A research team has developed a wheat powdery mildew index (WPMI) that enables accurate detection and monitoring of a destructive fungal disease across leaf, ground canopy, and UAV scales. The study, published in the Journal of Remote Sensing, integrates hyperspectral remote sensing with hot-spot analysis to identify infection clusters and track disease dynamics in smallholder wheat fields.
Wheat powdery mildew (WPM) caused by Blumeria graminis f. sp. tritici is a major threat to winter wheat production, capable of causing severe yield losses. Traditional diagnosis relies on visual inspection, which is subjective and labor-intensive. While hyperspectral remote sensing has shown promise, existing vegetation indices are often not disease-specific. The new WPMI addresses this gap by targeting disease-sensitive bands in the green, red, and near-infrared regions.
The team, from China Agricultural University, the Beijing Academy of Agriculture and Forestry Sciences, and the Chinese Academy of Agricultural Sciences, reported their findings in the Journal of Remote Sensing (DOI: 10.34133/remotesensing.0955). They developed two index forms: WPMIG = (R760 − R554)/(R661 + R554) and WPMIR = (R760 − R661)/(R661 + R554). These indices consistently distinguished healthy from infected wheat and quantified disease index across scales.
Data were collected from greenhouse and field experiments (2022-2024), including 1,260 leaf spectra and 804 canopy spectra. At the leaf scale, WPMI achieved classification accuracy up to 86% in greenhouse conditions and 81% in field conditions. For disease severity estimation, WPMIG reached R² values of 0.55 to 0.93 at the ground scale and 0.48 to 0.90 at the UAV scale.
UAV-derived WPMIG maps, combined with Getis–Ord Gᵢ* hot-spot analysis, identified clusters of infection and tracked spatiotemporal changes across smallholder plots over three growing seasons. The researchers noted that a disease-specific spectral index moves crop disease monitoring beyond simple image comparison, revealing where disease is emerging, expanding, or declining.
Leaf spectra were collected using a handheld hyperspectral camera, while canopy spectra used a ground spectrometer and a DJI M600 UAV with a Pika L hyperspectral camera. Linear discriminant analysis selected sensitive bands, and disease index was measured via field surveys. Linear regression assessed the relationship between WPMI and disease index.
With further validation, WPMI-based UAV monitoring could support precision plant protection and early warning systems for wheat production. This approach may help farmers identify disease hot spots before severe outbreaks, reduce pesticide use, and improve decision-making. The strategy also provides a framework for developing disease-specific remote sensing indices for other crop–pathogen systems.
