.

Grape moth

Four species, by geography

Generational flights modelled by degree-days; the species is selected by where the vineyard is.

Pierce's disease

Xylella fastidiosa

Bacterial vine killer, assessed only where the pathogen and its vectors occur.

Trunk disease

Esca complex

Enters through pruning wounds. Managed through wound-susceptibility windows, not sprays after the fact.