Detection of downy mildew of opium poppy using high-resolution multi-spectral and thermal imagery acquired with an unmanned aerial vehicle

Downy mildew (DM) caused by the biotrophic obligate oomycete Peronospora arborescens (Berk.) is one of the most economically limiting diseases of opium poppy (Papaver somniferum L.) worldwide. The first symptoms appear as small chlorotic leaf lesions, which can evolve to curled and thickened tissues that become deformed and necrotic as the disease develops. The present study explored the use of high-resolution thermal and multi-spectral imagery as an indicator of DM infection. Work was conducted in two opium poppy field plots artificially infected by P. arborescens. Airborne thermal and multi-spectral imagery were acquired at 200 mm resolution on three dates in spring of 2009 using an unmanned aerial vehicle (UAV). Leaf reflectance and transmittance spectra of DM asymptomatic and symptomatic opium poppy leaves were measured using an integrating sphere. Simulation work was conducted with the coupled PROSPECT + SAILH radiative transfer model to assess the effects of the variability found in an opium poppy plot developing a DM epidemic on the normalized difference vegetation index (NDVI) and the green/red index (R550/R670) calculated from the multi-spectral imagery. The airborne flights enabled DM detection by using image-derived canopy temperature (Tc) normalized by air temperature (Tc − Ta) and the green/red index (R550/R670). Tmin for each grid unit was calculated to estimate pure-vegetation temperature removing background and soil effects. Tmin − Ta and R550/R670 were assessed as a function of aggregated NDVI clusters to compare asymptomatic and symptomatic plants normalized by similar growth levels. Results demonstrated that Tc − Ta and the R550/R670 index were related to physiological stress caused by DM infection. In addition, Tmin − Ta was found to decrease as the NDVI increased and symptomatic plants reached significantly higher (P < 0.05) temperatures for an NDVI ≥0.6. The R550/R670 index was positively correlated with the NDVI, showing significantly higher v