MONITORING SCOTCH PINE INFESTED BY SIREX NOCTILIO USING HYPERSPECTRAL DATA: A LABORATORY STUDY

摘要

Sirex noctilio is an invasive insect native to Europe and north Africa that has the potential to devastate North American softwoods. The current extent of Sirex infestations in the United States has been approximated using expensive, and marginally effective, ground-based surveys. This study is part of a project aimed at using remote sensing to perform individual tree-based assessment of Sirex infestation. Because the Sirex: toxin causes rapid chlorophyll breakdown, it is expected to create a spectral response in early infestations that is different from the response from other insects or disease. In this study, samples from healthy and infested Scotch pines were collected in the field and then hyperspectral data were acquired in the laboratory using a hand-held spectroradiometer. Two vegetation indices—i.e. red edge position (REP) and normalized difference water index (NDWI)—were investigated for indicating health status of Scotch pine based on the lab-measured hyperspectral data. Three different methods— i.e. four point interpolation, maximum first derivative of reflectance spectrum, and linear extrapolation—were used to calculate REPs of healthy and infested needle samples. Linear extrapolation presented its advantage in separating healthy from infested samples based on the results of paired t-tests. The optimal wavelength grouping for NDWI calculations was selected from all possible combinations of two narrow bands within the range of 350-2500 nm using paired Mests. Compared to REP, NDWI was superior in discriminating between healthy and Sirex-infested Scotch pine, therefore this index is recommended for future work.