Endmember Spectra Research Articles

Variations in Reflectance of Tropical Soils: Spectral-Chemical Composition Relationships from AVIRIS data

The relationships between Airborne Visible/Infrared Imaging Spectrometer (AVIRIS) surface reflectance values and constituents (total iron, organic matter, TiO 2, Al 2O 3, and SiO 2) of samples representative of three important soil types from central Brazil [Terra Roxa Estruturada (S TE), Latossolo Vermelho-Escuro (S LE), and Areia Quartzosa (S AQ)] were analyzed. End member spectra for green vegetation (GV), nonphotosynthetic vegetation (NPV), water (W), and the three soil types were selected by inspecting scatter plots derived from the principal components analysis (PCA) of 140 AVIRIS bands. They were then used to compose a six end member unmixing model to characterize the spectral reflectance variations associated with the different scene components, the spatial distribution of the soil types, and the effects of spectral mixing on the spectral-chemical composition relationships. Finally, regression equations fitted to soil constituents and their highly correlated spectral bands were used to produce maps showing the chemical variability in the scene for areas dominated by the presence of exposed soils, as indicated by the results from the unmixing model. The results showed a very good agreement between the spatial variability of the soil types and of the soil constituents. The largest squared correlation results were obtained for Fe 2O 3, TiO 2, and Al 2O 3, but the relationships were affected in the transition from the red to the near-infrared interval by the presence of nonsoil residues (e.g., senescent vegetation or litter) over the soil surfaces. In comparison with the light and loamy sand S AQ, the dark-red clay S TE and S LE presented higher contents of Fe 2O 3, Al 2O 3, and TiO 2, and consequently lower overall reflectance in the scene, because of the presence of greater amounts of opaque minerals. The prediction of these constituents from remote sensing data and their close association with the spatial distribution of the different soil types demonstrate the importance of the present investigation for soil mapping and soil erosion studies. All Rights Reserved.

Spectral Mixture Analysis: Linear and Semi-parametric Full and Iterated Partial Unmixing in Multi- and Hyperspectral Image Data

As a supplement or an alternative to classification of hyperspectral image data linear and semi-parametric mixture models are considered in order to obtain estimates of abundance of each class or end-member in pixels with mixed membership. Full unmixing based on both ordinary least squares (OLS) and non-negative least squares (NNLS), and the partial unmixing methods orthogonal subspace projection (OSP), constrained energy minimization (CEM) and an eigenvalue formulation alternative are dealt with. The solution to the eigenvalue formulation alternative proves to be identical to the CEM solution. The matrix inversion involved in CEM can be avoided by working on (a subset of) orthogonally transformed data such as signal maximum autocorrelation factors, MAFs, or signal minimum noise fractions, MNFs. This will also cause the partial unmixing result to be independent of the noise isolated in the MAF/MNFs not included in the analysis. CEM and the eigenvalue formulation alternative enable us to perform partial unmixing when we know one desired end-member spectrum only and not the full set of end-member spectra. This is an advantage over full unmixing and OSP. The eigenvalue formulation of CEM inspires us to suggest an iterated CEM scheme. Also the target constrained interference minimized filter (TCIMF) is described. Spectral angle mapping (SAM) is briefly described. Finally, semi-parametric unmixing (SPU) based on a combined linear and additive model with a non-linear, smooth function to represent end-member spectra unaccounted for is introduced. An example with two generated bands shows that both full unmixing, the CEM, the iterated CEM and TCIMF methods perform well. A case study with a 30 bands subset of AVIRIS data shows the utility of full unmixing, SAM, CEM and iterated CEM to more realistic data. Iterated CEM seems to suppress noise better than CEM. A study with AVIRIS spectra generated from real spectra shows (1) that ordinary least squares in this case with one unknown spectrum performs better than non-negative least squares, and (2) that although not fully satisfactory the semi-parametric model gives better estimates of end-member abundances than the linear model.

BRDF laboratory measurements

Laboratory studies of the bidirectional reflectance distribution function (BRDF) have greatly improved our understanding of the bidirectional reflectance characteristics of natural and man‐made surfaces. Explorative research and model validation studies in astronomy, remote sensing, material science, and computer graphics have likewise increased the knowledge of BRDF and its relationship to surface properties. In this review, a concise summary of laboratory BRDF studies relevant to remote sensing is given along with a discussion of major advantages and drawbacks of laboratory BRDF experiments and an overview of laboratory goniometric radiometers. Recommendations made for future research directions include the standardization of BRDF laboratory measurements, the initiation of hyperspectral BRDF databases for multidirectional endmember spectra, and the investigation of the BRDF phenomenon at different levels of scale by jointly analyzing laboratory, field, and remote sensing data.

Valence measurement of Mn oxides using Mn K β emission spectroscopy

High resolution Mn K β emission spectra provide a direct method to probe the effective spin state and charge density on Mn sites. Direct comparison of MnF 2 and MnO reveals significant changes due to the degree of covalency. The detailed shape and energy shift of the spectra for the perovskite LaMnO 3 and CaMnO 3 compounds are found to be very similar to Mn 2O 3 and MnO 2, respectively. Detailed Mn K β X-ray emission results on La 1− x Ca x MnO 3 can be well fit by linear superpositions of the end member spectra. However, for x<0.3, a retarded response is found. No evidence for Mn 2+ is found.

Synthesis and structural properties of clinopyroxenes of the hedenbergite CaFe2+Si2O6 - aegirine NaFe3+Si2O6 solid-solution series

Clinopyroxenes along the solid solution hedenbergite-aegirine M2[Ca2+1-xNa+xM1{Fe2+1-xFe3+x}Si2O6 were synthesized using hydrothermal techniques at 4 kbar. Different temperatures and redox conditions were used to determine optimum synthesis conditions and the stability range of individual compositions in the T - log fO2 field. Synthesized samples were characterized using microprobe analysis, X-ray powder diffraction and Mossbauer spectroscopy at 298 K and 80 K. The structure was refined in the C 2/ c space group by means of the Rietveld method. Along the solid-solution series between hedenbergite (a = 9.8448(6) A, b = 9.0296(6) A, c = 5.2452(4) A, β = 104.813) and aegirine endmembers (a = 9.6547(6) A, b = 8.7941(8) A, c = 5.2944(4) A, β = 107.398), the changes in unit cell dimensions show significant deviations from linearity. Mean and individual M1-O distances decrease linearly from hedenbergite to aegirine; mean M2-O and T-O distances do not change significantly, whereas individual length may vary. While in hedenbergite the coordination of the M2 site is 6+2, it is 4+4 in aegirine. The Mossbauer spectra of the solid-solution endmembers display narrowly split resonance absorption lines with hyperfine parameters typical for Fe2+ (δ = 1.18 mm/s, Δ = 2.25 mm/s at 298 K) and Fe3+ (δ = 0.38 mm/s, Δ = 0.30 mm/s at 298 K). Fe occupies only the M1 site. The Fe2+ resonance absorption is somewhat broadened in the 80 K spectra of the solid solution, which is due to a distribution of quadrupole splittings.

Improving the results of spectral unmixing of Landsat Thematic Mapper imagery by enhancing the orthogonality of end-members

Spectral unmixing is a technique that has been developed to derive fractions of spectrally pure materials that contribute to observed spectral reflectance characteristics of a mixture through a inverse least-squares deconvolution using end-member spectra. This technique has been shown to be very successful when applied to high spectral resolution imaging or non-imaging data where subtle diagnostic absorption features largely determine the spectral characteristics of the data. A large and vastly growing number of papers where spectral unmixing is applied to analyse low resolution image data (e.g. Landsat Thematic Mapper (TM), NOAA AVHRR, etc.) often to derive abundances of different materials as input parameters for models (i.e. land degradation models, crop growth models, hydrologic models, etc.) has evolved throughout recent years. This justifies efforts put into the quality assessment of these abundance estimates. In this paper we evaluate the effect of end-member redundancy on the deconvolution of spectral mixtures in unconstrained unmixing using simulated, one-dimensional spectral mixtures of three end-members that we unmix with two out of three of these components. Our analysis shows a relationship between the unmixing error and the difference between the true and estimated abundance with an index which combines (1) the weighted correlation of end-members in the mixture, (2) the correlation between the end-members used in unmixing this mixture, and (3) the amount of 'information' mapped in the end-members. Given this result we investigate the reduction of correlation in the spectral unmixing process and present an application of unmixing to decorrelated Landsat TM data using the minimum noise fraction transformation. The statistical evaluation of this experiment shows that over-and undershooting rather than the error in the unmixed spectrum can be significantly improved when decorrelating the data.

An investigation of the geological and geomorphological features of Fowlers Gap using thermal infrared, radar and airborne geophysical remote sensing techniques.

This study summarises the application of several remote sensing techniques to investigate various components of a land surface in the semi-arid environment of Fowlers Gap. These remote sensing techniques included NASA's Thermal Infrared Multispectral Scanner (TIMS) and CSIRO's Mid-Infrared Airborne CO, Laser Spectrometer (MIRACO,LAS), NASA's AIRSAR radar and geophysical airborne radiometrics. Linear spectral unmixing of extracted emissivities from the TIMS data produced four endmembers: quartz, clay minerals, dry vegetation (cellulose) in fine soils, and green vegetation/moisture. MIRAC0,LAS data identified spectral signatures similar to the spectra of endmembers derived from TIMS data. The sensitivity of both thermal infrared remote sensing techniques to the quartzklay contents and textures of the soils and sediments was confirmed by detailed laboratory spectral measurements. Surface roughness information from AIRSAR's band C radar backscatter assisted the discrimination of alluvial and colluvial quartz and clay-rich deposits from the outcropping geological units. In particular the C band AIRSAR radar discriminated the coarse grained sandstone and quartzite scree within the colluvial pediments, from the finer grained quartz-rich 'radar smooth' alluvium in the scalds. Airborne radiometrics were also found useful for further discriminating potassium and thorium- bearing phyllosilicate/clay minerals, within shales and ghyllites, from the kaolinite and montmorillonite- rich alluvium. This study found that TIMS data could identify some of the geomorphological features at Fowlers Gap, such as colluvial pediments, depositional scalds and gilgai landforms, that characterise some of the land systems in the Lowlands and Plains relief class of the Fowlers Gap land system classification. Thermal infrared remote sensing techniques also proved capable of discriminating areas of cellulose-rich dry vegetation and fine grained soils within the Plains relief class. The sensitivity of AIRSAR radar for topographic relief and surface roughness suggests that it is useful for distinguishing land systems in the Ranges relief class. Radiometrics appeared useful for land system definition when outcropping argillaceous units and alluvium assisted their classification. Key words: remote sensing, Fowlers Gap, thermal infrared, land system, spectral unmixing

Valence state of Mn in Ca-dopedLaMnO3studied by high-resolutionMnKβemission spectroscopy

$\mathrm{Mn}{K}_{\ensuremath{\beta}}$ x-ray emission spectra provide a direct method to probe the effective spin state and charge density on the Mn atom and is used in an experimental study of a class of Mn oxides. Specifically, the $\mathrm{Mn}{K}_{\ensuremath{\beta}}$ line positions and detailed spectral shapes depend on the oxidation and the spin state of the Mn sites as well as the degree of d covalency/itinerancy. Theoretical calculations including atomic charge and multiplet effects, as well as crystal-field splittings and covalency effects, are used as a guide to the experimental results. Direct comparison of the ionic system ${\mathrm{MnF}}_{2}$ and the covalent system MnO reveals significant changes due to the degree of covalency of Mn within atomic-type $\mathrm{Mn}{K}_{\ensuremath{\beta}}$ simulations. Moreover, comparisons of measurement with calculations support the assumed high spin state of Mn in all of the systems studied. The detailed shape and energy shift of the spectra for the perovskite compounds, ${\mathrm{LaMnO}}_{3}$ and ${\mathrm{CaMnO}}_{3},$ are, respectively, found to be very similar to the covalent ${\mathrm{Mn}}^{3+}{\ensuremath{-}\mathrm{M}\mathrm{n}}_{2}{\mathrm{O}}_{3}$ and ${\mathrm{Mn}}^{4+}{\ensuremath{-}\mathrm{M}\mathrm{n}\mathrm{O}}_{2}$ compounds thereby supporting the identical Mn-state assignments. Comparison to the theoretical modeling emphasizes the strong covalency in these materials. Detailed $\mathrm{Mn}{K}_{\ensuremath{\beta}}$ x-ray emission results on the ${\mathrm{La}}_{1\ensuremath{-}x}{\mathrm{Ca}}_{x}{\mathrm{MnO}}_{3}$ system can be well fit by linear superpositions of the end member spectra, consistent with a mixed-valent character for the intermediate compositions. However, an arrested Mn-valence response to the doping in the $x&lt;0.3$ range is found. No evidence for ${\mathrm{Mn}}^{2+}$ is observed at any x values seemingly ruling out proposals regarding ${\mathrm{Mn}}^{3+}$ disproportionation.

Infrared Spectral Imaging of Martian Clouds and Ices

Multispectral images of Mars, taken at the NASA Infrared Telescope Facility (IRTF) near and at the 1995 opposition, are used to identify and track its atmospheric clouds and ground ices. Band depth mapping is used to help distinguish between the composition of volatiles and provide a check for the techniques of principal components analysis (PCA) and linear mixture modeling (LMM). PCA/LMM are used to create maps that track clouds and volatiles, a technique that requires no a priori spectral information in order to create these maps. Band depth maps at 3.33 μm, which have been shown to trace CO2frosts, show some transient features which could indicate polar CO2clouds at the time of these observations. We show that band depth maps at 2.25 μm are good tracers of H2O frosts and that band depth maps at 3.69 μm can distinguish between coarse- and fine-grained water frosts. These maps have allowed the detection of fine-grained water frosts in the north polar region and along the morning and evening limb regions. From the PCA technique we find that just two principal components can account for over 99% of the data variance. The first of these is an infrared albedo unit and the second is an ice/thermal unit. Plotting the spectral data cubes in this new vector space, we find that most of the martian disk can be modeled by spectrally mixing three endmember spectra having extreme values of these principal components. The morning and evening regions of Mars are composed of 40–60% of the north polar ice/thermal component endmember, indicating a frost component there consistent with the band depth mapping results. With a combination of these techniques it is possible to not only identify the extensive martian clouds, but to also determine composition. These new results are particularly relevant in light of recent Mars Pathfinder descent temperature profile data that indicated upper atmosphere temperatures below the CO2frost condensation point, implying that CO2ice clouds may be an important radiative component of the current martian climate.

Multispectral and hyperspectral image analysis with convex cones

A new approach to multispectral and hyperspectral image analysis is presented. This method, called convex cone analysis (CCA), is based on the bet that some physical quantities such as radiance are nonnegative. The vectors formed by discrete radiance spectra are linear combinations of nonnegative components, and they lie inside a nonnegative, convex region. The object of CCA is to find the boundary points of this region, which can be used as endmember spectra for unmixing or as target vectors for classification. To implement this concept, the authors find the eigenvectors of the sample spectral correlation matrix of the image. Given the number of endmembers or classes, they select as many eigenvectors corresponding to the largest eigenvalues. These eigenvectors are used as a basis to form linear combinations that have only nonnegative elements, and thus they lie inside a convex cone. The vertices of the convex cone will be those points whose spectral vector contains as many zero elements as the number of eigenvectors minus one. Accordingly, a mixed pixel can be decomposed by identifying the vertices that were used to form its spectrum. An algorithm for finding the convex cone boundaries is presented, and applications to unsupervised unmixing and classification are demonstrated with simulated data as well as experimental data from the hyperspectral digital imagery collection experiment (HYDICE).

Effects of Band Positioning and Bandwidth on NDVI Measurements of Tropical Savannas

The consequences of shifting the spectral location of narrow and broad red (R) and near-infrared (NIR) bands on measurements of the normalized difference vegetation index (NDVI) of Brazilian savannas were investigated for different plant communities and seasonal phenology. The basic data came from an Airborne Visible/Infrared Imaging Spectrometer (AVIRIS) image obtained during the regional dry season and two Landsat 5-Thematic Mapper (TM) images from the dry and rainy seasons. Principal components analysis (PCA) was applied to a set of 64 AVIRIS bands from the interval of 500–1100 nm in order to select the R–NIR band pair of greatest spectral contrast and respective endmember spectra for the R and NIR band variations. For the endmember spectra, modeled AVIRIS-derived NDVI values, obtained by changing the spectral positioning and bandwidths of R–NIR band pairs, were compared with NDVI calculated from simulated R–NIR band pairs of orbital sensors. Furthermore, the suitability of different positions and bandwidths was also examined employing, as criterion, the calculated NDVI contrast between the endmember spectra. The results revealed that the NDVI of the green vegetation was essentially affected by the proximity of the R and NIR bands to the spectral interval of the red edge (690–750 nm). The strongest effects were observed for the R bands that partly extended longwards the 690 nm wavelength. On the other hand, the NDVI of the nonphotosynthetic vegetation varied up to 100% within the studied interval, especially as a function of the R–NIR spectral distance. The largest NDVI contrast between green and dry vegetation was obtained with the convergent displacement of the R and NIR bands, respectively, towards the 690 nm and 750 nm wavelengths, as predicted from PCA. Results from simulated orbital sensors demonstrated that bandwidths have no significant influence on the NDVI provided the R and NIR bands are not extended into the red edge domain. The variable NDVI contrasts along traverses presented the smallest values when associated with the conspicuous green vegetation, and the largest values when associated with the dried-out grasslands or exposed soils.

Optimal linear spectral unmixing

The optimal estimate of ground cover components of a linearly mixed spectral pixel in remote-sensing imagery is investigated. The problem is formulated as two consecutive constrained least-squares (LS) problems: the first problem concerns the estimation of the end-member spectra (EMS), and the second concerns the estimate, within each mixed pixel, of ground cover class proportions (CCPs) given the estimated EMS. For the EMS estimation problem, the authors propose a total least-squares (TLS) solution as an alternative to the conventional LS approach. The authors pose the CCP estimation problem as a constrained LS optimization problem. Then, they solve for exact solution using a quadratic programming (QP) method, as opposed to the Lagrange multiplier (LM)-based approximated solution proposed by Settle and Drake (1993). Preliminary computer experiments indicated that the TLS-estimated EMS always leads to better estimates of CCP than that of the LS-estimated EMS.

Iterative spectral unmixing (ISU)

Spectral unmixing techniques strive to find proportions of end-members within a pixel from the observed mixed pixel spectrum and a number of pure end-member spectra of known composition. The outcomes of such analysis are fraction (abundance) images for the selected (pure) end-members and a root mean square (RMS) error estimate representing the difference between the observed mixed spectrum and the calculated mixed spectrum. The RMS image can be used to select additional end-members and re-position existing ones. This is now done manually. In this Letter, an automatediterative approach is proposed using the RMS error image to select additional end-members and re-distribute older ones in order toincrease the accuracy of the spectral unmixing. Optimization criteria are proposed to drive the iterative process including minimization of the average RMS, minimizing the spread of the RMS values, minimizing the spatial structure of the RMS image, minimizing the spatial anisotropy of the RMS image and minimizing the local variance. The preliminary results of the analysis indicate that considerable improvement tothe spectral unmixing results are achieved using the iterative spectral unmixing (ISU) approach.

Raman spectra of silicate garnets

The single-crystal polarized Raman spectra of four natural silicate garnets with compositions close to end-members almandine, grossular, andradite, and uvarovite, and two synthetic end-members spessartine and pyrope, were measured, along with the powder spectra of synthetic pyrope-grossular and almandine-spessartine solid solutions. Mode assignments were made based on a comparison of the different end-member garnet spectra and, in the case of pyrope, based on measurements made on additional crystals synthesized with 26Mg. A general order of mode frequencies, i.e. R(SiO4)>T(metal cation)>T(SiO4), is observed, which should also hold for most orthosilicates. The main factors controlling the changes in mode frequencies as a function of composition are intracrystalline pressure (i.e. oxygen-oxygen repulsion) for the internal SiO4-vibrational modes and kinematic coupling of vibrations for the external modes. Low frequency vibrations of the X-site cations reflect their weak bonding and dynamic disorder in the large dodecahedral site, especially in the case of pyrope. Two mode behavior is observed for X-site cation vibrations along the pyrope-grossular binary, but not along the almandine-spessartine join.

Non-linear mixture modelling without end-members using an artificial neural network

Many methods of analysing remotely sensed data assume that pixels are pure, and so a failure to accommodate mixed pixels may result in significant errors in data interpretation and analysis. The analysis of data containing a large proportion of mixed pixels may therefore benefit from the decomposition of the pixels into their component parts. Methods for unmixing the composition of pixels have been used in a range of studies and have often increased the accuracy of the analyses. However, many of the methods assume linear mixing and require end-member spectra, but mixing is often non-linear and end-member spectra are difficult to obtain. In this paper, an alternative approach to unmixing the composition of image pixels, which makes no assumptions about the nature of the mixing and does not require end-member spectra, is presented. The method is based on an artificial neural network (ANN) and shown in a case study to provide accurate estimates of sub-pixel land cover composition. The results of this case study showed that accurate estimates of the proportional cover of a class and its areal extent may be made. It was also shown that there was a tendency for the accuracy of the unmixing to increase with the complexity of the network and the intensity of training. The results indicate the potential to derive accurate information from remotely sensed data sets dominated by mixed pixels.

Spectral mixture modelling and spectral stratigraphy in carbonate lithofacies mapping

Carbonate areas have been studied extensively since carbonate depositional environments have high economic potential and are likely to contain petroleum resources. Systematic mapping of carbonate lithofacies from remote sensing data can be done using the traditional practice of photogeologic interpretation. Additionally, spectral characterization yields compositional information complementary to the structural information that the photogeologic map provides. The Montalbán area in northern-central Spain is an ideal target for photointerpretation due to the impressive display of structural features and lithologic differentiation. Therefore most image interpretation studies in this area have focused on such an exercise neglecting largely the spectral information available. In this contribution, field and laboratory spectroscopy in conjunction with image analysis are used to construct a spectral stratigraphic column for the units exposed in the area. The findings from this analysis are used to select minerals likely to exist in the image. Furthermore, carbonate endmember spectra for a number of lithofacies are extracted from the image data. Both the pure mineral spectra and the lithofacies spectra are used as input for a spectral unmixing analysis to obtain abundance estimates of mineral- and rock-types for each pixel in the image. Image classification assumes that a pixel can be assigned to a single spectral class although most pixels are mixed in the sense that their spectral response is the product of a number of signatures together constituting the observed reflectance characteristics. Spectral unmixing acknowledges this fact and aims at separating the spectral contributions of a number of endmember spectra at a pixel support. Thus spectral unmixing yields abundance estimates for each endmember together summing-up to the 100% reflectance measured in the image. Besides this conceptual advantage of unmixing over classification, spectral unmixing, by contrast to classification, also provides an error assessment by comparing the observed pixel spectrum with the calculated mixed spectrum. This error matrix can be used to assess the accuracy of the unmixing systematics as well as the proper selection of endmembers.

Automated Mapping of Montane Snow Cover at Subpixel Resolution from the Landsat Thematic Mapper

A fully automated method uses Landsat Thematic Mapper data to map snow cover in the Sierra Nevada and make quantitative estimates of the fractional snow‐covered area within each pixel. We model winter and spring reference scenes as linear mixtures of image end member spectra to produce the response variables for tree‐based regression and classification models. Decision trees identify cloud cover and fractional snow‐covered area. We test the algorithm on a different Thematic Mapper scene and verify with high‐resolution, large‐format, color aerial photography. The accuracy of the automated classification of Thematic Mapper data equals that obtainable from the aerial photographs but is faster, cheaper, and covers a vastly larger area. The mapping method is insensitive to the choice of lithologic or vegetation end members, the water equivalent of the snow pack, snow grain size, or local illumination angle.

Analyses of Forest Foliage II: Measurement of Carbon Fraction and Nitrogen Content by End-Member Analysis

Chemical constituents of forest canopies are accepted as key indicators of ecosystem state and function. Remote sensing of these parameters offers the potential for rapid and accurate assessment of rates of key biogeochemical processes over large regions. NASA's Accelerated Canopy Chemistry Program was established to test the accuracy and generality of high-spectral resolution remote sensing in the measurement of lignin, cellulose and nitrogen concentrations in forest canopies. One of the most straightforward methods for extracting constituent concentration information from spectra is through simple linear mixing models of pre-determined end-member or pure compound spectra. While there are ample reasons to expect that linear mixing models will not work in foliar and whole-canopy samples, end-member analysis might still provide important information in support of standard and emerging statistical techniques, such as linear regression and partial least squares regression on first and second difference spectra, by indicating which parts of the spectrum should contain information on the concentrations of important constituents. The purpose of this paper is to present the results of two approaches to determining the value of end-member spectra for estimating constituent concentrations in foliage of temperate zone forest species. The first examines the spectral changes accompanying each step in the chemical proximate analysis method used to determine concentrations in the laboratory, and from them to infer the spectra of the fractions removed at each step. The second is the combination of known materials which approximate these same fractions in foliage into well-mixed samples to determine whether a simple linear mixing model can be used to predict the spectrum of the resulting mixture. Results confirm that the combination of linear mixing models and end-member analysis is not an appropriate technique for obtaining quantitative estimates of constituent concentrations for the major components of foliage of native woody plants, nor do we expect that more detailed analyses of plant ultrastructure or foliar spectra will correct the deficiencies identified. However, these results do suggest that the wet chemical procedures used to extract different carbon fractions produce consistent results with regard to the location of spectral features associated with the compounds removed at each step, and that the spectra of the cellulose and lignin isolated by this technique are very similar to those from pure materials. This in turn suggests that statistical techniques such as linear regression on first and second difference spectra and partial least squares methods which allow or correct for non-linear mixing, should be successful.

Green vegetation, nonphotosynthetic vegetation, and soils in AVIRIS data

An Airborne Visible / Infrared Imaging Spectrometer (AVIRIS) image collected over the Jasper Ridge Biological Preserve, California on 20 September 1989 was analyzed using spectral mixture analysis. The scene was calibrated to reflectance assuming a homogeneous atmosphere. The image was modeled initially as linear mixtures of the minimum number of reference endmember spectra that accounted for the maximum spectral variability. Over 98% of the spectral variation was explained by linear mixtures of three endmembers: green vegetation, shade, and soil. Additional spectral variation appeared as residuals. Nonlinear mixing was expressed as variations in the fraction of each endmember when a linear mixing model was applied to spectral subsets of the entire spectrum. After the fractions of the endmember spectra were calculated for each pixel, different types of soil were discriminated by the residual spectra. Nonphotosynthetic vegetation (NPV) (e.g., dry grass, leaf litter, and woody material), which could not be distinguished from soil when included as an endmember, was discriminated by residual spectra that contained cellulose and lignin absorptions. Distinct communities of green vegetation were distinguished by 1) nonlinear mixing effects caused by transmission and scattering by green leaves, 2) variations in a derived canopy-shade spectrum, and 3) the fraction of NPV. The results of the image analysis, supported by field observations in 1990 and 1991, indicate that the multiple bands of AVIRIS enhance discrimination of NPV from soil, and the separation of different types of green vegetation. The ability of the system to measure narrow absorption bands is one important factor; however, also important is the variation in continuum spectra expressed by the endmembers, and characteristic nonlinear mixing effects associated with green leaves.

End-Member Feldspar Concentrations Determined by FTIR Spectral Analysis

ABSTRACT Feldspar mineralogy, reported as concentrations of KAlSi3O8, NaAlSi3O8, and CaAl2Si2O8, can be determined from FTIR spectroscopy. In this study end-member FTIR spectra were constructed from sample spectra and their associated K-, Na- and Ca- feldspar concentrations using a least-squares full spectrum program. This was necessary because none of the samples have single-phase, pure end-member composition. The K-, Na-, and Ca-feldspar concentrations are computed by fitting the sample FTIR spectrum to the end-member FTIR spectra using a nonnegative least-squares program. For the feldspar samples examined we found that when the compositions determined from infrared spectroscopy are compared to the compositions deri ed from chemistry, the standard error of estimate for K-, Na-, and Ca-feldspar is 1.0, 0.9, and 0.7 absolute mole %, respectively.