HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS		QUICK SEARCH:   [advanced] Author: Keyword(s): Year:  Vol:  Page:

This Article Abstract Figures Only Full Text (PDF) Free Alert me when this article is cited Alert me if a correction is posted Services Similar articles in this journal Alert me to new issues of the journal Download to citation manager Citing Articles Citing Articles via HighWire Citing Articles via Google Scholar Google Scholar Articles by Stevens, A. Articles by Tychon, B. Search for Related Content PubMed Articles by Stevens, A. Articles by Tychon, B. Agricola Articles by Stevens, A. Articles by Tychon, B. Related Collections Carbon Sequestration Soil Classification and Mapping Soil Organic Matter

Soil & Water Management & Conservation

Detection of Carbon Stock Change in Agricultural Soils Using Spectroscopic Techniques

^a Université catholique de Louvain, Département de Géographie, Place Pasteur, 3, 1348 Louvain-La-Neuve, Belgium
^b Univerité de Liège, Campus d'Arlon, Département des Sciences et Gestion de l'Environnement, Av. de Longwy, 185, 6700 Arlon, Belgium

^* Corresponding author (stevens{at}geog.ucl.ac.be)

	ABSTRACT

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS AND DISCUSSION CONCLUSIONS REFERENCES

 
Soil organic carbon (SOC) represents one of the major pools in the global C cycle. Therefore, even small changes in SOC stocks cause important CO₂ fluxes between terrestrial ecosystems and the atmosphere. However, SOC stocks are difficult to quantify accurately due to their high spatial variability. The aim of this paper is to evaluate the potential of Imaging Spectroscopy (IS) using the Compact Airborne Spectrographic Imager (CASI; 405–950 nm) and field spectroscopy with an Analytical Spectral Devices spectrometer (ASD; 350–2500 nm) to measure SOC content in heterogeneous agricultural soils. We used both stepwise and partial least square (PLS) regression analysis to relate spectral measurements to SOC contents. Standard Error of Prediction (SEP) for the ASD ranged from 2.4 to 3.3 g C kg^–1 depending on soil moisture content of the surface layer. Imaging spectroscopy performed poorly, mainly due to the narrow spectral range of the CASI. Tests using both the CASI and the Shortwave infrared Airborne Spectrographic Imager (SASI; 900–2500 nm) showed better results. The variation in soil texture and soil moisture content degrades the spectral response to SOC contents. Currently, SEP allows to detect a SOC stock change of 7.2–9.9 Mg C ha^–1 in the upper 30 cm of the soil, and is therefore still somewhat high in comparison with changes in SOC stocks as a result of management or land conversion (0.3–1.9 Mg C ha^–1 yr^–1). A detailed SOC maps produced by IS reflected the patterns in SOC contents due to the recent conversion from grassland to cropland.

Abbreviations: ASD, analytical spectral devices spectrometer • ASD_d, ASD data in ‘dry’ conditions • CASI, Compact Airborne Spectrographic Imager • IS, imaging spectroscopy • PLS, partial least square • RPD, ratio of performance to deviation • SASI, Shortwave infrared Airborne Spectrographic Imager • SD, standard deviation • SEC, standard error of calibration • SEL, standard error of laboratory measures • SEP, standard error of prediction • SOC, soil organic carbon • VIS-NIR-SWIR, visible–near infrared–short wave infrared

	INTRODUCTION

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS AND DISCUSSION CONCLUSIONS REFERENCES

 
THE GLOBAL SOC pool is estimated at 1500 Pg C, which is more or less three times that of the vegetation (Lal et al., 1998). Hence, soils play an important role in the global C cycle and small relative changes in this pool can lead to large fluxes of CO₂ toward or from the atmosphere. Considering the important historic SOC loss on conversion from forest to agricultural land, SOC stocks in agricultural land are generally low and there is a potential of C sequestration through land use change and improved management practices (Lal, 2004). However, the capacity to detect temporal changes in SOC stocks using traditional sampling and analysis techniques is quite limited due to (i) the large spatial variability in SOC content and (ii) the slow response of SOC on land use conversion or change in land management (0.3–1.9 Mg C ha^–1 yr^–1 in the upper 30 cm of mineral soils; Freibauer et al., 2004). Conventional measurement campaigns result often in under-sampling, because of the time-consuming and costly sampling and analysis. One solution to reduce the uncertainty due to spatial variability consists in increasing the number of samples with new and more efficient techniques (McCarty and Reeves, 2001). Spectroscopy is one of these techniques that can produce the large number of samples needed. Moreover, this technique allows analyzing soil properties in situ. It is fast, cost effective, nondestructive and does not use chemical reagents. However, this method is less accurate than the conventional methods such as wet oxidation and dry combustion. Moreover, IS can only use the reflectance of bare surface to measure soil properties and are not able to detect vertical gradients in SOC within the topsoil. In addition, the conditions of the soil surface can affect the spectral signal. Some of the properties that are subject to variation both in time and in space are: the degree of soil crusting as a result of rain drop impact, soil texture, soil moisture, roughness and vegetation or crop residue cover. These disturbing factors may induce changes in soil reflectance that approach or exceed the spectral response of organic matter (Barnes et al., 2003).

Soil spectra generally show three peaks near 1400, 1900, and 2200 nm and a few smaller ones between 2200 and 2500 nm, which are overtones and combinations bands of C-H, N-H, and O-H bonds vibrating in the mid-infrared (MIR) region (Chang et al., 2001; Fidêncio et al., 2002a; Martin et al., 2002). Reflectance characteristics of soils are related to chemical groups known as ‘chromophores’. Soil organic matter contains such ‘spectrally active’ groups like chlorophyll, oil, cellulose, pectin, starch, lignin, and humic acid in the visible, near infrared and short wave infrared (VIS-NIR-SWIR; 400–2500 nm) region (Ben-Dor et al., 1997). In general, soil reflectance decreases with organic matter content (Stoner and Baumgardner, 1981; Galvão et al., 2001; Kooistra et al., 2003).

Spectroscopy has demonstrated its capability to accurately determine SOC contents in the laboratory (Sudduth and Hummel, 1993; Ben-Dor et al., 1997; Reeves et al., 1999; Chang and Laird, 2002; Martin et al., 2002; Coûteaux et al., 2003; Sørensen and Dalsgaard, 2005), directly in the field with a portable spectrometer (Shonk et al., 1991; Barnes et al., 2003), or from airborne-hyperspectral sensors (Ben-Dor et al., 2002; Selige et al., 2003). Spectrometers have already been extensively used to determine SOC in the laboratory for the study of, for example, the spatial variability of SOC at the field level (Martin et al., 2002; Odlare et al., 2005), the dynamics of organic matter during a controlled decomposition process (Ben-Dor et al., 1997) or the determination of SOC fractions in forest soils (Ludwig and Khanna, 2001; Coûteaux et al., 2003). Few studies have investigated the use of field spectrometers or IS for the analysis of SOC. Imaging spectroscopy uses airborne or satellite based hyperspectral sensors to spatialize the spectral information. Hyperspectral sensors differ from multispectral instruments in the greater number of wavebands, enabling a precise recording of the spectrum and a detailed analysis of spectral properties of the soil. Each pixel provides the full spectrum of the surface in the sensor range, giving the possibility to study spatial distribution of soil properties with a high spatial resolution.

Field spectrometers have the advantage of portability but give less accurate estimation of SOC compared with the fore-mentioned laboratory techniques due to non-controlled environmental conditions like soil surface roughness and moisture. Until now, field spectrometry has mainly been used as a rapid tool for soil characterization (Kooistra et al., 2003; Udelhoven et al., 2003). Imaging spectroscopy can yield reasonable results (Ben-Dor et al., 2002; Selige et al., 2003) and has the advantage of providing large amounts of spatial data. From the discussion above, it appears that these techniques are promising in the context of SOC stock inventories (McCarty and Reeves, 2001; Martin et al., 2002; McCarty et al., 2002; Reeves et al., 2002) because they offer an instant analysis of SOC content and in the case of IS provide a full regional cover. In general, spectroscopy can be useful in studies where a large number of samples and spatial information are required like in precision agriculture and soil monitoring surveys.

This paper aims to evaluate the potential of VIS-NIR-SWIR spectroscopic techniques—field spectroscopy and imaging spectroscopy—to measure SOC content in agricultural soils within a region with heterogeneous soil types. The paper will further discuss the benefits that such techniques can offer in different applications such as determining CO₂ fluxes induced by SOC stock change or detecting the spatial pattern of SOC contents caused by land use history or natural factors.

	MATERIALS AND METHODS

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS AND DISCUSSION CONCLUSIONS REFERENCES

 
Study Site, Flight Campaign, Sampling, Soil Organic Carbon Analysis, and Acquiring Spectral Data
To increase the robustness of the calibration, data sets from two different studies with a variation in soil properties were used. The two test sites cover the localities of Ortho and Attert in Belgium and provide a range of soil types. In general, the soils are poorly developed (Incepticols) except for the clay soils (Luvisols; Table 1). The first study area (Ortho) is located in the Belgian Ardennes (50°9'15''N 5°33'50''E and 50°6'34''N 5°33'36''E). Cropland (60% cereal and 30% silage corn) represents ± 8% of this hilly area. Agricultural lands are generally located on high plateaus covered by silty and shallow soils. The Ardennes were used as a case study area, since the SOC contents in this region are among the highest in Belgium (Lettens et al., 2004). The second study area (Attert) is located in Belgian Lorraine (49°47' N 5°46' E and 49°43' N 5°42' E). This area is characterized by a heterogeneous land use of meadows and fodder maize with cereal crops. The zone was selected because of its variation in soil texture over a short distance (sandy to clayey soils; Table 1).

One hundred twenty two soil spectra were taken at Ortho and 40 at Attert with a field spectrometer Fieldspec Pro FR (Analytical Spectral Devices Inc., Boulder, CO). The ASD measures the reflectance for every nanometer in the range from 350 to 2500 nm. Soil spectra were obtained by averaging repeated scans—approximately six ASD measurements—taken around the marked sampling site within a square corresponding to the pixel size of the airborne instrument.

Airborne hyperspectral data was acquired from the Compact Airborne Spectrographic Imager (CASI-2, Itres Research Ltd, Alberta, Canada) mounted on a Dornier 228 aircraft from the NERC (Natural Environment Research Council). The CASI-2 operates in the VIS-NIR region (405–950 nm) with a spectral resolution of 96 bands (every 6 nm). The flight took place on a clear and windy day, 15 Oct. 2003. The aircraft flew over Ortho at an altitude of ±2700 m with a spatial resolution of 6 by 6 m and over Attert at an altitude of ±1500 m, providing a pixel size smaller than 2.5 by 2.5 m. Images were atmospherically, radiometrically, and geometrically corrected. Spectra were extracted from the data cube using ENVI (Research Systems Inc., Boulder, CO). A window of 3 by 3 pixels was used to extract spectral signatures of all the soil sampling at the Attert test site to make comparable spectral signatures of the two study areas.

Statistical Analysis
First, spectral data were converted into absorbance by taking the logarithm, to the base 10, of the inverse reflectance. To eliminate the noise, soil spectra were preprocessed using Matlab (The MathWorks Inc., Natick, MA) with combinations of pretreatments (Savitsky-Golay smoothing and derivative algorithm (Savitzky and Golay, 1964), gap derivative, moving average and skip). Datasets were randomly split into two sets for calibration and validation purposes. One quarter of soil samples was used as the validation set. Calibrations of SOC from spectral data were developed for each pretreatment using both stepwise and PLS regression using the SAS statistical package (SAS Institute Inc., Cary, NC). Partial least square is an alternative and useful regression method to determine SOC from spectroscopic data since multiple linear regression has shown some limitations when the number of samples is greater than the number of variables (Fidêncio et al., 2002a). The PLS approach seeks linear combinations of the predictors, called factors, that explain both response and predictor variation. The maximum number of PLS factors was set to 10 and determined using the predicted residual sum of squares statistic and leave-one-out cross validation.

The best treatment was considered to be the one with the lowest standard error of prediction (SEP). The SEP is an average prediction error estimate and can be used as an approximation of the standard deviation (SD) of the prediction error for all future prediction samples. It is given by Eq. [1]:

[1]

where n denotes the number of samples in the validation set, y_pred,i is the value of C predicted for sample i and y_ref,i is the associated reference value. The SEP overestimates the true prediction uncertainty since it includes measurement errors in the reference values (Faber et al., 2003). Standard errors of calibration (SEC) were computed as well. The SEC is the standard deviation of all the points from the reference values in the calibration set. The SEC and the SEP are related to the SD of the calibration and validation set: the lower the SD, the lower the standard errors. Therefore, they have to be compared with their corresponding SD. The ratio of SD in the validation set to the SEP is referred to as the ratio of performance to deviation (RPD) and can be used as an index of model accuracy (Chang et al., 2001). We also used the ratio between the SEC and the SD of the samples involved in the calibration phase to measure the predictive power of the model (Coûteaux et al., 2003). Bias and r² were also computed to assess the predictive ability of the model. To assess the relative efficiency of the technique to measure SOC content, the SEP will be compared with the standard error of laboratory analysis (SEL), that is, SD of differences between duplicate samples of the reference method. The SEL of the Walkley–Black method applied in the provincial centers for soil and crop analysis of the Walloon Region is 0.52 g C kg^–1 (Colinet, personal comm.).

Eight CASI samples, located near parcels borders and clearly showing a spectral response influenced by the vegetation were removed from the analysis. These samples present typically a steep slope in the spectral signature due to chlorophyll content at wavelengths between 690 and 740 nm, known as the "red edge" (Curran et al., 1990). Samples from the portable spectrometer with a low signal/noise ratio (SNR) were also removed from the analysis. Here, SNR is defined as the ratio between the average signal reflectance and the standard deviation of the repeated measurements. Signal/noise ratio > 3 were considered satisfactory. For ASD spectra, one range of the signature corresponding to water vapor band (1815–1940 nm) was removed. Beyond 2385 nm, the signal tends to be noisy due to a low level of incoming radiation (Kooistra et al., 2003) and was eliminated. During the calibration procedure, samples having t-statistic 2.5 were considered as spectral outliers and removed. The expression for t is given in Eq. [2]:

[2]

where X_pred is C predicted by spectroscopy and X_obs is C analyzed by the Walkley–Black method.

	RESULTS AND DISCUSSION

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS AND DISCUSSION CONCLUSIONS REFERENCES

 
Calibration and Validation
Initially, two different data sets were analyzed: field spectra from the ASD and airborne spectra from the CASI. However, we noticed that the soil surface dried rather quickly during the first day when the field spectra were measured. Therefore, the question arises if soil moisture has to be taken into account in the regression models. The soil moisture content of the soil surface was not measured, but rather the soil moisture content of the 0- to 6-cm topsoil was determined both by means of the Theta-Probe (volumetric soil moisture content) and by drying sealed soil samples (gravimetric soil moisture content). Due to the strong vertical gradient, neither method represents the soil moisture at the surface, for which the reflectance is measured. During 3 d of field spectroscopic measurements, different soil surface moisture conditions were encountered: wet during the first day and dry during the second and third day. To take into account the effect of moisture, a third dataset was created with ASD measurements taken only on a dry soil surface. Hyperspectral data, collected on the second day, were considered to be representative for a dry surface. Finally, three different data sets were processed: ASD field spectra, ASD field spectra in ‘dry’ conditions (ASD_d) and airborne data from the CASI.

To relate the spectra to SOC content, stepwise and PLS regression were applied to each pretreatment. Each model was ranked according to its SEP. Stepwise regressions yielded generally good SEC but SEP were not good enough to produce reasonable predictions (results not shown). The high SEPs were probably due to colinearity and overfitting. These problems were fixed by the standard PLS procedure since the full spectrum is used in the calibration phase. Table 2 shows, for each data set, the PLS output statistics of the pretreatment yielding the best model. The best combinations of pretreatments were a moving average with a window size of five bands and the skipping of every 10 data points for ASD, a first derivative with a gap of 1 band and the skipping of every 10 data points for ASD_d and a moving average with a window size of three bands and a first derivative with a gap of one band for CASI spectra. The SEC ranged from 1.7 g C kg^–1 for ASD_d to 2.8 g C kg^–1 for ASD. The SEC/SD ratio of ASD_d was quite low (0.25) but ASD has a higher value (0.45), indicating that quantitative prediction should be treated with caution. The SEP varies between 2.4 g C kg^–1 for ASD_d and 3.3 g C kg^–1 for ASD (Table 2). The SEP is an important statistic since it assesses the ability of the model to predict SOC content at unsampled sites. The r² of ASD was >0.8 and the r² of ASD_d as high as 0.9 denoting reliable models (Coûteaux et al., 2003). According to the classification of Chang et al. (2001), the RPD of the best model for ASD_d falls in the highest category meaning that SOC content of dry soils can be well predicted by a field spectrometer. The predictive ability of ASD based on its RPD falls in the intermediate category. Bias, which is a measure of the difference between reference and predicted means, were low (ranging from –0.12 to 0.08 g C kg^–1). This implies that models using the data from the portable spectrometer do not systematically under- or overestimate C contents determined by the classic Walkley and Black method.

Model predictive accuracy indexes in this study are within the range of those found in the literature (Table 3). Some studies show much lower SEPs, but this is often caused by a low variation in SOC content. Predictive statistics of ASD_d and CASI + SASI are relatively good, suggesting that those field spectrometer and hyperspectral remote sensors could efficiently predict SOC. However, SEP values are at least five times higher than the SEL (0.52 g C kg^–1). This difference could be due to the diversity of soils found in our two study areas, giving a large variation in soil texture and SOC content (Table 1). Although one needs some variability in soil properties to detect a trend and obtain a good relationship between C content and spectral characteristics of the soil, the variation can be too large. For instance, Ben-Dor and Banin (1995) used samples with a large variation in organic matter (3–132 g C kg^–1) and noticed that, above 40 g C kg^–1, the agreement between predicted and observed values was weak. They assumed that soil samples were divided into two groups having different spectral properties: soils with highly decomposed soil organic matter (0–40 g C kg^–1) and those with less decomposed organic matter (40–132 g C kg^–1). As it seems that the relationship between SOC and spectral properties is nonlinear, this can pose a problem when a universal calibration is sought (Sørensen and Dalsgaard, 2005). Previous efforts have shown that this problem could be partially solved by neural networks (Fidêncio et al., 2002b). Likewise, as Hill and Schütt (2000) pointed out, a good relationship can hardly be found if soil texture varies too widely. For this reason, the development of a predictive model with spectroscopy should be restricted to a homogeneous area with similar soil or geological unit (Kooistra et al., 2003; Udelhoven et al., 2003). Local calibrations are thus probably more appropriate than a regional calibration. Nevertheless, Kemper et al. (2005) showed that calibration models for the estimation of SOC content from laboratory data could demonstrate a tolerable stability on a regional/countrywide level.

Converted to SOC stock in the upper 30 cm of a soil using the bulk density measured in the field, the SEP of ASD corresponds to 7.2 to 9.9 Mg C ha^–1. This detection limit is still high in comparison with rates of SOC stock changes after land conversion or change in management of agricultural soils (0.3–1.9 Mg C ha^–1 yr^–1) found in the literature (Freibauer et al., 2004). The fact that biases are low is critical. Indeed, such biases are not reduced by averaging and would greatly influence the estimation of SOC stocks (Christy et al., 2003). Therefore, decreasing the SEP or in other words the detection limit of spectroscopic techniques is the first research priority. The decreasing of the detection limit could be achieved by a better understanding of the influence of disturbing factors on the spectral signal and the use of a sensor covering a wide spectral range. In this respect, McCarty et al. (2002) found that MIR region (2500–25 000 nm) is even more robust than NIR region 400–2500 nm) when developing calibrations for highly diverse soil samples. Furthermore, MIR calibrations performed significantly better than those of NIR (about twice that of NIR).

Mapping Spatial Variability of Soil Organic Carbon
Despite the lack of accuracy in the estimation of SOC with the CASI alone, as pointed out in previous sections, a SOC map of neighboring fields was produced to detect spatial patterns induced by land use or land management change (Fig. 1 ). This can be useful for instance in precision farming studies (e.g., Selige et al., 2003). Indeed, spatial variability of SOC is often too detailed to be captured by a coarse sampling grid (Odlare et al., 2005). We ran the best model on a pixel-by-pixel basis on selected fields of the data cube. Figure 1 shows an agricultural field extracted from the SOC map. According to recent aerial photographs dating from 1998, this large field was originally made up of smaller parcels of both grassland and cropland. We can clearly see the effect of land use change. Arable parcels to the west were permanent grassland 1 yr before the flight. The new arable parcel in the eastern part of the field (grassland tilled 1 d before the flight) has higher C content (mean: 30.2 g C kg^–1) than surrounding cropland parcels (mean: 27.9 g C kg^–1). Converted to SOC stocks in the 0- to 30-cm topsoil, this results in a difference of 6.9 Mg C ha^–1 in 1 yr or a decrease of about 7% in SOC stock. Nevertheless, the difference of 2.3 g C kg^–1 is smaller than the SEP (5.1 g C kg^–1), so that this result should be regarded only as a trend. The decline in SOC content on conversion from grassland to cropland is often observed (e.g., Burke et al., 1989; Bowman et al., 1990) and is mainly due to increased erosion and agricultural practices such as tillage and residue removal. The difference observed here is far greater than the value of –0.95 ± 0.3 Mg C ha^–1 yr^–1 by converting permanent grassland to cropland found by Soussana et al. (2004).

View larger version (85K): [in this window] [in a new window]   Fig. 1. Map of soil organic C content in a freshly plowed field after land consolidation. The borders of the original fields that were joined are indicated with dashed lines.

	CONCLUSIONS

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS AND DISCUSSION CONCLUSIONS REFERENCES

	ACKNOWLEDGMENTS

 
This work was funded under the PRODEX program of the European Space Agency (contract number C90166) and the support is gratefully acknowledged. We are grateful to the Vlaamse Instelling voor Technologisch Onderzoek (VITO) for organizing the flight campaign and carrying out the spectral measurements on the ground. We thank T. Stevens, C. Schmit and M. Bravin for their assistance during the field measurements. We also thank two anonymous reviewers for their helpful comments.

Received for publication January 19, 2005.

	REFERENCES

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS AND DISCUSSION CONCLUSIONS REFERENCES

 

• Barnes, E.W., K.A. Sudduth, J.W. Hummel, S.M. Lesch, D.L. Corwin, C. Yang, C.S.T. Daughtry, and W.C. Bausch. 2003. Remote- and ground-based sensor techniques to map soil properties. Photogramm. Eng. Remote Sens. 69:619–630.
• Ben-Dor, E.Y., and A. Banin. 1995. Near-infrared analysis as a rapid method to simultaneously evaluate soil properties. Soil Sci. Soc. Am. J. 59:364–372.[Abstract/Free Full Text]
• Ben-Dor, E., Y. Inbar, and Y. Chen. 1997. The reflectance spectra of organic matter in the visible near-infrared and short wave infrared region (400–2500 nm) during controlled decomposition process. Remote Sens. Environ. 61:1–15.[CrossRef]
• Ben-Dor, E., K. Patkin, A. Banin, and A. Karnieli. 2002. Mapping of several soil properties using DAIS-7915 hyperspectral scanner data- a case study over clayey soils in Israel. Int. J. Remote Sens. 23:1043–1062.
• Bowman, R.A., J.D. Reeder, and R.W. Lober. 1990. Change in soil properties in a Central Plains rangeland soil after 3, 20 and 60 years of cultivation. Soil Sci. 150:851–857.
• Burke, I.C., C.M. Yonker, W.J. Parton, C.V. Cole, K. Flach, and D.S. Schimel. 1989. Texture, climate, and cultivation effects on soil organic matter content in U.S. grassland soils. Soil Sci. Soc. Am. J. 53:800–805.[Abstract/Free Full Text]
• Chang, C.-W., D.A. Laird, M.J. Mausbach, and C.R. Hurburgh, Jr. 2001. Near-infrared reflectance spectroscopy–Principal components regression analysis of soil properties. Soil Sci. Soc. Am. J. 65:480–490.[Abstract/Free Full Text]
• Chang, C.-W., and D.A. Laird. 2002. Near-infrared reflectance spectroscopic analysis of soil C and N. Soil Sci. 167:110–116.[CrossRef]
• Christy, C., E. Lund, and D. Laird. 2003. Mapping soil carbon with on-the-go near infrared spectroscopy. Advances in Terrestrial Ecosystem Carbon Inventory Measurements and Monitor. Available online at http://www.oznet.ksu.edu/ctec/Fall%20Forum%20pdf%20files/Papers_Abstracts/Christy_Veris.pdf (verified 8 Dec. 2005).
• Coûteaux, M.-M., B. Berg, and P. Rovira. 2003. Near infrared reflectance spectroscopy for determination of organic matter fractions including microbial biomass in coniferous forest soils. Soil Biol. Biochem. 35:1587–1600.[CrossRef]
• Curran, P.J., J.L. Dungan, and H.L. Gholz. 1990. Exploring the relationship between reflectance red edge and chlorophyll content in slash pine. Tree Phys. 7:33–48.
• Faber, N.M., X.-H. Song, and P.K. Hopke. 2003. Sample-specific standard error of prediction for partial least squares regression. Trends Anal. Chem. 22:330–334.[CrossRef]
• Fidêncio, P.H., R.J. Poppi, J.C. de Andrade, and H. Cantarella. 2002a. Determination of organic matter in soil using near-infrared spectroscopy and partial least squares regression. Commun. Soil Sci. Plant Anal. 33:1607–1615.[CrossRef]
• Fidêncio, P.H., R.J. Poppi, and J.C. de Andrade. 2002b. Determination of organic matter in soils using radial basis function networks and near infrared spectroscopy. Anal. Chim. Acta 453:125–134.[CrossRef]
• Freibauer, A., M.D.A. Rounsevell, P. Smith, and J. Verhagen. 2004. Carbon sequestration in the agricultural soils of Europe. Geoderma 122:1–23.[CrossRef][ISI]
• Galvão, L.S., M.A. Pizarro, and J.C. Neves Epiphanio. 2001. Variations in reflectance of tropical soils: Spectral-chemical composition relationships from AVIRIS data. Remote Sens. Environ. 75:245–255.[CrossRef]
• Hill, J., and B. Schütt. 2000. Mapping complex patterns of erosion and stability in fry Mediterranean ecosystems. Remote Sens. Environ. 74:557–569.[CrossRef]
• Hummel, J.W., K.A. Sudduth, and S.E. Hollinger. 2001. Soil moisture and organic matter prediction of surface and subsurface soils using an NIR soil sensor. Comp. Electron. Agric. 32:149–165.[CrossRef]
• Islam, K., B. Singh, and A. McBratney. 2003. Simultaneous estimation of several soil properties by ultra-violet, visible, and near-infrared reflectance spectroscopy. Aust. J. Soil Res. 41:1101–1114.[CrossRef]
• Kemper, T., K. Böttcher, M. Machwitz, S. Sommer, and W. Mehl. 2005. An approach to chemometric methods for soil organic matter estimation from laboratory and remote sensing data. p. 50–51. In Abstract book of the 4th EARSeL Workshop on Imaging Spectroscopy. Warsaw, April 27–29 2005. Available online at http://www.wgsr.uw.edu.pl/zts/workshop/docs/abstracts4.pdf (verified 8 Dec. 2005).
• Kooistra, L., J. Wanders, G.F. Epema, R.S.E.W. Leuven, R. Wehrens, and L.M.C. Buydens. 2003. The potential of field spectroscopy for the assessment of sediment properties in river floodplains. Anal. Chim. Acta 484:189–200.[CrossRef]
• Lal, R., J. Kimble, R.F. Follett, J.M. Kimble, and B.A. Stewart. 1998. Pedospheric processes and the carbon cycle. p. 1–8. In Soil processes and the carbon cycle. CRC Press Inc., Boca Raton, FL.
• Lal, R. 2004. Soil carbon sequestration to mitigate climate change. Geoderma 123:1–22.[CrossRef][ISI]
• Lettens, S., J. Van Orshoven, B. van Wesemael, and B. Muys. 2004. Soil organic and inorganic carbon contents of landscape units in Belgium derived using data from 1950 to 1970. Soil Use Manage. 20:40–47.[CrossRef]
• Ludwig, B., and P.K. Khanna. 2001. Use of near infrared spectroscopy to determine inorganic and organic carbon fractions in soil and litter. p. 361–370. In R. Lal (ed.) Assessment methods for soil carbon. Lewis Publishing, Boca Raton, FL.
• Martin, P.D., D.F. Malley, G. Manning, and L. Fuller. 2002. Determination of soil organic carbon and nitrogen at the field level using near-infrared spectroscopy. Can. J. Soil Sci. 82:413–422.
• McCarty, G.W., and J.B. Reeves, III. 2001. Development of rapid instrumental methods for measuring soil organic carbon. p. 371–380. In R. Lal (ed.) Assessment methods for soil carbon. Lewis Publishing, Boca Raton, FL.
• McCarty, G.W., J.B. Reeves, III, V.B. Reeves, R.F. Follett, and J.M. Kimble. 2002. Mid-infrared and near-infrared diffuse reflectance spectroscopy for soil carbon measurement. Soil Sci. Soc. Am. J. 66:640–646.[Abstract/Free Full Text]
• Odlare, M., K. Svensson, and M. Pell. 2005. Near infrared spectroscopy for assessment of spatial soil variation in an agricultural field. Geoderma 126:193–202.[CrossRef][ISI]
• Reeves, J.B., III, G.W. McCarthy, and J.J. Meisinger. 1999. Near-infrared reflectance spectroscopy for the analysis of agricultural soil. J. Near Infrared Spectrosc. 7:179–193.
• Reeves, J., III, G.W. McCarty, and T. Mimmo. 2002. The potential of diffuse reflectance spectroscopy for the determination of carbon inventories in soils. Environ. Pollut. 116:277–284.[CrossRef]
• Salgó, A., J. Nagy, J. Tarnóy, P. Marth, O. Pálmai, and G. Szabó-Kele. 1998. Characterisation of soils by the near infrared technique. J. Near Infrared Spectrosc. 6:199–203.
• Savitzky, A., and M.J.E. Golay. 1964. Smoothing and differentiation of data by simplified least squares procedures. Anal. Chem. 36:1627–1638 [erratum: 44:1906–1909].[CrossRef]
• Selige, T., L. Nätscher, and U. Schmidhalter. 2003. Spatial detection of topsoil properties using hyperspectral sensing. p. 633–638. In J. Stafford and A. Werner (ed.) Precision agriculture. Wageningen Academic Publisher, Wageningen.
• Shonk, J.L., L.D. Gaultney, D.G. Schulze, and G.E. Van Scoyoc. 1991. Spectroscopic sensing of soil organic matter. Trans. ASAE 34:1978–1984.
• Sørensen, L.K., and S. Dalsgaard. 2005. Determination of clay and other soil properties by Near Infrared Spectroscopy. Soil Sci. Soc. Am. J. 69:159–167.[Abstract/Free Full Text]
• Soussana, J.-F., P. Loiseau, N. Vuichard, E. Ceschia, J. Balesdent, T. Chevallier, and D. Arrouays. 2004. Carbon cycling and sequestration opportunities in temperate grasslands. Soil Use Manage. 20:219–230.[CrossRef]
• Stoner, E.R., and M.F. Baumgardner. 1981. Characteristic variations in reflectance of surface soils. Soil Sci. Soc. Am. J. 45:1161–1165.[Abstract/Free Full Text]
• Sudduth, K.A., and J.W. Hummel. 1993. Soil organic matter, CEC, and moisture sensing with a portable NIR spectrophotometer. Trans. ASAE 36:1571–1582.
• Tavernier, R., and R. Maréchal. 1958. Carte des associations de sols de la Belgique. Pédologie 8:134–182.
• Touré, S., and B. Tychon. 2003. Estimation of soil organic matter by means of hyperspectral data analysis. In CASI-SWIR2002 Workshop, Bruges. 4 Sept. 2003.
• Udelhoven, T., C. Hemmerling, and T. Jarmer. 2003. Quantitative analysis of soil chemical properties with diffuse reflectance spectrometry and partial least-squares regression: A feasibility study. Plant Soil 251:319–329.[CrossRef]
• Van Orshoven, J., J. Maes, H. Vereecken, J. Feyen, and R. Didal. 1988. A structured database of Belgian soil profile data. Pédologie 38:191–206.
• Walkley, A., and I.A. Black. 1934. An estimation of the Degtjareff method for determining soil organic matter and a proposed modification of the chromic acid titration method. Soil Sci. 37:29–37.

This article has been cited by other articles:

				F. Chen, D. E. Kissel, L. T. West, W. Adkins, D. Rickman, and J. C. Luvall Mapping Soil Organic Carbon Concentration for Multiple Fields with Image Similarity Analysis Soil Sci. Soc. Am. J., January 11, 2008; 72(1): 186 - 193. [Abstract] [Full Text] [PDF]

This Article Abstract Figures Only Full Text (PDF) Free Alert me when this article is cited Alert me if a correction is posted Services Similar articles in this journal Alert me to new issues of the journal Download to citation manager Citing Articles Citing Articles via HighWire Citing Articles via Google Scholar Google Scholar Articles by Stevens, A. Articles by Tychon, B. Search for Related Content PubMed Articles by Stevens, A. Articles by Tychon, B. Agricola Articles by Stevens, A. Articles by Tychon, B. Related Collections Carbon Sequestration Soil Classification and Mapping Soil Organic Matter