Detection of natural and stress-induced variability in reflectance spectra of apple trees using hyperspectral analysis

Early detection of biotic and abiotic stresses and subsequent steering of agricultural systems using hyperspectral sensors potentially could contribute to the pro-active treatment of production-limiting factors. Venturia inaequalis (apple scab) is an important biotic factor that can reduce yield in apple orchards. Previous hyperspectral research focused on (i) determining if Venturia inaequalis leaf infections could be differentiated from healthy leaves and (ii) investigating at which developmental stage Venturia inaequalis infection could be detected. Logistical regression and partial least squares discriminant analysis were used to select the hyperspectral bands that best define differences among treatments. It was clear that hyperspectral data provide the contiguous, high spectral resolution data that are needed to detect subtle changes in reflectance values between healthy and stressed vegetation. The research was extended to include tree-based modeling as an alternative classification method. Results suggested that good predictability could be achieved when classifying infected plants based on this supervised classification technique. It was concluded that the spectral domain around 1600 nm was best suited to discriminate between infected and non-infected leaves immediately after infection, while the visible spectral region became more important at a well-developed infection stage. Research was focused on young leaves, because of the decreased incidence of infection in older leaves, the so-called 'ontogenic resistance'. Additional research was performed to gain a better understanding of the processes occurring during the first days after leaf unfolding and to evaluate the natural spectral variability among leaves. An undisturbed 20-day growth profile was examined to assess variations in the reflectance spectra due to physiological changes at the different growth stages of the leaves. Results suggested that an accurate distinction could be made between different leaf developmental stages using the 570 nm, 1940 nm, and 1460 nm wavelengths, and the red edge inflection point. Based on these results and the outcome of some existing chlorophyll indices, it was concluded that the chlorophyll content in leaves increased remarkably during the first 20 days after unfolding.

[1]  Lalit Kumar,et al.  Imaging Spectrometry and Vegetation Science , 2001 .

[2]  Pablo J. Zarco-Tejada,et al.  Simple reflectance indices track heat and water stress-induced changes in steady-state chlorophyll fluorescence at the canopy scale , 2005 .

[3]  Pablo J. Zarco-Tejada,et al.  Natural and stress-induced effects on leaf spectral reflectance in Ontario species , 2000 .

[4]  P. Ralph,et al.  Changes in photosynthesis during leaf expansion in Corymbia gummifera , 2003 .

[5]  John R. Miller,et al.  Scaling-up and model inversion methods with narrowband optical indices for chlorophyll content estimation in closed forest canopies with hyperspectral data , 2001, IEEE Trans. Geosci. Remote. Sens..

[6]  Pol Coppin,et al.  DETECTION OF BIOTIC STRESS (VENTURIA INAEQUALIS) IN APPLE TREES USING HYPERSPECTRAL ANALYSIS , 2005 .

[7]  W. E. Machardy,et al.  Apple Scab: Biology, Epidemiology, and Management , 1996 .

[8]  J. Dungan,et al.  The effect of a red leaf pigment on the relationship between red edge and chlorophyll concentration , 1991 .

[9]  Roger Newson,et al.  Parameters behind "non-parametric" statistics: Kendall's ¿a, Somers' D and median dierences , 2001 .

[10]  Leo Breiman,et al.  Classification and Regression Trees , 1984 .

[11]  D. Roberts,et al.  Deriving Water Content of Chaparral Vegetation from AVIRIS Data , 2000 .

[12]  O. Mutanga Hyperspectral Remote Sensing of Tropical Grass Quality and Quantity , 2004 .

[13]  S. I. Pogosyan,et al.  Application of Reflectance Spectroscopy for Analysis of Higher Plant Pigments , 2003, Russian Journal of Plant Physiology.

[14]  Y. Lespinasse,et al.  A microscopic study of the different classes of symptoms coded by the Vf gene in apple for resistance to scab (Venturia inaequalis) , 1991 .

[15]  D. M. Moss,et al.  Red edge spectral measurements from sugar maple leaves , 1993 .

[16]  Pablo J. Zarco-Tejada,et al.  Hyperspectral Remote Sensing of Closed Forest Canopies: Estimation of Chlorophyll Fluorescence and Pigment Content , 2000 .

[17]  John A. Gamon,et al.  Assessing leaf pigment content and activity with a reflectometer , 1999 .