Changes in spectral characteristics of rice canopy infested with brown planthopper and leaffolder
Journal
Crop Science
Journal Volume
47
Journal Issue
1
Start Page
329
End Page
335
ISSN
0011-183X
Date Issued
2007-01
Date
2007-01
Author(s)
Yang, Chwen-Ming
Cheng, Ching-Huan
Chen, Rong-Kuen
Abstract
Potted experiments of rice (Oryza sativa L.) plants were conducted to produce various scales of brown planthopper [Nilaparvata lugens (Stal), Homoptera:Delphacidae] and leaffolder [Cnaphalocrosis medinalis Guenee (Lep., Pyralidae)] infestations, respectively, for canopy hyperspectral reflectance measurements, and then to identify spectral characteristics (SCs) associated with insect infestations leading to the establishment of spectral models for severity assessment. By linear correlation intensity analysis, correlation coefficients (r) along the spectral domain of 350 to 2400 nm were determined and narrow bands related to infestation severity were selected as SCs. The reflectance at green light (490-560 nm) maximum (R-GREEN), red light (640-740 nm) minimum (R-RED), and near-infrared (740-1300 nm) peak (R-NIR) were also considered. For canopies infested with brown planthopper, r value at 426 nm was the highest (r=0.878**). Among the calculated spectral indices using two SCs, the determination coefficient of R-NIR/R-RED ratio was the highest (R-2-0.922, P < 0.001). For leaffolder infested canopies, the most negative r value located at 757 nm (r=0.613*) in active tillering stage but shifted to 445 nm (r=0.928**) in heading stage. The index R-NIR-R-RED in the active tillering increased R-2 value to 0.422 (P < 0.001), while no increase in R-2 was found in the examined SIs in heading stage. Models with more than two SCs yielded from multiple linear regression analysis exhibited a further improvement for discriminating infestation severity.
Publisher
Crop Science Society of America
Type
journal article