Imaging spectrometry for remote sensing of ecosystem processes

Adv. Space Res. Vol. 12, No.7, pp. (7)361—(7)368, 1992 Printed in Great Britain. All rights reserved.

0273—1177/92 $15.00 Copyright © 1992 COSPAR

IMAGING SPECTROMETRY FOR REMOTE SENSING OF ECOSYSTEM PROCESSES C. A. Wessman Centerfor the Study ofEarth from Space/CIRES University of Colorado, Boulder,CO 80309-0449, U.S.A.

ABSTRACT Broad-band reflectance measurements of vegetation have had wide application in the form of indices based on the unique differential between chlorophyll absorption in the red wavelengths and structure-induced reflectance in the near infrared. However, background and atmospheric effects are irretrievably convolved in the measured signal and only partially removed through ratioing of wavebands. The numerous high resolution, contiguous spectral bands acquired by imaging spectrometers enhance the spectral separation of scene components such as shade, vegetation and soil. Further, absorption feature characteristics can be quantified through definition of inflection points, band depth, etc. Alterations in the chlorophyll absorption feature are indicative of phenological and stress-induced changes in chlorophyll activity. Research suggests that other canopy biochemical constituents, such as cellulose and lignin, influence reflectance in the shortwave infrared and can potentially be quantified using imaging spectrometry. Capability to estimate biochemical properties in terrestrial ecosystems would aid in the assessment of carbon fixation/allocation patterns, metabolic processes and nutrient availability altered by a changing environment, INTRODUCTION The plant canopy acts as a major interface between the earth surface and the atmosphere. As a consequence, it has a strong influence on the manner in which mass and energy exchanges occur within an ecosystem. Canopy structure exerts a strong influence on the temperature, radiation and vapor concentration regime of the plant canopy. The surface area available for gas and moisture exchange is a function of leaf area; the efficiency of exchange varies with physiological characteristics and the geometric display of the foliage. The chemistry of canopy foliage can reflect metabolic processes, nutrient availability and altered carbon allocation patterns. Changes in canopy chemistry may express changes in ecosystem processes such as productivity and decomposition. Quantitative assessment of parameters such as photosynthetic capacity, leaf area and canopy chemistry and their dynamics over time would provide knowledge of the spatial extent and variation of carbon/nutrient sources and sinks crucial to understanding gas exchange between vegetation and the atmosphere. Since the canopy is the principle surface sensed by remote instruments, we need to examine canopy characteristics influencing optical properties, and, importantly, which are indicative of ecosystem function. Remote sensing is viewed as an important tool for landscape or regional estimation of ecosystem function. However, while it provides a way to obtain data over large areas, it does not always permit direct measurement of ecosystem processes, in particular terrestrial biogeochemical cycles. Better definition of remotely sensible properties that can serve as surrogates for ecosystem processes (such as the relationship of pigment concentration to absorbed photosynthetically active radiation) or that can serve as input variables to ecosystem models that calculate process rates is possible with the information available in imaging spectrometry data. This paper presents an overview of the information imaging spectrometry may offer with respect to the study of largescale ecosystem processes and how it differs from the information derived from discrete broad-band measurements made by currently operating systems. The discussion will include current capabilities to assess vegetation productivity using broad-band instruments; the potentials in (7)361

C. A. Wessman

(7)362

estimating canopy chemistry through high spectral resolution measurements of the visible and shortwave infrared region; the concept of spectral mixture analysis; and finally consideration of invertible ecosystem models using remote observations as drivers. VEGETATION PRODUCTIVITY ESTIMATES FROM BROAD-BAND INSTRUMENTS Broad-band reflectance measurements of vegetation have had wide application in the form of indices based on the unique differential between chlorophyll absorption in the red wavelengths and structure-induced reflectance in the near infrared. Field studies have demonstrated that crop canopy reflectance in the chlorophyll-absorbing red (R) spectral region ratioed to the highly reflected near infrared (NIR) region is linearly related to the canopy’s absorbed photosynthetically active radiation (APAR) /1,2/. As surrogates of APAR, the simple ratio (R/NIR) and the normalized difference (NDVI, (NIR-R)/(NIR+R)) vegetation indices can be related, in principle, to net productivity through an ‘efficiency’ term defining the carbon fixed per radiation intercepted /3/. Sellers’ /4,5/ theoretically demonstrated this relationship in a comparison of simple canopy photosynthesis models to radiative transfer of the red and infrared spectral wavebands within a uniform plant canopy. He established that, due to the strong association between chlorophyll density and APAR, plant canopy reflectance is indicative of instantaneous biophysical rates and, thus, supported near-linear relationships between canopy properties of APAR, photosynthetic capacity (Pc), and minimum canopy resistance (rc), and either of the two vegetation indices. The NDVI has been used extensively with data from the Landsat sensors (Multispectral Scanner, MSS, and Thematic Mapper, TM) and NOAA’s Advanced Very High Resolution Radiometer (AVHRR) for studies of vegetation productivity. AVHRR data are used to derive continental NDVI’s to monitor vegetation dynamics /6-8/. NDVI values correspond to expected seasonal variations in density and extent of green-leaf vegetation, and when integrated over the entire growing season, relate closely to annual net primary productivity (NPP). Assuming that the NDVI is a means of monitoring and quantifying the temporal and spatial variations of the terrestrial biosphere on a global scale, Fung and colleagues have gone one step further by utilizing the NDVI as an indicator of photosynthetic drawdown of CO 2 from the atmosphere in their modeling of global atmosphere dynamics /9,10/. These relationships, while appearing robust, have still to be verified in a wide variety of natural ecosystems with varying metabolic pathways, architecture and environmental resources. First, since biophysical processes are sensitive to environmental variables that may not be related to spectral indices, an understanding of these sources of variability will be required to adequately model ecosystem behavior / 11/. For example, a stressed system suffering a reduction in efficiency of light capture will experience a change in the relationship between absorbed photosynthetically active radiation (PAR) and carbon fixation. Efficiencies may also change with phenological phase as demands on resources shift throughout the growing season. Secondly, the relationship between APAR and NDVI is strong in continuous canopies above relatively dark soil. However, if canopy cover is only partial, the relationships can be very sensitive to variable background reflectance (e.g., soil and plant litter) and sun-sensor geometry /12,13/. Measurements of reflectance across wide regions of the spectrum (on the order of 50 to 100 nm) averages the contributions of surface components together (including shadowing), making it difficult or impossible to retrieve individual fractions. CANOPY CHEMISTRY Most remote sensing studies, such as those described above, are based on chlorophyll feature depth as measured by ratios of the red and near infrared reflectance. Understanding of the influence of vegetation properties on canopy reflectance is being extended by using more defined measurements of spectral shape available from the high spectral resolution instruments. Variables of spectral shape such as width, depth, skewness, and symmetry of absorption features may be more indicative of biochemical state and canopy physiology than average reflectance over relatively broad spectral regions. The visible spectral region is dominated by the presence of photosynthetic pigments. Wavelengthspecific absorption differences among chlorophylls, xanthophylls, and carotenes may permit quantification of their concentrations, and these may be related to photosynthetic activity and primary productivity. In one case, a spectral change in green reflectance resulted from a lightinduced change in a xanthophyll pigment that is closely linked to changes in photosynthetic capacity (Gamon et al., 1990). Variations in absorption feature shape in the visible region relate to

Imaging Spectrometry for Remote Sensing of Ecosystem Processes

(7)363

chlorophyll concentration /15-18/ and chlorophyll degradation /18/. Miller and colleagues /19/ have used the red edge and absorption maximum of the chlorophyll absorption feature to describe the phenological state of vegetation. Derivatives of high spectral resolution reflectance data in the visible and near infrared regions appear to be strongly related to AFAR and relatively insensitive to the reflectance of nonphotosynthetically active background /20/. Due to its relative linearity in the visible/near infrared, soil reflectance is less apt to play a significant role in the combined soil/vegetation reflectance and its influence can be reduced through curvature analysis or spectral mixing models /20-22/. Advances in near-infrared reflectance spectroscopy have prompted research of organic mixtures using field and airborne spectrometers. Such work is based on the fact that the short-wave infrared spectra of organic compounds consist of mixtures of organic and combination bands due to vibrations of strong molecular bonds between light atoms /23,24/. Spectroscopy principles suggest that pure leaf components such as cellulose, lignin, and protein, while not immediately apparent in the composite leaf reflectance, are important factors in its shape. Spectral reflectance of dried leaves exhibit diagnostic absorption features of major foliar chemical constituents such as cellulose, protein, and lignin /25,26/. A liquid water absorption curve of known concentration fit to a green leaf spectrum produces a residual spectrum resembling a dry leaf spectrum with absorption features due to materials other than water /27/. A review, of the absorption characteristics of individual foliar constituents can be found in Wessman /28/. The nature of absorptions within organic mixtures (such as a leaf or canopy) are weak and complex since they consist of overlapping overtone and combination bands. The origins of the observed vibrations are limited and they are all associated with primary constituents of vegetation. Knowledge of absorption characteristics of each of the major leaf constituents (e.g., cellulose, starch and protein) may permit remote assessment of canopy level concentrations if high spectral resolution reflectance information is acquired. A simpler taxonomy of constituents and their effective spectral and ecological combinations may be required if two materials influence the canopy spectrum in similar ways as to be indifferentiable (e.g. structural constituents such as lignin and cellulose). Nonetheless, an estimate of their combined concentration may be very useful for large scale applications to ecosystem research. A limited number of published successes /29,30/ have provided enough impetus to support such research, and several studies are currently underway in both the laboratory and field. Remotely-sensed measurements of vegetation canopy reflectance will not be as sensitive as those of a laboratory spectrophotometer to foliar chemical constituents. In the complete leaf or canopy condition, complications arise from the attenuation and multiple scattering of radiation within a plant canopy, additional influences from soil, shadow and other background components, and atmospheric scattering and attenuation of the reflected signal. High variation in measurements of in situ canopy reflectance resulting from environmental and sensor effects may introduce enough noise to limit interpretation of the electromagnetic signal. Nevertheless, if the reflectance of canopies is sampled at sufficient spectral detail through the use of imaging spectrometers, it is possible to perform curvature and spectral mixing analyses to separate factors of interest and derive more information with regard to changing biological state and functioning than with broad-band measurements alone. SPECTRAL MIXTURE ANALYSIS Spectral mixture analysis presupposes that most of the variance within remotely sensed data is due to combinations of a relatively small number of surface materials in the scene. Spectral “unmixing” separates the spectral contributions of the important scene components, including those at sub-pixel scales (such as leaves, branches, and litter) where individual components are not spatially resolved. Procedures for this approach have been pioneered by Adams and Smith and their colleagues /31-36/ as a means of relating the scene variance to a constant frame of reference: a library of spectra of scene components of known characteristics. The scene components or spectral endmembers are identified by matching their spectra against a library of known surface materials; concentrations are then determined on a pixel-by-pixel basis. Spectral mixture analysis serves as a strategy to stratify a spectrum or an image into fractional proportions of component classes. Image endmembers are selected from image spectra using a set of equations for each band as follows /35/: DNb

=

~ F~DN~,b+ Eb

C. A. Wessman

(7)364

DN represents the digital number in band b of an image pixel; F~is the fraction of endmember i where i = 1 to n; DNi,b is the radiance of endmember i in band b; and Eb is the error for band b of the fit of n spectral endmembers. The model calculates for each pixel the fraction of the endmember or image components represented by vegetation, soil, etc. For n spectral end-members, mixing is described within an (n - 1) dimensional volume. Imaging spectrometry provides better definition of endmember spectra and increased ability to identify both major and minor sources of variation within an image. A simple example of a mixture model is the separation of shaded versus illuminated surfaces. The fractions of shade at the pixel and image scales can be used to derive parameters of canopy architecture /37/. For example, two canopies illuminated at the same sun angle have a minimum of four endmembers at the pixel level: illuminated and shaded vegetation and illuminated and shaded soil (Fig. 1). A dimensionless index of the height to width of the canopy can be derived from the proportions of shade to illumination within a given pixel. Shadow

~Shadow

Fig. 1. Two canopies of different height will project proportionally different shadows at the same sun angle. Consequently, at the pixel level the proportions of shade to illumination will be different in each canopy type. Inter-pixel variance will be driven by the spacing or width of the canopy /38-41/. Case A (Fig. 2) shows a canopy where individual crowns are large relative to the pixel size. Shadowing will vary considerably from zero to large portions of shade depending upon where ,the pixel falls on the ground. As a result, the image variance will be high. Case B illustrates a situation where crowns are smaller than the relative pixel size. Variation from pixel to pixel will be minimal since change in the shadow component is relatively small.

Case A

Case B~

~

Fig. 2. Canopy size with respect to pixel size will affect image variance. In Case A, the crown is large compared with the pixel. The pixel can be filled with the crown or can fall on a gap, creating high image variance. In Case B, the crown is small with respect to the pixel and the pixel will have relatively the same proportion of shadow to illuminated crown wherever it is located.

Imaging Spectrometry for Remote Sensing of Ecosystem Processes

(7)365

Spectral mixture analysis forms a framework for the systematic separation and quantification of vegetative and non-vegetative components at sub-pixel spatial resolution /35,36/. This provides the means to unambiguously separate image variance associated with vegetative and nonvegetative processes. Many of the major sources of temporal and spatial variance in remotely sensed images result from factors independent of those of greatest biological significance. Spectral or temporal differences related to spectral measures of ecological process phenomena can only be sorted out after accounting for these other sources of variance. Spectral concentrations of the vegetative components defined by the spectral mixture analysis can then be used to parameterize canopy radiative models, spatial models, physiological models and other types of models which provide the appropriate variables needed for ecosystem models REMOTE SENSING AND ECOSYSTEM MODELS Sophisticated ecosystem models driven by remote observations would permit monitoring of ecosystem dynamics at local to global scales. A growing body of theory relates light interception by plant canopies to ecosystem functioning. However, remote sensing presents a combination of temporal and spatial resolution relatively unfamiliar to the ecosystem modeler. On one hand, remotely sensed data are instantaneous measurements and may be strongly influenced by high frequency changes in vegetation. This is the realm of very detailed biophysical or physiological models used to predict high temporal resolution vegetation dynamics. Ecosystem models that traditionally operate at the same large spatial scales provided by remote sensing have a coarse time resolution and limited ability to predict the instantaneous state of the vegetation. Most efforts to resolve this problem have used time-integrated remote measurements, as in annual integrated APAR /10,42/. This, however, ignores the temporal data, with potential information on nutrient uptake, herbivory and other dynamic aspects of ecosystems. One solution is to couple models across time scales, so that a model of “slow” processes (nutrient uptake, NPP) constrains a model of fast processes (ET) /43/. The fast model would provide integrated values to the slow model. Thus a slow model might pass LAT and photosynthetic capacity to a model of biophysical exchange, and receive integrated light interception and heat sums with which to calculate NPP. A combination of high spectral/low temporal resolution data and low spectral/high temporal remote sensing data may provide the important variables required by ecosystem models. For example, in addition to estimates of APAR, chlorophyll density and photosynthetic capacity are strongly related to canopy nitrogen and their time series may indicate nitrogen uptake and dynamics in the canopy. High temporal data at very coarse spatial resolution is currently provided by the AVHRR and in the future by the Moderate Resolution Imaging Spectrometer (MODIS) on NASA’s proposed Earth Observing System (EOS). The EOS High Resolution Imaging Spectrometer (HIRIS) will be used to define variability of these parameters at high spatial resolution with respect to the MODIS and with better precision due to its contiguous, high resolution spectral information. The potential estimates of canopy lignin and cellulose concentrations from airborne imaging spectrometers and the HIRIS may be used to infer carbon allocation patterns and potential litter decomposition rates /44,45/. Spectral mixture analysis, particularly suited to EOS’ multiband image data sets, can provide a framework for systematically defining both large and small scale features in the image data. This procedure can be used with a combination of MODIS and HIRIS data to sequentially model the smaller scale spectral features of plant canopies useful for quantification of ecosystem functioning. REFERENCES 1.

G. Asrar, B.M. Fuchs, E.T. Kanemasu, and J.L. Hatfield, Estimating absorbed photosynthetic radiation and leaf area index from spectral reflectance in wheat, Agron. 1. 76, 300-306 (1984).

2.

J.L. Hatfield, E.T. Kanemasu, G. Asrar, R.D. Jackson, P.J. Pinter, Jr., R.J. Reginato, and S.B. Idso, Leaf area estimates from spectral reflectance measurements from spectral reflectance measurements over various planting dates of wheat, mt. I. Remote Sens. 46, 651-656 (1984).

3.

J.L. Monteith, Vegetation and the Atmosphere, Vol. 2, Academic Press, New York, (1976).

4.

P.J. Sellers, Canopy reflectance, photosynthesis and transpiration, 1335-1372 (1985).

JASR 12:7-i

mt. I.

Rem. Sens. 6, 8,

(7)366

C. A. Wessman

5.

P.J. Sellers, Canopy reflectance, photosynthesis, and transpiration. II. The role of biophysics in the linearity of their interdependence, Remote Sens. Environ. 21, 143-183 (1987).

6.

C.J. Tucker, J.R.G. Townshend and T.E. Goff, African land-cover classification using satellite data, Science 227, 369-375 (1985).

7.

C.O. Justice, J.R.G. Townshend, B.N. Holben and C.J. Tucker, Analysis of the phenology of global vegetation using meteorological satellite data, mt. I. Remote Sens. 6, 8, 1271-1318 (1985).

8.

C.J. Tucker and B.J. Choudhury, Satellite remote sensing of drought conditions, Remote Sens. Environ. 23, 243-251 (1987).

9.

C.J. Tucker, I.Y. Fung, C.D. Keeling, and R.H. Gammon, Relationship between atmospheric C02 variations and a satellite-derived vegetation index, Nature 319, 195-199 (1986).

10.

I.Y. Fung, C.J. Tucker, and K,C. Prentice. Application of advanced very high resolution radiometer vegetation index to study atmosphere-biosphere exchange of CO2, 1. Geophys. Res. 92, 2999-3015 (1987).

11.

D.S. Bartlett, G.J. Whiting and J.M. Harman, Use of vegetation indices to estimate intercepted solar radiation and net carbon dioxide exchange of a grass canopy, Remote Sens. Environ. 30, 115-128 (1990).

12.

B.J. Choudhury, Relationships between vegetation indices, radiation absorption and net photosynthesis evaluated by a sensitivity analysis, Remote Sens. Environ. 22. 209-234 (1987).

13.

A.R. Huete, R.D. Jackson, and D.F. Post, Spectral response of a plant canopy with different soil backgrounds, Remote Sens. Environ. 17, 37-53 (1985).

14.

J.A. Gamon, C.B. Field, W. Bilger, A. Bjorkman, A.L. Fredeen and J. Penuelas, Remote sensing of the xanthophyll cycle and chlorophyll fluorescence in sunflower leaves and canopies, Oecologia (1990).

15.

D.M. Gates, H.J. Keegan, J.C. Schleter, and V.R. Weidner, Spectral properties of plants, App!. Opt. 4, 11-20 (1965).

16.

D.N.H. Horler, M. Dockray and J. Barber, The red edge of plant leaf reflectance, Sensing 4, 2, 273-288 (1983).

17.

B. Curtiss, and S.L. Ustin, The characterization of sources of illumination in a Ponderosa pine (Pinus ponderosa) forest community using the Portable Instantaneous Display and Analysis Spectrometer, SPIE Tech. Symp. on Optics, Electro-Optics, and Sensors, Orlando, FL, lop (1988).

18.

S.L. Ustin, V.C. Vanderbilt, J. Way, E.S. Kasischke, F.T. Ulaby, and M.C. Dobson, Opportunities for the EOS instrument synergism in monitoring of forest ecosystems, Proc. IEEE Geosci. and Remote Sensing Symp, 2301-2304 (1990).

19.

J.R. Miller, E.W. Hare, and J. Wu, Quantitative characterization of the vegetation red edge reflectance I. An inverted-Gaussian reflectance model, mt. I. Remote Sens. 10, 1755-1773 (1990).

20.

F.G. Hall, K.F. Huemmrich, and SN. Goward, Use of narrow-band spectra to estimate the fraction of absorbed photosynthetically active radiation, Remote Sens. Environ. 47-54 (1990).

21.

E.R. Stoner and M.F. Baumgardner, Characteristic variations in reflectance of surface soils, Soil Sci. Soc. Am. 1. 45, 6, 1161-1165 (1981).

22.

S.L. Ustin and B. Curtiss, Spectral characteristics of ozone-treated conifers, Environ. Exper. Botany 30, 3, 293-308 (1990).

23.

DL. Wetzel, Near-infrared reflectance analysis: sleeper among spectroscopic techniques, Analytical Chemistry 55, 1165A-1171A (1983).

mt. I. Remote

Imaging Spectrometry for Remote Sensing ofEcosystem Processes

(7)367

24.

L.G. Weyer, Near-infrared spectroscopy of organic substances, App!. Spectr. Rev. 21, 1&2, I43 (1985).

25.

CA. Wessman, J.D. Aber, D.L. Peterson, and J.M. Melillo., Foliar analysis using near infrared reflectance spectroscopy, Can I. For. Res. 18, 6-11 (1988).

26.

C.D. Elvidge, Visible and near infrared reflectance characteristics of dry plant materials, Remote Sensing 11, 1775-1795 (1990).

27.

A.F.H. Goetz, B.-C. Gao, C.A. Wessman and W.D. Bowman, Estimation of biochemical constituents from fresh, green leaves by spectrum matching techniques, Proc. Int. Geoscience and Remote Sens. Symp. 2, 971-974 (1990).

28.

C.A. Wessman, Evaluation of canopy biochemistry, in: Remote Sensing of Biosphere Functioning, ed. R. Hobbs and H. Mooney, Springer-Verlag, New York. pp. 135-156 (1990).

29.

C.A. Wessman, J.D. Aber, D.L. Peterson, and J.M. Melillo, Remote sensing of canopy chemistry and nitrogen cycling in temperate forest ecosystems, Nature 335, 154-156 (1988).

30.

N.A. Swanberg and P.A. Matson, Determining experimentally induced variation in coniferous canopy chemistry with airborne imaging spectrometer data, Proc. IEEE Geosci. and Remote Sensing Symp, 4, 2090-2092 (1989).

31.

J.B. Adams, M.O. Smith, and P.E. Johnson, Spectral mixture modeling: a new analysis of rock and soil types at the Viking Lander I. site, I. Geophys. Res. 91, 8098-8112 (1986).

32.

S.L. Ustin, J.B. Adams, C.D. Elvidge, M. Rejmanek, B.N. Rock, M.O. Smith, R.W. Thomas, and R.W. Woodward, Thematic Mapper studies of semiarid shrub communities, Bioscience 36, 446-452 (1986).

33.

J.B. Adams, M.O. Smith, and A.R. Gillespie, A mixing model strategy for analyzing and interpreting hyperspectral images. in: Remote Geochemical Analysis: Elemental and Minerological Composition, eds. CM. Pieters and P. Englert, LPI and Cambridge University Press. (in press).

34.

M.O. Smith, P.E. Johnson, and J.B. Adams, Quantitative determination of mineral types and abundances from reflectance spectra using principal components analysis, I. Geophysical Res. 90, C797-C804 (1985).

35.

MO. Smith, and S.L. Ustin, J.B. Adams, and A.R. Gillespie, Vegetation in deserts I. A regional measure of abundances from multispectral images, Remote Sens. Environ. 29, 1-26 (1990).

36.

M.O. Smith, and S.L. Ustin, J.B. Adams, and A.R. Gillespie. Vegetation in deserts II. Environmental influences on regional abundance, Remote Sens. Environ. 29, 27-52 (1990).

37.

C.A. Wessman, S. Ustin, B. Curtiss, Potential for large-scale ecosystem modeling using parameters derived from imaging spectrometer data, Proc. SPIE Int. Soc. for Opt. Eng. 1298, 164-170 (1990).

38.

X. Li, and A.H. Strahler, Geometric-optical modeling of a conifer forest canopy, IEEE Trans. Geosci. Remote Sens. GE-23, 508-514 (1985).

39.

X. Li, and A.H. Strahler, Geometric-optical bidirectional reflectance modeling of a conifer forest canopy, IEEE Trans. Geosci. Remote Sens. GE-24, 906-919 (1986~.

40.

X. Li, and A.H. Strahier, Modeling the gap probability of a discontinuous vegetation canopy, IEEE Trans. Geosci. Remote Sens. GE-26, 161-170 (1988).

41.

J. Franklin

42.

SN. Goward, C.J. Tucker and D.G. Dye, North American vegetation patterns observed with the NOAA-7 advanced very high resolution radiometer, Vegetatio 64, 3-14 (1985).

mt. I.

and A.H. Strahler, Invertible canopy reflectance modeling of vegetation structure in semiarid woodland, IEEE Trans. Geosci. and Remote Sens. GE26, 809-825 (1988).

(7)368

C. A. Wessman

43.

D. Schimel, private communication (1990).

44.

J.M. Melillo, J.D. Aber, and J.F. Muratore, Nitrogen and lignin control of hardwood leaf litter decomposition dynamics, Ecology 63, 3, 621-626 (1982).

45.

J. D.

Aber, C.A. Wessman, D.L. Peterson, J.M. Melillo, and J.H. Fownes, Remote sensing of litter and soil organic matter decomposition in forest ecosystems. in: Remote Sensing of Biosphere Functioning, eds. R. Hobbs and H. Mooney. Springer-Verlag, New York. pp. 87101 (1990).