Mapping Broadleaf Plant Species and Chemical Constituents Using SEBASS

Spectral emissivity features in plants are due to structural elements and chemical constituents on leaf surfaces that can be unique for each species. This work evaluates the use of one-meter spatial resolution SEBASS imagery for recognizing chemical constituents and for species identification. SEBASS is an airborne thermal infrared (TIR) hyperspectral sensor with 256 bands between 2.44 and 13.55 µm.

SEBASS data collected over the State of Virginia Arboretum on 07/06/07 were calibrated to radiance and atmospherically corrected by using the In-Scene Atmospheric Compensation (ISAC) algorithm. An emissivity image was produced by dividing the corrected surface radiance image by the Planck function (28.5° C) that best fit the average radiance curves of the tree canopies. Laboratory directional hemispherical reflectance measurements of collected leaf samples were converted to emissivity using Kirchhoff's Law (e=1-R) and then were compiled into a spectral library. The emissivity image was analyzed by using this library in conjunction with the Matched Filtering module in the ENVI 4.2 software (Research Systems Inc.).

Of the 40 local species represented in the spectral library, 10 were not discerned in the imagery, and 30 were discerned with varying degrees of success. Species having relatively large leaves, high contrast spectral features, and/or well-developed planophile canopies, (e.g. Liriodendron tulipifera) gave the best matches. Confusion occurred when spectral features of leaves were similar, for example, between species rich in silica (e.g. Fagus grandifolia and Morus alba). Library spectra of lignin and cellulose matched with mulch piles, wooden structures, logs, lawns and pastures.

These preliminary results show that there is good potential for using TIR hyperspectral sensors to map plant species and various chemical constituents. Further work will examine possible effects of canopy structure on image spectra, and the use of other algorithms and techniques for improving atmospheric corrections and species mapping.