Field research was conducted to determine the potential of hyperspectral remote sensing for discriminating plots of soybean intermixed with pitted morningglory and weed-free soybean with similar and different proportions of vegetation ground cover. Hyperspectral data were collected using a handheld spectroradiometer when pitted morningglory was in the cotyledon to two-leaf, two- to four-leaf, and four- to six-leaf growth stages. Synthesized reflectance measurements containing equal and unequal proportions of reflectance from vegetation were obtained, and seven 50-nm spectral bands (one ultraviolet, two visible, and four near-infrared) derived from each hyperspectral reflectance measurement were used as discrimination variables to differentiate weed-free soybean and soybean intermixed with pitted morningglory. Discrimination accuracy was 93 to 100% regardless of pitted morningglory growth stage and whether equal or unequal proportions of reflectance from vegetation existed in weed-free soybean and soybean intermixed with pitted morningglory. Discrimination accuracy was 88 to 98% when using the discriminant model developed for one experiment to discriminate soybean intermixed with pitted morningglory and weed-free soybean plots of the other experiment. Reflectance in the near-infrared spectrum was higher for weed-free soybean compared with soybean intermixed with pitted morningglory, and this difference affected the ability to discriminate weed-free soybean from soybean intermixed with pitted morningglory.
Mapping Ground Cover Using Hyperspectral Remote Sensing after the 2003 Simi and Old Wildfires in Southern California
