pixel in the Harvard Forest AVIRIS image to estimate nitrogen concentration (Figure 2). 
Similarly, canopy lignin concentrations were predicted with equation 2 using four bands 
of the first difference absorption spectra: 
%Lignin — 33.36 — (0.048*822nm) + (0.106 *627nm) + (0.005 *756 nm) + (0.052 * 1660rcm) 
( 2 ) 
Absorption at 1660nm is related to absorption overtones of unsaturated or phenolic C-C 
bonds which are abundant in lignin molecules. The three shorter wavelengths used in this 
equation correspond to a region of high absorbance observed in the laboratory spectra 
of lignin. Figure 3 shows the relationship between field measured and AVIRIS predicted 
lignin concentrations. Figure 4 shows the AVIRIS predicted lignin concentration for each 
pixel in the Blackhawk Island scene. 
Previous research at Blackhawk Island has demonstrated a very strong (R 2 = .96, n = 
7,p < .001) relationship between canopy lignin concentration and annual net nitrogen 
mineralization, or nitrogen cycling [10]. This relationship has been used with remote 
sensing data from a low-elevation airborne platform to produce a verified map of nitrogen 
mineralization for Blackhawk Island [10]. A nearly identical map is generated from an 
image of estimated lignin concentrations from 1992 AVIRIS data (Figure 5). 
At the Harvard Forest, a simple model of monthly carbon balances driven largely by foliar 
nitrogen concentrations, has been validated against monthly carbon balance data obtained 
by eddy-correlation methods [11]. Applying this model to an image of foliar nitrogen 
concentrations at the Harvard Forest, yields an estimate of net ecosystem exchange of 
carbon for the entire research site (Figure 6). 
4 DISCUSSION 
These results demonstrate the potential for high resolution remote sensing to increase both 
the accuracy of spatially averaged estimates of carbon and nitrogen cycling in temperate 
forest ecosystems, and to increase the spatial detail of those estimates. Continuing work 
on this project will include the incorporation of additional canopy chemistry data for sites 
at which field and AVIRIS data are available, and efforts to derive additional ecosystem 
modeling variables from AVIRIS data. 
5 REFERENCES 
[1] K. H. Norris, R. F. Barnes, J. E. Moore, and J. S. Shenk. Predicting forage quaility 
by infrared reflectance spectroscopy. J. Animal Sci., 43:889-897, 1976. 
[2] C. A. Wessman, J. D. Aber, D. L. Peterson, and J. M. Melillo. Foliar analysis using 
near infrared reflectance spectroscopy. Can. J. For. Res., 18:6-11, 1988. 
203