Skip to main content
ARS Home » Research » Publications at this Location » Publication #118453
Title: DETECTION AND DELINIATION OF VARIOUS WEED SPECIES FROM SOYBEAN USING REMOTE SENSING

Author
item BUNTING, JEFFREY - UNIV OF ILLINOIS
item SPRAGUE, CHRISTY - UNIV OF ILLINOIS
item WAX, LOYD
item COPENHAVER, KENNETH - NASA REMOTE SENSING
item GRESS, TIMOTHY - NASA REMOTE SENSING
item VARNER, BENJAMIN - NASA REMOTE SENSING

Submitted to: North Central Weed Science Society US Proceedings
Publication Type: Abstract Only
Publication Acceptance Date: 1/8/2000
Publication Date: N/A
Citation: N/A

Interpretive Summary:

Technical Abstract: Monocultures of six broadleaf species and three grass species were grown in separate 6 m by 15 m plots planted within 76 cm row soybeans. Spectral signatures were collected from an airborne hyperspectral system and a handheld spectral radiometer mounted on an all-terrain vehicle at the University of Illinois Crop Sciences Research and Education Center in Urbana, Illinois. The objectives of this experiment were to (1) determine if remote sensed imagery can be used to delineate weeds mixed with soybeans, and (2) determine if remote sensed imagery can be used to differentiate grass from boradleaf species. Separability analysis from the radiometer signatures determined that shattercane, common waterhemp, and common lambsquarters could be delineated from soybeans in most of the visible spectrum (0.4 to 0.7 nm). However, giant foxtail could only be separated from soybeans in the green visible spectrum (0.5 to 0.6 nm). shattercane and common waterhemp were also separable from soybeans in the near-infrared spectrum. Broadleaf weeds mixed with soybeans could be statistically separated from grasses mixed with soybeans for a number of wavelengths in the visible and near-infrared collected using a handheld radiometer. Classification algorithms including spectral angle mapper, minimum distance, and maximum likelihood were performed on the hyperspectral imagery. The minimum distance classification provided the most accurate assessment in separating the individual weed species. While weed species may be statistically separable, it is more difficult to perform separation using classification techniques on airborne imagery.