Combining multispectral imagery with habitat specific elements to locate eastern hemlock

Eastern Hemlock (Tsuga Canadensis) fills an important role in the forests of the Eastern United States. This hemlock is threatened by the spread of an invasive species called the Hemlock Woolly Adelgid, and requires human assistance to survive this pest. The goal of this work is to help forest managers direct their management efforts by developing accessible methodology for locating Eastern Hemlock using readily available and affordable data. Leaf on, leaf off, and spring Landsat ETM+ imagery and a 10m Digital Elevation model are the foundation of the model, and several layers are derived from topographic data to provide information on hemlock habitat. The Classification and Regression Tree technique was chosen to unlock the complex relationships and interactions among the datasets that affect the growth and prosperity of hemlock. Ground reference plots collected by the Forestry Organization Remote Sensing Technology Project were used to train and assess the accuracy of models created by the Cubist data mining software.