PS 85-213 - Using hyperspectral data to detect insect pests in an agricultural system

Friday, August 11, 2017
Exhibit Hall, Oregon Convention Center
Ganesh P. Bhattarai and Brian McCornack, Entomology, Kansas State University, Manhattan, KS

Remotely sensed spectral data are increasingly used in ecological and agricultural researches. Wide range of sensors, from color and infra-red to multi- and hyper-spectral, mounted on various platforms are effectively used in studying vegetation properties such as species identity, leaf chemistry, plant stress, and disease. However, these approaches are still underutilized in detecting insect pests and their spread in natural and agricultural systems.

In this study, we evaluated reflectance spectra of wheat leaves infested by different species of aphids. In a greenhouse experiment, two-month old wheat plants were subjected to feeding by three species of aphid (Russian wheat aphid, bird cherry oat aphid, and greenbug). Aphid colonies were initiated on an individually caged plant with two (for low-density treatment) or five (high-density) adult aphids. A handheld leaf spectrometer was used to collect hyperspectral data (400 – 1000 nm in wavelengths) from each experimental plant in every second day until the plants began to senesce. Using these data, we calculated several photosynthetic pigments (chlorophyll a and b, carotenoids, anthocyanin, and phytochromes), and vegetation indices (NDVI and leaf water content) for each plant for each measurement day.


Reflectance spectra differed between control and aphid infested plants. Moreover, major principal components representing photosynthetic pigments and vegetation indices differed significantly among plants infested by different aphid species. Leaf reflectance started to differ significantly between control and infested plants in two weeks when the aphid colony size reached about 180 aphids per infested plant. These results suggest that physiochemical changes associated with aphid infestation can be detected using hyperspectral sensors. Furthermore, these pigment and vegetation indices could be used to detect pest infestation and their spread in the field using various aerial platforms.