📈 Disease Forecasting Models

Disease forecasting uses meteorological data, pathogen biology, and host phenology to predict the likelihood and severity of disease outbreaks. Accurate forecasts allow farmers to apply fungicides only when necessary, reducing costs and environmental impact — a key principle of IPDM.

---

Principles of Disease Forecasting

Forecasting models are based on the observation that disease epidemics require specific environmental conditions. If those conditions are not met, disease will not develop regardless of inoculum presence. Models quantify the relationship between weather variables and infection events.

Types of Forecasting

Type	Description	Example
Calendar-based	Fixed spray schedule irrespective of weather	Traditional approach (not true forecasting)
Threshold-based	Spray triggered when specific weather criteria are met	Disease Severity Values (DSV)
Simulation models	Computer models simulate pathogen life cycle	EPIPRE, LATEBLIGHT
Risk maps	Geographic mapping of disease risk using GIS	Regional advisory systems

---

Major Forecasting Systems

Late Blight of Potato (Phytophthora infestans)

The Beaumont period rule (UK) states that late blight risk is high when temperature remains above 10 degrees C and relative humidity exceeds 75% for 48 consecutive hours. The improved Smith period uses minimum temperature above 10 degrees C with RH above 90% for at least 11 hours on two consecutive days.

Modern systems like BLITECAST (USA) and NegFry (Europe) combine temperature and rainfall data to calculate Disease Severity Values (DSV). Fungicide sprays are recommended when cumulative DSV reaches a threshold.

Apple Scab (Venturia inaequalis)

Mills' Table (1944) relates temperature and leaf wetness duration to ascospore infection. For example, at 15 degrees C, a minimum of 12 hours of continuous leaf wetness is required for light infection. Mills' criteria have been refined into computer-based systems like RIMpro and A-Scab.

Wheat Rusts

The Indian Wheat Rust Monitoring System uses trap nurseries across the Indo-Gangetic plains to track rust pathotype migration. Weather-based forecasts combine temperature, humidity, and wind direction data to predict epidemic initiation.

---

Components of a Forecasting Model

1. Weather monitoring — automated weather stations recording temperature, humidity, rainfall, leaf wetness, and wind
2. Biological parameters — spore trapping, disease surveys, inoculum quantification
3. Model algorithms — mathematical relationships between weather and infection probability
4. Decision rules — thresholds that trigger management actions
5. Communication — advisories delivered via SMS, mobile apps, radio, or extension services

---

---

Modern Advances

• Remote sensing — satellite and drone imagery detect disease symptoms at landscape scale
• IoT sensor networks — real-time microclimate data from in-field sensors
• Machine learning — AI models trained on historical disease and weather datasets improve prediction accuracy
• Mobile decision support — smartphone apps integrate weather APIs with forecasting algorithms for on-farm use

---

---

Benefits for IPDM

• Reduces unnecessary sprays — fungicides applied only when infection risk is high
• Improves timing — interventions target the critical infection window
• Lowers costs — fewer applications mean reduced input expenditure
• Minimizes resistance risk — fewer fungicide applications slow resistance evolution
• Supports advisory services — government and extension agencies disseminate region-specific recommendations

Disease forecasting transforms IPDM from a reactive to a proactive, precision-based approach.

---

---

Summary Cheat Sheet

Forecasting Basics

Element	Purpose
Weather thresholds	Detect likely infection windows
Disease severity metrics	Trigger action decisions
Advisory dissemination	Turn model output into field action

Classic Models to Remember

• Beaumont and Smith periods for potato late blight.
• Mills' table for apple scab infection risk.
• DSV-based tools like BLITECAST for timing fungicides.

Exam Traps

• Calendar spraying is not true forecasting.
• Forecast validity depends on local weather data quality.
• Prediction without communication channel has little field value.

References

2 sources • [1] [2]

[1]

Plant disease forecasting concepts and decision support systems

Book

[2]

BLITECAST and late blight forecasting resources

Website