Near Real Time Monitoring of Potato Late Blight Disease Severity using Field Based Hyperspectral Observation

Abstract

Field spectroscopic study was performed using a hand-held Field Spectroradiometer to examine spectral variability between healthy and late blight disease infected potato canopy. The purpose was to select a suitable spectral feature in the visible-near infrared region for late blight disease initiation and severity during winter season. The spectral regions at which significant differences in bio-optical response was observed between healthy and diseased canopy include, 680-730nm (47.84%), 750-900nm (76.14%) and 860-1040nm (68.60%). Correlation study was carried out among different spectral variables, sensitive to PLB disease, and field measured PLB disease severity. Out of all the spectral variables, Red-Edge Normalised Difference Vegetation Index (NDVI705) and Disease Water Stress Index (DWSI) showed significant negative correlation (0.87 and 0.84 at 95% confidence level) and capable to predict the different level of potato late blight disease severity. A remote sensing based novel disease severity method was developed from the above mentioned spectral variables using multi-linear regression model and validated with very high correlation (R2 = 0.883). The scoring method developed could be a good proximal indicator for real-time field monitoring of potato late blight disease.