Page Banner

United States Department of Agriculture

Agricultural Research Service


item Radhakrishnan, Jayakumar
item Teasdale, John
item Mcmurtry, John
item Daughtry, Craig
item Liang, Shunlin
item Shuey, Chad

Submitted to: Meeting Abstract
Publication Type: Abstract Only
Publication Acceptance Date: 2/12/2001
Publication Date: N/A
Citation: N/A

Interpretive Summary:

Technical Abstract: Spectral weed detection techniques to distinguish between weeds and crops will contribute to weed management, especially in precision farming situations. This can be achieved if the differences between the spectral reflectance in certain wavelength regions of crop and weed spectra can be recognized. These differences are influenced by temporal variations during gthe growing season, such as cell size, chemical composition, water concentration and other factors that effect the characteristics of the crops and weeds. The physical environment such as directional reflection, shading, and spatial variation also need to be taken into account as they also influence spectral signals. The objectives of this study were to investigate 1) spectral separability of agronmically important weeds, 2) if temporal profiles of fluorescence and reflectance can aid in distinction of weed species, and 3) development of non-parametric models to estimate biochemical and biophysical variables of weed canopies. Spectral reflectance measurements of the canopy and soil in the field, leaf area index, leaf reflectance and fluorescence, chlorophyll and moisture content were obtained on corn and agronomically important weed species planted and replicated for this purpose. Two approaches to analysis are being conducted. Discriminate analysis is a statistical tool which assesses from a set of attributes those combiantions that are most characteristic of a species. These combinations, referred to as "functions", are used to determine which characteristics discriminate between two or more weed/crop species. Biological, chemical and physical properties of different features on the ground control the reflectance signal. A non-parametric model is being developed to make use of the above study for identification.

Last Modified: 10/19/2017
Footer Content Back to Top of Page