Identification of red and NIR spectral regions and vegetative indices for discrimination of cotton nitrogen stress and growth stage

Authors:
D. H. Zhao;J. L. Li;J. G. Qi
Affiliations:
Department of Biological Science and Technology, Nanjing University, Nanjing 210093, China;Department of Biological Science and Technology, Nanjing University, Nanjing 210093, China;Center for Global Change and Earth Observations, Michigan State University, East Lansing, MI 48823, USA
Venue:
Computers and Electronics in Agriculture
Year:
2005

Citing 0
Cited 5

Selection of the most efficient wavelength bands for discriminating weeds from crop

Computers and Electronics in Agriculture
Design of a hyperspectral nitrogen sensing system for orange leaves

Computers and Electronics in Agriculture
Assessing nutritional status of Festuca arundinacea by monitoring photosynthetic pigments from hyperspectral data

Computers and Electronics in Agriculture
Comparative analysis of three chemometric techniques for the spectroradiometric assessment of canopy chlorophyll content in winter wheat

Computers and Electronics in Agriculture
Fusion of remotely sensed data from airborne and ground-based sensors to enhance detection of cotton plants

Computers and Electronics in Agriculture

Quantified Score

Hi-index	0.00

Visualization

Abstract

Studies have demonstrated the value of spectral vegetation indices (VIs) based on red and near-infrared (NIR) spectral reflectance in agriculture. The objective of this research was to analyze the effects of central wavelengths and bandwidths of red and NIR bands on VIs, and to evaluate the potential of red-NIR VIs in discriminating cotton canopies by nitrogen stress and growth stage. A completely randomized experiment was conducted in a cotton (Gossypium hirsutum L. cv. Sumian 3) field treated with four nitrogen application rates: 0%, 50%, 100% and 200% of the recommended rate. Hyperspectral reflectance was measured at 2.3m above the cotton canopy on July 15, August 14 and October 1 using a FieldSpec^(R) FR spectroradiometer. Using one-way analysis of variance (ANOVA) for 150x150=22,500 combinations of wavelengths and bandwidths in the normalized difference vegetation index (@l"2-@l"1)/(@l"1+@l"2), results suggested that the proper central wavelengths of @l"1 and @l"2 were at 680-730nm (not 640-660nm, the central wavelengths of the red channels of most multi-spectral sensors on the current generation satellites) and 750-850nm, respectively. However, the effect of bandwidth on VIs was complicated. A single VI was not enough for nitrogen stress and growth stage detection. A single VI, in the 96 VIs used in this paper, was able to correctly classify 30-45% of the samples by nitrogen rate and growth stage. Using the 96 VIs as independent variables, a canonical discriminant analysis resulted in an accuracy of 62.4% with a six-VI model by the stepwise procedure.