Determining forest species composition using high spectral resolution remote sensing data

Abstract

Airborne hyperspectral data were analyzed for the classification of 11 forest cover types, including pure and mixed stands of deciduous and conifer species. Selected bands from first difference reflectance spectra were used to determine cover type at the Harvard Forest using a maximum likelihood algorithm assigning all pixels in the image into one of the 11 categories. This approach combines species specific chemical characteristics and previously derived relationships between hyperspectral data and foliar chemistry. Field data utilized for validation of the classification included both a stand-level survey of stem diameter, and field measurements of plot level foliar biomass. A random selection of validation pixels yielded an overall classification accuracy of 75%.

Department

Natural Resources and the Environment

Publication Date

1-1-1998

Journal Title

Remote Sensing of Environment

Publisher

Elsevier Science Inc

Digital Object Identifier (DOI)

10.1016/S0034-4257(98)00035-2

Document Type

Article

Rights

© 1998 Elsevier Science Inc.

Share

COinS