USGS Spectroscopy Lab

http://speclab.cr.usgs.gov

Investigating a physical basis for spectroscopic estimates of leaf nitrogen concentration

Raymond F. Kokaly

U.S. Geological Survey
MS 973, Box 25046
Denver Federal Center
Denver, CO 80225
(303) 236-1359
(303) 236-3200 FAX

raymond@speclab.cr.usgs.gov

This on-line version was submitted to Remote Sensing of Environment March 20, 2000
Final version accepted with minor revisions to Remote Sensing of Environment July 10, 2000

ABSTRACT
 

The reflectance spectra of dried and ground plant foliage are examined for changes directly due to increasing nitrogen concentration. A broadening of the 2.1 µm absorption feature is observed as nitrogen concentration increases. The broadening is shown to arise from two absorptions at 2.054 µm and 2.172 µm. The wavelength positions of these absorptions coincide with the absorption characteristics of the nitrogen-containing amide bonds in proteins. The observed presence of these absorption features in the reflectance spectra of dried foliage is suggested to form a physical basis for high correlations established by stepwise multiple linear regression (SMLR) techniques between the reflectance of dry plant samples with their nitrogen concentration. The consistent change in the 2.1 µm absorption feature as nitrogen increases and the offset position of protein absorptions compared to those of other plant components together indicate that a generally applicable algorithm may be developed for spectroscopic estimates of nitrogen concentration from the reflectance spectra of dried plant foliage samples.

INTRODUCTION
 

Estimates of biochemical concentrations in plants have been made using statistical regression methods applied to the spectral reflectance of dry foliage measured in the laboratory. However, a strong physical basis for high correlations between nitrogen concentration and the near-infrared reflectance spectra of natural vegetation has not been fully developed. This research on estimation of plant biochemical concentrations has been motivated by studies which linked nitrogen, lignin and cellulose concentrations to canopy biophysical processes, for example nutrient cycling (Aber and Federer, 1992 and Stuedler et al., 1989). Ultimately, in order to study canopy level biophysical and biochemical processes, estimates of canopy biochemistry using remotely sensed spectral measurements are desired. Several studies have shown that high correlation between canopy reflectance and nitrogen concentration can be achieved using stepwise multiple linear regression (SMLR) (Martin and Aber 1997, LaCapra et al 1996, Gastellu-Etchegorry et al 1995, Johnson et al 1994, Matson et al 1994, Wessman et al 1989). In each study, different wavelengths were found to produce the high correlations. Martin and Aber (1997) and LaCapra et al (1996) demonstrated that equations for estimating nitrogen derived from one site were unable to predict the nitrogen concentrations for other sites. The suggested reasons for this are many, including: inaccurate atmospheric correction, variability in the magnitudes of nitrogen levels between data sets, canopy architecture effects, sensor limitations, and background vegetation and soil influences. However, Grossman et al (1996) found similar inconsistencies in wavelength selection and inter-site nitrogen predictions at the leaf level and, furthermore, that the band selection by SMLR does not appear to be based on the absorption characteristics of the chemical being examined. Thus, many issues remain to be examined in developing a robust remote sensing method of estimating canopy chemistry, including: 1) are there observable influences of variable nitrogen concentration in the reflectance spectra of plants and 2) can nitrogen predictions be made across leaf level data sets.
 

The second issue of successful cross-site predictions of nitrogen concentration may be addressed by examining the history of past laboratory studies. Estimation of plant biochemical concentrations with spectroscopy began with laboratory analysis of dried and ground samples of animal forage (Norris et al., 1976 and Marten et al., 1989). These methods stressed the importance of controlled laboratory methods for reducing noise levels and the limited application of regression equations to samples of the same type used in calibration (Marten et al., 1989). Such methods have been extended to less controlled data sets of forest foliage (Kokaly and Clark, 1999, Bolster et al., 1996, Johnson and Billow, 1996, and McLellan et al., 1991a). In these studies, reflectance (R), log(1/R), and calculated derivatives of log(1/R) at a variety of wavelength positions in the near-infrared were highly correlated with biochemical concentrations. Recently, Kokaly and Clark (1999) presented a method for estimating biochemical concentrations using an analysis of the absorption features present in reflectance spectra of dried vegetation. Nitrogen predictions between data sets were made with low root mean square error (0.17% N by dry weight). This method showed a strong correlation (R2 = 0.95) between nitrogen concentration and the continuum removed and normalized band depths at five locations in the broad absorption feature at 2.1 µm for 840 samples of evergreen needles, deciduous leaves and other plant foliage. Interpretation of the wavelengths selected by regression analyses has been difficult because the many leaf components have absorption features which overlap in the near-infrared wavelength region. However, early research did point to the fact that protein has absorptions in the near infrared at 2.054 and 2.172 µm as justification for the use of near-infrared spectra for estimations (Davies and Grant, 1988 and Murray, 1988). The recent research on the near-infrared spectra of forest foliage has focused more on the statistical methods to achieve high regressions and less on the physical basis of such high correlations.
 

This paper partly addresses the first issue regarding remotely sensing canopy chemistry by examining reflectance spectra of dried and ground foliage in order to explore a physical basis for high correlations between near-infrared reflectance spectra and the nitrogen concentration of dried plant foliage. Selected plant spectra were examined to define changes in absorption features associated with increases in nitrogen concentration. These changes were examined for consistency across species and ranges in nitrogen concentration. Next, an attempt was made to link the reflectance changes with the absorption features of nitrogen-containing biochemical components of plants. Specifically, the wavelength positions of protein absorption features were compared to the changes in reflectance spectra that occurred as nitrogen concentration increased. The spectra of dried and ground leaves were examined to alleviate the strong influence of leaf water content on the reflectance spectra of fresh leaves in the near-infrared wavelength region.
 

METHODS
 

Spectral Feature Analysis
 

In order to compare the shapes of the absorption features between samples, this study uses a method of normalization called continuum removal. Continuum removal, or baseline normalization, is a method commonly used in laboratory infrared spectroscopy (Ingle, 1988). Clark and Roush (1984) discussed the application of this method to remotely-sensed reflectance spectra. Continuum removal is a numerical method to estimate the absorptions not due to the band of interest and remove their effects (Clark and Roush, 1984 and Clark, 1999). The continuum removal may be made by using linear segments; however, more rigorous treatment might be done with gaussian band shapes (Clark and Roush, 1984). Figure 1 shows the effect of continuum removal for the major absorption features in dried plant samples. The initial step is calculating the equation for a line between the continuum endpoints of the reflectance data (Figure 1a). Subsequently, the reflectance value for each point in the absorption band is divided by the reflectance level of the continuum line at the corresponding wavelength to establish the continuum removed spectral features in Figure 1b.
 


Figure 1a at 72dpi.

Figure 1b at 72dpi.

Figure 1a at 150dpi
Figure 1a at 300dpi

Figure 1b at 150dpi
Figure 1b at 300dpi

Figure 1. The reflectance spectrum of a sample of dried and ground white pine needles (Figure 1a) and the continuum lines for three major absorption features in a dry leaf. The isolated continuum removed reflectance for the three absorption features after the continuum removal calculation (Figure 1b).



 

Continuum removed absorption features can be compared by scaling them to the same depth at the band center, thus, allowing a comparison of the shapes of absorption features (see Kokaly and Clark, 1999). Figure 2 shows a comparison of band-depth normalized and continuum-removed shapes of the 2.1 µm absorption feature for two samples of dried and ground foliage. In this study, the short and long wavelength continuum endpoints of the 2.1 µm absorption feature were established at 2.01 µm and 2.222 µm, respectively. These endpoints correspond to the approximate minimum in absorption on either side of the 2.1 µm absorption feature. In Figure 2, the shape of the spectrum of the deciduous leaf sample of a basswood tree is broader than that of the rice sample. The challenge is to determine the cause(s) of the shape difference. Is the shape change due to variations in biochemical content, cell structure, or other properties of the samples?
 


Figure 2 at 72dpi.

Figure 2 at 150dpi
Figure 2 at 300dpi

Figure 2. The continuum removed reflectance of the 2.1 µm absorption feature showing the disparate absorption characteristics for two samples of dried and ground foliage: rice (dashed line) and basswood (solid line).



 

NASA ACCP Data Set

The NASA ACCP data set was compiled by an interdisciplinary team to investigate the feasibility of making estimates of canopy biochemical composition from remote sensing observations (ACCP, 1994). The full data set consists of plant materials collected from several sites: three eastern U.S. forests, a slash pine plantation, rice fields, and Douglas-fir seedlings grown in a greenhouse. More than 30 deciduous and coniferous tree species were represented by the 809 samples. All data except for the Douglas-fir were analyzed by Newman et al. (1994) for nitrogen concentration (presented in % dry weight) using a Perkin-Elmer CHN Elemental Analyzer. Lignin and cellulose concentration were determined by a modified wood-products chemistry procedure (see Newman et al., 1994). The Douglas-fir samples were analyzed for nitrogen by Johnson and Billow (1996) with a Perstorp Analytical RFA/2 continuous flow auto-analyzer.
 

Reflectance data (R) used in this study were converted from log(1/R) data for the specimens gathered by the NASA Accelerated Canopy Chemistry Program (ACCP, 1994). Prior to analysis, all samples were oven dried at 70o C for 48 hours, ground through a 1 mm mesh, and homogenized by Newman et al. (1994) as part of the NASA ACCP. Spectral reflectance was measured with a NIRSystems(1) Model 6250 scanning monochromator with spinning sample cup module by Bolster et al. (1996). Reflectance data were gathered over the wavelength range from 1.100-2.498 µm at a 2 nm sampling interval with a 10 nm bandpass. For Douglas-fir samples, Johnson and Billow (1996) measured reflectance with a NIRSystems Model 6500 scanning monochromator over the wavelength range 0.400-2.498 µm at the same sampling interval and bandpass as the NIRSystems 6250 instrument.
 

Sorted Samples from the NASA ACCP Data Set

To evaluate changes in spectral features observed in the reflectance of dry vegetation in terms of changes in nitrogen concentration only, the entire NASA ACCP data set of chemical concentrations was sorted to define a subset of samples which vary in nitrogen concentrations but not lignin and cellulose concentrations. Table 1 shows three selected pairs of samples and their associated biochemical concentrations resulting from this sort. The results include same-species and different-species matches in lignin and cellulose concentration. The first match, the same-species pair of Douglas-Fir samples, has a large difference in the concentration of nitrogen (2.06% N). The Douglas-Fir fertilization experiments were designed to reduce differences in other leaf biochemical constituents and measurements of specific leaf area suggest large differences in carbon accumulation were avoided in the fertilized seedlings (Dungan et al, 1996). However, the lignin and cellulose concentrations were not determined leaving an ambiguity as to whether differences in reflectance spectra may be caused by changes in these unmeasured leaf components. The second match, the same-species pair of slash pine samples, has a moderate variation in nitrogen concentration (0.87% N) and extremely close concentrations of lignin and cellulose. The final comparison, the interspecies pair of Norway spruce and basswood, has an extreme difference in nitrogen concentration (2.22% N) and variations in lignin and cellulose of under 0.5%. There are many other leaf components in addition to lignin and cellulose; thus, it is possible that any changes in reflectance may arise from these unmeasured components instead of nitrogen. In order to directly link shape changes in the 2.1 µm absorption feature to nitrogen, this study assessed whether the absorption features of nitrogen-containing compounds are in the proper wavelength position to cause the shape changes.
 

RESULTS
 

Figure 3a shows the scaled continuum removed 2.1 µm absorption feature of the two Douglas-fir samples (DF). The 2.1 µm absorption feature changes from a relatively narrow feature in the low nitrogen sample (df_sample14) to a more broad absorption feature of the higher nitrogen sample (df_sample77). Figure 3b shows the ratio of the two features which enhances the differences. The ratio of the continuum removed and scaled absorption features suggest that the broadening arises from absorptions centered at 2.054 µm and 2.172 µm.
 


Figure 3a at 72dpi.

Figure 3b at 72dpi.

Figure 3a at 150dpi
Figure 3a at 300dpi

Figure 3b at 150dpi
Figure 3b at 300dpi

Figure 3. The continuum removed reflectance (Figure 3a) of the 2.1 µm absorption feature for two dried and ground Douglas-fir samples: a low nitrogen sample (dashed line) and a high nitrogen sample (solid line). The ratio of these two continuum removed reflectance spectra are shown in Figure 3b indicating that absorptions at 2.054 µm and 2.172 µm cause a broadening of the 2.1 µm absorption feature as nitrogen concentration increases.



 

Figure 4a shows the change in the 2.1 µm absorption feature as the nitrogen concentration increases for the Slash Pine subset (SP). The spectral changes of the SP pair (Figure 4b) are similar to the DF pair as the nitrogen concentration increases (Figure 3b). The magnitude of the change is less than that observed between the Douglas-fir samples. Finally, Figure 5a shows the change in the 2.1 µm absorption feature as the nitrogen concentration increases for the interspecies pair of Norway Spruce-Basswood samples (NS-BW). Again, as concentration of nitrogen increases from a low value of 1.29% N in the Norway Spruce sample to the high value of 3.51% in the Basswood sample, the corresponding spectra show a broadening of the 2.1 µm absorption feature. The ratios of all three pairs (Figures 3b, 4b, and 5b) show the broadened absorption feature to arise from absorptions centered near 2.054 µm and 2.172 µm
 


Figure 4a at 72dpi.

Figure 4b at 72dpi.

Figure 4a at 150dpi
Figure 4a at 300dpi

Figure 4b at 150dpi
Figure 4b at 300dpi

Figure 4. The continuum removed reflectance (Figure 4a) of the 2.1 µm absorption feature for two dried and ground slash pine samples: a low nitrogen sample (dashed line) and a relatively higher nitrogen sample (solid line). The ratio of these two continuum removed reflectance spectra are shown in Figure 4b.


Figure 5a at 72dpi.

Figure 5b at 72dpi.

Figure 5a at 150dpi
Figure 5a at 300dpi

Figure 5b at 150dpi
Figure 5b at 300dpi

Figure 5. The continuum removed reflectance (Figure 5a) of the 2.1 µm absorption feature for two samples: a low nitrogen sample of dried and ground Norway spruce needles (dashed line) and a high nitrogen sample of dried and ground basswood leaves (solid line). The ratio of these two continuum removed reflectance spectra are shown in Figure 5b again indicating that absorptions at 2.054 µm and 2.172 µm cause a broadening of the 2.1 µm absorption feature as nitrogen concentration increases.



 

DISCUSSION
 

Relating Protein Absorption Features to Reflectance Spectra of Dried and Ground Foliage
 

The results presented in this study show that the broadening of the 2.1 µm absorption feature in dry plant spectra arise from two absorption features centered at 2.054 µm and 2.172 µm. To determine if these observed absorptions may be related to nitrogen containing compounds in the plants the wavelength positions of the observed absorptions were compared to wavelength positions of protein absorptions. Protein is the major nitrogen containing biochemical in plants. Davies and Grant (1988) present the spectra of various proteins. Elvidge (1990) shows the reflectance spectrum of the protein D-ribulose 1-5-diphosphate carboxylase (rubisco), an important plant protein with a major role in photosynthesis. According to Elvidge (1990), this protein accounts for 30 to 50 percent of the nitrogen in green leaves and is the most abundant protein in chloroplasts and perhaps the most plentiful protein on the planet. The protein spectra presented in Davies and Grant (1988) and the Elvidge (1990) rubisco spectrum all show absorptions centered near 2.054 µm and 2.172 µm of varying intensity. Thus, the wavelength positions of these persistent protein absorptions match those of the absorption features related to the broadening of the 2.1 µm absorption feature of the dry leaves as nitrogen concentration increases from low to high concentration.
 

To firmly attribute the changes observed in the shape of the 2.1 µm absorption feature to an increase in the concentration of nitrogen, the protein absorptions at 2.054 µm and 2.172 µm must be shown to arise from bonds containing nitrogen. In the near infrared region of the electromagnetic spectrum, absorption features are commonly the result of overtones and combinations of fundamental absorptions at longer wavelengths (i.e., in the middle infrared region). The wavelength positions of the infrared absorptions of the polypeptide linkages in proteins are listed in Table 2. The assignment of the chemical bonds from which these absorptions derive from are based on infrared reflectance measurements of model compounds made by Miyazawa et al. (1956). The absorptions of the peptide linkage involve the carbon and nitrogen bond, which is the peptide bond center, and the interaction with attached oxygen and hydrogen atoms. Thus, C-N, C=O, and N-H bonds all contribute to the characteristic absorptions related to the peptide linkage. The protein absorptions commonly observed near 2.054 µm and 2.172 µm arise from combinations of several nitrogen related absorptions as shown in Table 3 (Harris and Chapman, 1994). Thus, the protein absorptions at 2.054 µm and 2.172 µm absorptions all have some character related to nitrogen containing bonds.
 

In summary, protein reflectance spectra show absorptions due to combinations of the characteristic absorptions of the amide bond (CONH2) between peptide units. The 2.054 µm and 2.172 µm absorptions of proteins result directly from nitrogen in the molecular structure, in particular from C-N and N-H bonds. As shown in Figures 3-5, the plant spectra of dry leaf matter show changes due to absorptions centered at 2.054 µm and 2.172 µm as nitrogen concentration increases. This coincidence suggests that the 2.1 µm feature in dry leaf reflectance spectra broadens directly as a result of increased concentration of protein, and therefore nitrogen, within the plant material. Thus, this study demonstrates a link between increased nitrogen concentration and specific changes in the reflectance spectra of dry foliage.
 

Estimating Biochemical Concentrations in the Presence of Overlapping Absorptions
 

Figure 6 shows the reflectance spectra of protein (a vegetable protein nutrient supplement) and cellulose (100% cotton-bond paper). Both spectra show absorption features in the 2.0 - 2.2 µm region. Figure 6 shows that cellulose, a polymer of glucose and the major constituent of plants by dry weight, absorbs near the center of the 2.1 µm absorption feature observed in dry plant foliage. Starch, another polymer of glucose, has absorption properties similar to cellulose due to the same carbohydrate composition but slightly different configuration (i.e., the structural arrangement of bonds) of the molecules. All major plant components, other than protein, show an absorption centered at 2.1 µm (see figures in Elvidge, 1990). As cellulose, starch and hemi-cellulose have very similar spectra, the estimation of cellulose concentration may be subject to a sizable error because its reflectance properties can easily be confused with these other plant components.
 


Figure 6 at 72dpi.

Figure 6 at 150dpi
Figure 6 at 300dpi

Figure 6. Reflectance spectra of leaf constituents: protein (vegetable protein in powder form) and cellulose (100% cotton-bond paper). The protein spectrum shows two absorptions at 2.055 µm and 2.172 µm which lie on the shoulders of the broad 2.104 µm absorption feature of cellulose.



 

Protein absorptions centered near 2.054 µm and 2.172 µm are uniquely situated compared to cellulose (Figure 6) and the other major plant constituents, including lignin (see spectra in Elvidge, 1990). The offset in the position of absorptions indicates that, though nitrogen is a small fraction of the leaf by weight, changes in nitrogen concentration within the leaf may have significant impact on reflectance. This is an important consideration, suggesting that reflectance changes due to increased nitrogen may be detectable even when other components have variable concentrations.
 

In addition to protein, other nitrogen-containing plant constituents potentially have absorption features in the near infrared. Chlorophylls are other major nitrogen-containing components of plants. The chemical structures of chlorophylls are markedly different from proteins. The nitrogen atoms in chlorophylls are present in the porphyrin ring of the molecule around a central magnesium ion and do not have attached hydrogen. Chlorophyll a and b in plants do not contain amide bonds and N-H absorption features do not appear in their infrared spectra (Katz et al 1966). Chlorophylls exhibit strong absorption in the visible region arising from conjugated carbon-carbon single and double bonds of the porphyrin ring and the Mg ion. The infrared spectra of chlorophylls show strong absorption due to C-H bonds in the phytol tail of the molecule (Katz et al 1966). Thus, the spectra of chlorophylls differ greatly from proteins because of their different chemical structures. The differential absorption properties of protein and chlorophyll might lead to the development of algorithms which use both visible and near-infrared wavelength regions to estimate nitrogen concentration.
 

Past Empirical Studies Examined in Terms of Results in this Study
 

This study demonstrates that increases in nitrogen concentration have a consistent influence on the overall shape of the 2.1 µm absorption feature due to absorptions centered near 2.054 µm and 2.172 µm Previous studies have used statistical methods to link reflectance at certain wavelengths to changes in nitrogen or protein concentration. Table 4 presents the wavelength selection in previous statistical studies. In general, wavelengths in the 2.1 µm absorption feature were selected. In particular, wavelengths close to the 2.054 µm and 2.172 µm were found to be correlated with nitrogen concentration in nearly every study (as shown by boldface type in Table 4). The results from these empirical studies support the current hypothesis that nitrogen-containing protein absorptions represent a sound physical basis for estimating nitrogen concentration using reflectance spectra of dried and ground leaves.
 

Some previous studies have concluded that nitrogen containing compounds may have different spectral signatures from plant to plant (LaCapra et al 1996 and Jacquemoud et al 1995). However, the consistent spectral changes in this study taken together with success at cross-species prediction of nitrogen by Kokaly and Clark (1999) indicate that nitrogen containing compounds such as proteins have consistent absorption characteristics in all plants. Thus, it may be possible to develop a generally applicable algorithm capable of predicting nitrogen concentration in dry plant foliage from their reflectance spectra.
 

As the absorptions due to water are very strong, the weaker absorptions due to the plant biochemical components are less apparent in the reflectance spectra of fresh leaves. However, the regressions of nitrogen concentration in fresh and intact Douglas-fir needles with derivative and logarithmic transforms of reflectance spectra by Johnson and Billow (1996) showed strong correlations with the 2.1 µm absorption feature. In that study, samples of a single species (Psuedotsuga menseii) were manipulated by fertilization treatments to induce a wide range of nitrogen concentrations in the foliage. On the basis of that study, it appears that the reflectance spectra of same-species samples with large variations in nitrogen concentrations may differ enough to overcome the strong influence of water.
 

CONCLUSIONS
 

The near-infrared spectra of leaves result from a complex combination of scattering processes and overlapping absorptions arising from water and biochemical components. The present study demonstrated a physical basis for the link between the reflectance spectra of dried and ground foliage and nitrogen concentration based on protein absorptions at 2.054 µm and 2.172 µm The major difference in the 2.1 µm absorption features of samples with different nitrogen concentration is the change from a relatively narrow feature for samples with low nitrogen concentration to a broad absorption feature for samples with higher nitrogen concentration. Ratios of continuum removed and scaled absorption features for three pairs of samples differing only in their measured nitrogen concentrations showed the broadened absorption feature to arise from absorptions centered near 2.054 µm and 2.172 µm. These absorptions were attributed to protein absorptions which originate from nitrogen in the molecular structure, in particular from C-N and N-H bonds. A review of past studies showed that statistical nitrogen estimation methods consistently choose wavelengths near these two absorptions. Although many generic applications of stepwise multiple linear regression have yielded high correlations for only isolated data sets, this study suggested that a restriction of the input set of wavelengths to areas of known physical absorption properties of the protein or other component of interest and an avoidance of strong water absorption features may give more consistent inter-site results.
 

The observed broadening of the 2.1 µm absorption feature as nitrogen concentration increases indicates that a generally applicable algorithm may be developed at the laboratory level to make estimates of nitrogen from the spectra of dried plant foliage. However, other leaf constituents have absorptions in the 2.1 µm wavelength region. Thus, the general assumption that the absorption feature broadens as nitrogen increases may be valid only for constant cellulose and lignin levels. Independent changes in the concentration of these other two components could cause the shape of the absorption feature to alter. However, the offset position of the protein absorptions compared to other leaf components suggests changes in nitrogen concentration may be observable in spectra despite complications arising from the absorptions of other leaf constituents. Additional studies should be conducted for other leaf components to improve the understanding of the near-infrared spectra of dry leaves. In this manner the physical basis of generally applicable algorithms for spectroscopic estimates of plant biochemical concentrations may be further developed.
 
 
 

ACKNOWLEDGEMENTS
 

This research was supported, in part, by the NASA Accelerated Canopy Chemistry Program through Interagency Agreement W-18,644. The author wishes to acknowledge the members of the NASA ACCP for their contributions of data sets important to spectroscopic studies of vegetation. Dr. Roger N. Clark provided his expert knowledge of spectroscopy to assist the author in the interpretations presented in this study. The author appreciates the tutelage of Dr. William C. Stickler in organic chemistry which aided in the understanding of plant biochemistry and protein spectra. Dr. Trude V.V. King and Dr. Ralph Root provided helpful reviews of this paper. Chemistry measurements and near-infrared spectra are available for the NASA ACCP data set at http://www-eosdis.ornl.gov/daacpages/accp.html.

REFERENCES
 

Aber, J.D., and Federer, C.A (1992), A generalized, lumped-parameter model of photosynthesis, evapotranspiration and net primary production in temperate and boreal forest ecosystems. Oecologia 92: 463-474.

ACCP (1994), Accelerated Canopy Chemistry Program Final Report to NASA-EOS-IWG. National Aeronautics and Space Administration, Washington D.C., 19 October 1994.

Bolster, K.L., Martin, M.E., and Aber, J.D. (1996), Determination of carbon fraction and nitrogen concentration in tree foliage by near infrared reflectance: a comparison of statistical methods. Canadian Journal of Forest Research 26: 590-600.

Card, D.H., Peterson, D.L., Matson, P.A., and Aber, J.D. (1988), Prediction of leaf chemistry by the use of visible and near infrared reflectance spectroscopy. Remote Sensing of Environment 26: 123-147.

Clark, R.N. (1999), Spectroscopy of rocks and minerals and principles of spectroscopy. In Remote Sensing for the Earth Sciences: Manual of Remote Sensing 3rd Edition, Vol. 3 (A.N. Rencz Ed.), John Wiley and Sons, Inc., New York, pp. 3-58.

Clark, R.N., Ghallagher, A.J., Swayze, G.A. (1990), Material absorption band depth mapping of imaging spectrometer data using a complete band shape least-squares fit with library spectra. Proceedings of the Second Airborne Visible/Infrared Imaging Spectrometer (AVIRIS) Workshop June 4-5, 1990 (R.O. Green Ed.), JPL Publication 90-54, pp. 176-186.

Clark, R.N. and Roush, T.L. (1984), Reflectance spectroscopy: Quantitative analysis techniques for remote sensing applications. Journal of Geophysical Research 89: 6329-6340.

Davies, A. and Grant, A.M.C. (1988), Near infrared spectroscopy for the analysis of specific molecules in food. In Analytical Applications of Spectroscopy (C.S. Creaser and A.M.C. Davies, Eds.), Royal Society of Chemistry, London, pp. 46-51.

Dungan, J., Johnson, L., Billow, C., Matson, P., Mazzurco, J., Moen, J., and Vanderbilt, V. (1996), High spectral resolution reflectance of douglas fir grown under different fertilization treatments: Experiment design and treatment effects. Remote Sensing of Environment 55:217-228.

Elvidge, C.D. (1990), Visible and near infrared reflectance characteristics of dry plant materials. Remote Sensing of Environment 11: 1775-1795. Gastellu-Etchegorry, J.P., Zagolski, F., Mougin, E., Marty, G., and Giordano, G. (1995), An assessment of canopy chemistry with AVIRIS - A case study in the Landes Forest, South-west France. International Journal of Remote Sensing 16: 487-501.

Grossman, Y.L., Ustin, S.L., Jacquemoud, S., Sanderson, E.W., Schmuck, G., and Verdebout, J. (1996), Critique of stepwise multiple linear regression for the extraction of leaf biochemistry information from leaf reflectance data. Remote Sensing of Environment 56: 1-12.

Harris, P.I. and Chapman, D. (1994), Analysis of polypeptide and protein structures using Fourier transform infrared spectroscopy. In Methods in Molecular Biology, Vol. 22: Microscopy, Optical Spectroscopy, and Macroscopic Techniques (C. Jones, B. Mulloy, and A.H. Thomas, Eds.), Humana Press Inc., Totowa, NJ.

Ingle, J.D. (1988), Spectrochemical Analysis, Prentice-Hall, Inc., Engelwood Cliffs, New Jersey, pp. 424-425.

Jacquemoud, S., Verdebout, J., Schmuck, G., Andreoli, G., and Hosgood B. (1995), Investigation of leaf biochemistry by statistics. Remote Sensing of Environment 54: 180-188.

Johnson, L.F. and Billow, C.R. (1996), Spectrometric estimation of total nitrogen concentration in Douglas-fir foliage. International Journal of Remote Sensing 17: 489-500.

Johnson, L.F., Hlavka, C.A., and Peterson, D.L. (1994), Multivariate analysis of AVIRIS data for canopy biochemical estimation along the Oregon transect. Remote Sensing of Environment 47:216-230

Katz, J.J., Dougherty, R.C., and Boucher, L.J. (1966), Infrared and nuclear magnetic resonance spectroscopy of chlorophyll. In The Chlorophylls (L.P. Vernon and G.R. Seely, Eds.), Academic Press, New York, p. 186-249.

Kokaly, R.F. and Clark, R.N. (1999), Spectroscopic determination of leaf biochemistry using band-depth analysis of absorption features and stepwise multiple linear regression. Remote Sensing of Environment 67: 267-287.

LaCapra, V.C., Melack, J.M., Gastil, M., and Valeriano, D. (1996), Remote sensing of foliar chemistry of inundated rice with imaging spectrometry. Remote Sensing of Environment55: 50-58.

Marten, G.C., Shenk, J.S., and Barton, F.E. II. Eds. (1989), Near-Infrared Reflectance Spectroscopy (NIRS): Analysis of Forage Quality, U.S. Dept. of Agric. Handbook 643, USDA, Washington D.C., pp. 1-96.

Marten, G.C., Halgerson, J.L., and Cherney, J.H. (1983), Quality prediction of small grain forages by near infrared reflectance spectroscopy. Crop Science 23: 94-96.

Martin, M.E., and Aber, J.D. (1997), High spectral resolution remote sensing of forest canopy lignin, nitrogen, and ecosystem processes. Ecological Applications 7:431-443.

Matson, P., Johnson, L.F., Billow, C., Miller, J. and Pu, R. (1994), Seasonal patterns and remote spectral estimation of canopy chemistry across the Oregon transect. Ecological Applications 4:280-298.

McLellan, T.M., Martin, M.E., Aber, J.D., Melillo, J.M., Nadelhoffer, K.J., and Dewey, B. (1991a), Comparison of wet chemistry and near infrared reflectance measurements of carbon-fraction chemistry and nitrogen concentration of forest foliage. Canadian Journal of Forest Research 21: 1689-1693.

McLellan, T.M., Aber, J.D., Martin, M.E., Melillo, J.M., and Nadelhoffer, K.J. (1991b),Determination of nitrogen, lignin and cellulose content of decomposing leaf material by near infrared reflectance spectroscopy. Canadian Journal of Forest Research 21: 1684-1688.

Miyazawa, T., Shimanouchi, T., and Mizushima, S. (1956), Characteristic infrared bands of monosubstituted amides. Journal of Chemical Physics 24: 408-418.

Murray, I. (1988), Description of NIR spectra in terms of X-H chromophores. In Analytical Applications of Spectroscopy (C.S. Creaser and A.M.C. Davies, Eds.), Royal Society of Chemistry, London, pp. 39-45.

Newman, S. D., Soulia, M.E., Aber, J.D., Dewey, B., and Ricca, A. (1994), Analyses of forest foliage I: Laboratory procedures for proximate carbon fractionation and nitrogen determination. Journal of Near Infrared Spectroscopy 2:5-14.

Norris, K.H., Barnes, R.F., Moore, J.E., and Shenk, J.S. (1976), Predicting forage quality by infrared reflectance spectroscopy. Journal of Animal Science 43:889-897.

Stuedler, P., Bowden, R., Melillo, J.M., and Aber, J.D. (1989), Influence of nitrogen fertilization on methane uptake in temperate forest soils. Nature 341: 314-316.

Wessman, C.A., Aber, J.D., and Peterson, D.L. (1989), An evaluation of imaging spectrometry for estimating forest canopy chemistry, International Journal of Remote Sensing 10: 1293-1316.
 
 
 

Table 1. Matched pairs of samples from the NASA ACCP data set with nearly equal lignin and cellulose concentrations and varied nitrogen concentrations.
 
Match Designator Sample ID Sample Type Biochemical Concentrations

(% by dry weight)

Nitrogen Lignin Cellulose
DF df_sample14 Douglas-fir needles 1.01 *** ***
df_sample77 Douglas-fir needles 3.07 *** ***
SP 92UCS3A116 Slash pine needles 0.62 21.52 36.86
92UCS3A154 Slash pine needles 1.49 20.79 36.16
NS-BW 92HFS12NS2 Norway spruce needles 1.29 22.61 42.62
92BHS10BW2 Basswood leaf 3.51 22.43 43.09
*** indicates this measurement was not made for the sample
 
 
 
 
 

Table 2. Characteristic Infrared Absorptions of the Amide Bonds in Proteinsa.
 
Designation Approximate Wavelength (µm) Approximate Frequency (cm-1) Chemical Bond Origin
A 3.030 3300 N-H stretching
B 3.225 3100 N-H stretching
I 5.917 - 6.25 1690-1600 C=O stretching, N-H bending, C-N stretching
II 6.349 - 6.757 1575-1480 N-H bending, C-N stretching
III 7.686 - 8.137 1301-1229 C-N stretching, N-H bending, C=0 stretching, O=C-N bending
IV 13.03 - 16.00 767-625 O=C-N bending, others
V 12.50 - 15.63 800-640 N-H bending
VI 16.50 - 18.62 606-537 C=O bending
VII 50.0 200 C-N torsion
aAdapted from Harris and Chapman (1994).
 
 
 
 
 

Table 3. Center Wavelength Positions of Combination Vibrational Absorptions Arising from Amide Bonds in Polypeptides and Proteinsa.
 
Wavelength (µm) Combination
2.054 N-H stretch + Amide II
2.168 2 x Amide I + Amide III
aAdapted from Davies and Grant (1988).
 
 
 
 
 

Table 4. Wavelengths of Reflectance and Derivatives Correlated with Nitrogen or Crude Protein Concentration By Statistical Methods for Dry Vegetation Matter.
 
Reference Correlated Wavelengths (µm) Sample Types Spectral Data Type
Kokaly and Clark, 1999 2.036, 2.050, 2.078, 2.152, 2.180 Primarily dried and ground coniferous and deciduous forest foliage (NASA ACCP) Normalized and continuum removed reflectance
Bolster et al., 1996 1.192, 1.98, 2.056, 2.168 Primarily dried and ground coniferous and deciduous forest foliage (NASA ACCP) 2nd derivative of

Log 1/R

McClellan et al., 1991a 1.686, 1.978, 2.170 Coniferous and deciduous forest foliage 2nd derivative of

Log 1/R

McClellan et al., 1991b 1.230, 2.090, 2.174, 2.378 Forest litter samples 2nd derivative of

Log 1/R

Card et al., 1988 2.03, 2.11, 2.12, 2.18, 2.21, 2.31 Dried and ground forest foliage  Log 1/R
Marten et al., 1983* 1.822, 2.142, 2.174, 2.490 Dried and ground forage matter Log 1/R
Norris et al., 1976* 1.574, 1.610, 1.786, 1.818, 2.084, 2.100, 2.164, 2.254 Dried and ground forage matter 2nd derivative of

Log 1/R

Johnson and Billow, 1996 1.722, 2.142, 2.346 Dry and intact Douglas-fir needles 1st derivative of

Log 1/R

Johnson and Billow, 1996 2.156-2.160, 2.172, and others Fresh and intact Douglas-fir needles 1st derivative of

Log 1/R

    Boldface indicates wavelengths near the absorptions observed in protein reflectance spectra which arises from the combination of the characteristic absorptions of the peptide linkage.
    * indicates that regressions were made with crude protein
 
 
 
 

Figure Captions
 

Figure 1. The reflectance spectrum of a sample of dried and ground white pine needles (Figure 1a) and the continuum lines for three major absorption features in a dry leaf. The isolated continuum removed reflectance for the three absorption features after the continuum removal calculation (Figure 1b).

Figure 2. The continuum removed reflectance of the 2.1 µm absorption feature showing the disparate absorption characteristics for two samples of dried and ground foliage: rice (dashed line) and basswood (solid line).
 

Figure 3. The continuum removed reflectance (Figure 3a) of the 2.1 µm absorption feature for two dried and ground Douglas-fir samples: a low nitrogen sample (dashed line) and a high nitrogen sample (solid line). The ratio of these two continuum removed reflectance spectra are shown in Figure 3b indicating that absorptions at 2.054 µm and 2.172 µm cause a broadening of the 2.1 µm absorption feature as nitrogen concentration increases.
 

Figure 4. The continuum removed reflectance (Figure 4a) of the 2.1 µm absorption feature for two dried and ground slash pine samples: a low nitrogen sample (dashed line) and a relatively higher nitrogen sample (solid line). The ratio of these two continuum removed reflectance spectra are shown in Figure 4b.
 

Figure 5. The continuum removed reflectance (Figure 5a) of the 2.1 µm absorption feature for two samples: a low nitrogen sample of dried and ground Norway spruce needles (dashed line) and a high nitrogen sample of dried and ground basswood leaves (solid line). The ratio of these two continuum removed reflectance spectra are shown in Figure 5b again indicating that absorptions at 2.054 µm and 2.172 µm cause a broadening of the 2.1 µm absorption feature as nitrogen concentration increases.
 

Figure 6. Reflectance spectra of leaf constituents: protein (vegetable protein in powder form) and cellulose (100% cotton-bond paper). The protein spectrum shows two absorptions at 2.055 µm and 2.172 µm which lie on the shoulders of the broad 2.104 µm absorption feature of cellulose.


1.

Use of trade names is for descriptive purposes only and does not constitute endorsement by the U.S. Geological Survey.