228-5 Quantification of Leaf Pigments in Soybean Based on Wavelet Decomposition of Hyperspectral Features.



Tuesday, October 18, 2011
Henry Gonzalez Convention Center, Hall C, Street Level

Shardendu Singh1, Larry Purcell2, Jeffery D. Ray3, James R. Smith3 and Felix Fritschi4, (1)University of Missouri-Columbia, University of Missouri, Columbia, MO
(2)Crop, Soil, and Environmental Sciences, University of Arkansas, Fayetteville, AR
(3)USDA-ARS, Stoneville, MS
(4)Division of Plant Sciences, University of Missouri, Columbia, MO
The concentration of leaf pigments is a good indicator of a plant’s health due to wavelength-selective absorption of solar radiation. Canopy reflectance of 385 soybean genotypes was collected under field conditions at 55 and 105 days after planting over two years. The chlorophyll (Chl) a, Chl b and carotenoid concentrations were also determined during the same periods. Spectral features related to pigments were extracted before and after applying various wavelet transformations on the hyperspectral data. Several commonly used wavelet families such as Haar, Mexican hat, Mayer, Daubechies, Gaussian and Biorthogonal were tested using continuous and/or discrete transformation. The selected spectral features were used in stepwise regression to derive hyperspectral models for detection of leaf pigments. Spectral features varied between stages of plant growth and among the wavelet families. Results indicated higher predictive capability of the wavelet-transformed spectra irrespective of the number of predictors in the models. However, the differences between the predictability of untransformed and transformed data tend to be narrowed with an increase in the number of predictors. Accurate prediction of leaf pigments from spectral reflectance is important because it may allow non-destructive, rapid assessment of crop-nitrogen status under field conditions.
See more from this Division: C03 Crop Ecology, Management & Quality
See more from this Session: Oilseed and Fiber Crop Ecology, Management and Quality