Sequential application of hyperspectral indices for delineation of stripe rust infection and nitrogen deficiency in wheat

• Published in: Precision agriculture 2015-10, Vol.16 (5), p.477-491
• Main Authors: Devadas, R, Lamb, D. W, Backhouse, D, Simpfendorfer, S
• Format: Article
• Language: English
• Subjects: Agriculture; Atmospheric Sciences; Biomedical and Life Sciences; canopy; Carotenoids; Cereal crops; Chemistry and Earth Sciences; Chlorophyll; chlorosis; color; Computer Science; Crop diseases; Crop science; Crops; developmental stages; Flowers & plants; Infections; Leaves; Life Sciences; Nitrogen; Nutrition; Physics; Physiology; Pigments; Plant diseases; plant pigments; Reflectance; Remote sensing; Remote Sensing/Photogrammetry; Sensors; Soil Science & Conservation; Statistics for Engineering; stripe rust; Studies; Triticum aestivum; Vegetation; vegetation index; wavelengths; Wheat
• ISSN: 1385-2256; 1573-1618
• DOI: 10.1007/s11119-015-9390-0

Nitrogen (N) fertilization is crucial for the growth and development of wheat crops, and yet increased use of N can also result in increased stripe rust severity. Stripe rust infection and N deficiency both cause changes in foliar physiological activity and reduction in plant pigments that result in chlorosis. Furthermore, stripe rust produce pustules on the leaf surface which similar to chlorotic regions have a yellow color. Quantifying the severity of each factor is critical for adopting appropriate management practices. Eleven widely-used vegetation indices, based on mathematic combinations of narrow-band optical reflectance measurements in the visible/near infrared wavelength range were evaluated for their ability to discriminate and quantify stripe rust severity and N deficiency in a rust-susceptible wheat variety (H45) under varying conditions of nitrogen status. The physiological reflectance index (PhRI) and leaf and canopy chlorophyll index (LCCI) provided the strongest correlation with levels of rust infection and N-deficiency, respectively. When PhRI and LCCI were used in a sequence, both N deficiency and rust infection levels were correctly classified in 82.5 and 55 % of the plots at Zadoks growth stage 47 and 75, respectively. In misclassified plots, an overestimation of N deficiency was accompanied by an underestimation of the rust infection level or vice versa. In 18 % of the plots, there was a tendency to underestimate the severity of stripe rust infection even though the N-deficiency level was correctly predicted. The contrasting responses of the PhRI and LCCI to stripe rust infection and N deficiency, respectively, and the relative insensitivity of these indices to the other parameter makes their use in combination suitable for quantifying levels of stripe rust infection and N deficiency in wheat crops under field conditions.