Using insitu hyperspectral remote sensing to discriminate pest attacked pine forests in South Africa

Abstract

Insitu hyperspectral remote sensing was used to identify optimal spectral bands capable of discriminating pine trees that were attacked by the wood boring pest, 'Sirex noctilio'. The pest attacks all commercial pine species in South Africa and the symptoms on infected trees can be represented on a severity scale, as the green, red and grey stages of attack The objective of this study was to determine whether there is a significant difference between the mean reflectance (%) at each measured spectral bands (from 400 –1300 nm) for the green, red and grey stages of attack. Next, for the bands that were significantly different (P<0.001) in this spectral region, we sought to test whether some bands had more discriminating power than others by using the Jeffries - Matusita distance analysis technique. Using a field spectrometer, ninety reflectance measurements were obtained from several infected Pinus patula trees in Kwazulu-Natal, South Africa. Results indicate that spectral bands located in the visible portion (350 – 700 nm) and some spectral bands in the red edge (670-737 nm) of the electromagnetic spectrum could spectrally discriminate the different levels of S. noctilio attack. Although no single band is capable of total separability, results of the Jeffries Matusita (J-M) analysis indicate that an acceptable separability of 99.22% (J-M value of 1403) for all attack classes was reached when using a four band combination comprising of bands located at 500 nm, 521 nm, 685 nm, and 760 nm. The results encourage canopy scale detection and mapping of S.noctilio attack in pine forest plantations using airborne or spaceborne hyperspectral sensors