Model Card: Disease Risk Data Layer

Developer setup

• Authenticate with Cropwise – Learn how to authenticate your requests using your credentials.
• Model API Reference - WFS endpoint reference for the regional data layer (typeName=DiseasePrediction30km).

What it is

Overview

The maps generate the in-season prediction for major foliar diseases for wheat and barley across most European countries. They aim to show the favorable conditions by which the disease is most likely to occur based on weather conditions.

Quantifying potential crop damages and alerting on disease risks can help growers make profitable and sustainable crop protection decisions. However, the model does not attempt to replace field scouting and in-field disease observations. This is a plot-level, mechanistic hybrid model.

The Winter Cereals model addresses the full disease triangle: it considers susceptible host (cultivar resistance), pathogen presence (inoculum/historical data), and favorable environment (weather).

Inputs and Outputs

Inputs

At the batch generation level, the underlying model is executed per grid centroid with the following inputs:

• Grid centroid coordinates (latitude and longitude), derived from the 30 × 30 km standardized grid
• Crop (winter cereals, depending on geographic scope)
• Required inputs: planting date, tillage practices, previous crops, crop variety traits, in-season crop stage observations, soil characteristics, fungicide applications, irrigations, nitrogen inputs, past disease observations, and estimated yield potential

Outputs

Model outputs are returned at disease, plant organ, and plot level in these feature categories:

• Daily disease risk: In-season daily disease risk indicators relative to crop stage and developed plant organs. The unit is a 0–3 integer scale where:
  • 0 = low risk
  • 1 = medium risk
  • 2 = high risk
  • 3 = no risk

Note: The Regional AgInsights view only displays the daily disease risk level.

Other outputs include:

• Daily crop stages: Crop stage and plant organ apparition forecast
• Daily treatment efficacy: Fungicide treatment efficacy per disease
• Potential yield loss: Pre-season yield loss estimates based on historical records
• Disease level: Early-season disease risk indicators linked to past infestations and early conditions

How it works

Algorithm principles

The winter cereal disease models are mechanistic (hybrid) models, with a mechanistic structure trained and calibrated on extensive field data.

They simulate in-field foliar disease risk daily per leaf layer using three sub-models:

• Phenological model
  Simulates plant growth stage based on soil-crop-atmosphere interactions; predicts leaf appearance, expansion, and senescence

• Disease model
  Simulates infection leaf-by-leaf starting from BBCH30; models pathogen cycles (incubation → sporulation → infection). Disease development depends on weather, cultivar resistance, and leaf age

• Fungicide efficacy model
  Accounts for chemical mode of action and degradation over time under weather conditions

Generation (batch → grid)

Regional AgInsights generates predictions on a standardized 30 × 30 km grid by:

1. Constructing grid centroid coordinates
2. Executing the Winter Cereals Diseases API per centroid
3. Managing execution metadata (timestamps, versions, success rates)
4. Publishing results as geospatial layers

Serving (consumption)

Layers are served via OGC-compliant WFS endpoints. To obtain the Winter Cereal Diseases layer, use the following typeName:

• DiseasePrediction30km

Consumers can:

• Apply CQL filters
• Request JSON or CSV outputs

Authentication is handled via OAuth2 client credentials.

How to use

Intended use

• In-season disease management
  Timely alerts to mitigate disease risks and avoid unnecessary applications

• Growth stage planning
  Crop stage forecasts support planning of farming activities

How to interpret insights

These maps consider weather data (rainfall, temperature, etc.) and agronomic factors influencing disease development.

The model:

• Is independent from field-specific management practices
• Provides high-level risk indicators
• Should NOT be used for precise field timing decisions

Limitations

• Accounts only for favorable weather conditions
• Does not replace field scouting
• Optimal forecast window: current day to 7 days ahead

System information

Data sources

• Historical and forecasted weather data
• Field-level agronomic practices (planting, varieties, treatments, etc.)
• Historical disease observations and inoculum data

Spatial and temporal resolution

• Grid size: 30 × 30 km
• Temporal granularity: Daily, up to 365 days

Geographic Scope, Crops, and Availability

Maps should be used from March to end of May (fungicide season), or until fungicide applications are no longer relevant.

Filtering & Querying via WFS

The Disease Prediction layer is queryable via a WFS endpoint using CQL filters. Below is an example request and the available filter keys.

Example query:

curl --location '{BASE_URL}/regional-aginsights/v1/wfs?typeName=DiseasePrediction30km&outputFormat=json&cql_filter=country_code%3D%27FR%27' \
--header 'accept: application/json' \
--header 'Authorization: Bearer $ACCESS_TOKEN'

Note: String values must be wrapped in single quotes ('). When used in a URL, the filter must be URL-encoded (e.g., = → %3D, ' → %27).

These keys can be used in the cql_filter parameter to query the table for Disease Prediction:

Last update: 20/05/2026