Analysis of hyper-spectral AVIRIS image data over a mixed-conifer forest in Maine
An introduction to some of the potential uses of hyperspectral data for ecosystem analysis is presented. The examples given are derived from a digital dataset acquired over a sub-boreal forest in central Maine in 1990 by the NASA-JPL Airborne Visible and Infrared Imaging Spectrometer (AVIRIS) instrument gathers data from 400 to 2500 nm in 224 channels at bandwidths of approximately 10 nm. As a preview to the uses of the hyperspectral data, several products from this dataset were extracted. They range from the traditional false color composite made from simulated Thematic Mapper bands and the well known normalized difference vegetation index to much more exotic products such as fractions of vegetation, soil and shade based on linear spectral mixing models and estimates of the leaf water content at the landscape level derived using spectrum-matching techniques. Our research and that of many others indicates that the hyperspectral datasets carry much important information which is only beginning to be understood. This analysis gives an initial indication of the utility of hyperspectral data. Much work still remains to be done in algorithm development and in understanding the physics behind the complex information signal carried in the hyperspectral datasets. This work must be carried out to provide the fullest science support for high spectral resolution data to be acquired by many of the instruments to be launched as part of the Earth Observing System program in the mid-1990's.