Multispectral and hyperspectral analysis from remote sensing, a precision tool in development, can detect differences in crop health. This information should be coupled with observations on the ground for ground-truthing of arthropod populations and other plant stress factors. This allows the information seen in the image to be associated with biological factors in the field. Field experiments were conducted in 2003 and 2004 investigating the potential for remote sensing to detect early infestationsof cotton aphid (Aphis gossypii Koch) and spider mites (Tetranychus spp.) in the San Joaquin Valley, CA. Differential populations of aphids and mites were established in field plots using selective and disruptive pesticides. Hyperspectral and multispectral airplane imagery were collected in 2003; hyperspectral imagery using a portable spectrometer and multispectral satellite imagery was collected in 2004 in the plots. Treatments with mite damage were found to have lower average reflectance, with mite infestations above treatment threshold levels in 2003, using the green band in the multispectral imagery, although the actual numbers of mites in the field were quite low. In the narrow band at ~579nm, uninfested cotton was found to have higher average reflectance levels than mite-infested and aphid-infested cotton using the hyperspectral imagery in 2003. Additionally, the treatment with the highest mite number was found to have lower average reflectance than the other treatments at ~579nm, using the hyperspectral imagery in 2003. Using the portable spectrometer in 2004, it was found that average reflectance levels tended to decrease as aphid numbers increased. Additionally, using the multispectral satellite image in 2004, it was found that aphids at economic threshold levels could be detected using a canopy index and near infrared images.
展开▼
