Predicting Wheat Rust Outbreaks from Low-Cost Field Sensor Networks

Key findings

• ▸Sensor-based model flagged outbreak conditions 5-8 days before visual scouting confirmed infection.
• ▸Leaf-wetness duration was a stronger predictor than humidity alone across all 18 monitored plots.
• ▸Sensor hardware cost under ₹2,000 per unit, built from off-the-shelf components.

Abstract

Wheat rust outbreaks are typically identified through visual field scouting, often after infection has already spread. This study trained a predictive model on humidity and leaf-wetness readings from low-cost field sensors (under ₹2,000 per unit) across 18 wheat plots over two growing seasons, comparing outbreak timing against traditional visual scouting records. The model flagged conditions consistent with impending outbreaks five to eight days before visual symptoms were confirmed by field scouts, offering a meaningful early-warning window for fungicide timing decisions in resource-constrained settings.