Linear unmixing of AVIRIS data to identify white ash in the Catskill Mountain region, New York

Identification of white ash and understanding its habitat has significance to understand the threat of a particular invasive species; the Emerald Ash Borer. This research uses a spectral linear unmixing classification of AVIRIS data to identify white ash from its five associated forest types including American beech, eastern hemlock, sugar maple, red maple and yellow birch. The contribution of this research is to generate the spectral identities in the region from 370 nm to 2500 nm for white ash and the other five forest types. These can be applied to distinguish white ash and produce the fraction map demonstrating the proportion of each endmember present within a subpixel area. The results indicate that the spectral identities can linear unmix the endmembers within a subpixel area and also that the estimated fraction values are generally close to field measured values. Future improvements can focus on adding more land cover types as endmembers and exploring alternate preprocessing approaches for atmospheric correction and geometric rectification.