Date of Award

Fall 1994

Project Type


Program or Major

Natural Resources

Degree Name

Doctor of Philosophy

First Advisor

John Aber


Remote sensing of foliar chemistry has been recognized as an important element in producing large scale, spatially explicit estimates of forest ecosystem function. Visible and near infrared reflectance measurements of forest foliage, at both the leaf and canopy scales, were studied in an effort to relate spectral data to foliar chemical composition. At the scale of the individual leaf, near infrared spectra and chemical data were collected from 211 fresh leaf samples, including 17 different hardwood and conifer species. Reflectance data at selected near infrared bands was closely related to the nitrogen, lignin, and cellulose concentration of these leaf samples, with correlation coefficients greater than 0.83. In order to make large-scale regional estimates of foliar chemistry, this study was extended to whole forest canopies. Foliage and leaf litter were sampled on forty plots at Blackhawk Island, Wisconsin and Harvard Forest, Massachusetts to determine canopy level nitrogen and lignin concentrations. At the time of the field sampling, spectral measurements of the canopy were made with NASA's Airborne Visible/Infrared Imaging Spectrometer, a high spectral resolution instrument. Calibration equations were developed from these field and spectral data (R$\sp2 =.87$ and.77 for nitrogen and lignin, respectively). These equations were applied to all image pixels to make spatially explicit estimates of canopy nitrogen and lignin for both study sites. These estimates of nitrogen and lignin concentrations were then used with existing ecosystem models to predict carbon balance at Harvard Forest and nitrogen mineralization rates at Blackhawk Island. The chemical analysis of leaf samples collected from the study sites revealed that foliar nitrogen and lignin concentrations could be used to identify samples by species. The spectral bands used to measure nitrogen and lignin were then used to identify 11 different species classes at the Harvard Forest. A comparison of field measured species composition to the classification showed an overall accuracy of 73.4%. These results indicate that airborne and spaceborne imaging spectrometers will provide important information for the study of forest ecosystems.