Printer Friendly

Simultaneous estimation of several soil properties by ultra-violet, visible, and near-infrared reflectance spectroscopy.


Soil is a non-ideal system, and is chemically and mineralogically more complex than the 'pure' systems often studied using traditional laboratory procedures. Mechanisms of soil processes are difficult to fully understand and the fundamental link between measured soil chemistry and particular soil attributes or properties may be complex. Chemical procedures used in the characterisation of soils may in fact further complicate the interpretation. The extraction procedures can change the equilibrium between solid and solution phases of soil and the analyte as a result of the extractant interactions in the solution and at the solution-particle interface. For this reason, there is an increasing tendency towards the development of techniques that preserve the basic integrity of the soil system (Janik et al. 1998). Sensing and scanning technologies are currently being developed to describe and obtain more efficiently and economically precise information on the extent and variability of soil attributes, which affect crop growth and yield (Viscarra Rossel and McBratney 1998).

The ultraviolet (UV, 250-400 nm) region, a zone of short-wavelength radiation that lies between the X-ray (30-250 nm) and the visible regions, has been used for identifying minerals such as iron and titanium oxides and hydroxides (Strens and Wood 1979). Although the visible spectrum (VIS, 400-700 nm) forms a small portion of the electromagnetic spectrum, it has obvious significance for soil classification. Soil visible reflectance, or colour, is a differentiating characteristic for many soil classes in all modern classification systems and it is an essential part of the definitions for both surface and subsurface diagnostic horizons (Baumgardner et al. 1985). Obukhov and Orlov (1964) reported that soils with an elevated content of iron could be easily distinguished by the inflection characteristic for pure [Fe.sub.2][O.sub.3]. They found the intensity of the reflection in the region from 500 nm to 640 nm to be inversely proportional to the iron content. The VIS region of soil spectrum has also been used for predicting soil organic matter. Krishnan et al. (1980) found that the VIS wavelength region (623.6 nm and 564.4 nm) provided a better prediction than the infrared wavelength region for determining the organic matter content of soil.

Wavelengths longer than the red portion of the visible spectrum are designated as the infrared region. Two important categories have been recognised. The first is the near-infrared (NIR) and second is the mid infrared (MIR). Both the NIR (700-2500 nm) and MIR (2500-25000 nm) regions have been used for determining various soil components. Analysis of soil using NIR has been used for the determination of pH, electrical conductivity (EC), soil moisture, organic carbon (OC), cation exchange capacity (CEC), total carbon, total nitrogen, and exchangeable cations (Dalal and Henry 1986; Morra et al. 1991; Ben-Dor and Banin 1995; Chang et al. 2001).

For sustainable and cost-efficient agricultural production, them is an increasing need to evaluate a wide range of nutrients in soil in routine or diagnostic tests in a cost-effective and timely manner. Although several procedures and techniques are available for the determination of soil fertility parameters in the laboratory, many have analytical problems and inefficiencies. These include reproducibility, reliability, time, and labour factors (Malley et al. 1999). Implementation of precision site-specific agriculture requires extensive information on soil properties, and conventional laboratory systems are too costly and labour-intensive to generate the necessary data (Dunnet al. 2002). Analysis of soil using reflectance technology is an alternative as it requires minimal sample preparation, and it is fast, cost-effective, non-destructive, and non-hazardous, and several constituents can be predicted simultaneously (Batten 1998).

In this study, we attempted to develop a relatively quick and cheap method based on the reflectance in the UV-VIS-NIR regions for simultaneous estimation of several soil properties. The soil properties studied were EC, pH, OC, air-dry gravimetric water content, free Fe, clay, sand, silt, CEC, and exchangeable Ca, Mg, K, and Na. We have used multivariate data analysis (principal component regression) of the reflectance spectra to develop a simple, fast, and reliable method for analysing soil properties.

Materials and methods

Soil sampling

One hundred and sixty-one soil samples, representing 11 of the 14 Soil Orders of the Australian Soil Classification System (Isbell 1996), namely Vertosols, Kandosols, Dermosols, Ferrosols, Kurosols. Sodosols, Tenosols. Rudosols, Calcarosols, Chromosols. and Podosols, were selected for this study. These samples were collected mostly from several common agricultural regions of New South Wales (n = 143) and a few from South East Queensland (n = 18). The sample locations are shown in Fig. 1. The sample population is a mix of surface (n = 112) and subsurface (n = 49) soil samples. A limited number of samples, or samples from a limited number of sites of similar soil types, can limit the calibration to a certain type of soil or only be applicable for the landscape studied. In this study we have selected a wider range of soil samples with a good range of chemical variation from different sites to develop calibration equations that may be useful for predicting soil samples generally.


Chemical analysis

Soil samples were dried at 40[degrees]C in an oven, ground, and then sieved to obtain the <2-mm traction. Standard laboratory methods (Rayment and Higginson 1992) were used to measure pH (1:5 soil water extract), air-dry gravimetric water content (gravimetric water content method), and EC (1:5 soil water extract). Organic carbon was measured by a modified Walkley and Black method (McCleod 1973), free iron by the citrate-bicarbonate-dithionate method (Mehra and Jackson 1960), particle size distribution by the hydrometer method (Gee and Bauder 1986), CEC and exchangeable Ca, Mg, K, and Na by the 0.01 M silver-thiourea method (Rayment and Higginson 1992).

Spectral measurement

The spectral reflectance of soil samples was measured using a UV-VIS-NIR spectrophotometer (Cary 500) equipped with a diffuse reflectance accessory (Labsphere DRA-CA-50D) at 1.1-nm intervals in the UV-VIS (250-700 nm) range and 3-nm intervals in the NIR (700-2500 nm) range. For each sample, ~20 g of air-dried soil was placed into a sample holder with a quartz window. Spectral reflectance curves were recorded digitally by 2 coordinates, reflectance (R%) and wavelength (nm). A standard sample supplied by the instrument manufacturer was used as a reference material for baseline correction. The standard sample is used to collect 100% uncorrected and 0% reflectance curves, and this information is used to a obtain corrected reflectance curve for each soil sample (known as zero/baseline correction). The original spectra consisting of reflectance data for 1065 in the 250-500 nm wavelength range were used to develop models. It requires approximately 3.5 min to scan a soil sample.

Soil sample selection for calibration and validation sets

The studied samples (n = 161) were divided randomly into calibration (n = 121) and validation (n = 40) set. The calibration set was used to develop a prediction equation, and the validation set was used to validate the predictive equation. After selecting the calibration and validation sets, soil samples were analysed for various chemical and physical properties in the laboratory. The overall strategy for using UV VIS-NIR spectra for soil sample selection was to make the selection by visual examination of 2-dimensional score plot of principal component 1 and 2. In principal component analyses (PCA), the first few components are known to account for 99% of the variation of the entire population. Here, principal components 1 and 2 account for 75.3% and 16.4% of the variation, respectively. Figure 2 shows the distribution of samples used for the calibration and the validation sets. It shows that samples used for validation are close, or in some cases overlapping, the samples used for calibration. The sample in the top right quarter away from the reference line was thought to be an outlier, but after investigation of its soil properties, it was found that this highly reflecting soil sample contained 98% sand. The colour and the texture of this sample may be responsible for its high reflectance and its distance from the other soil samples.


Statistical analysis

In this study, principal component regression analysis was useful to relate UV, VIS, and NIR spectra with measured soil properties. PCA is an example of a latent variable regression method (Martens and Naes 1987) and it reduces the set of orthogonal components, which represents most of the variability in the original data and contains a reduced amount of random measurement noise. It consists of first reducing the dimensionality of the data by PCA of the [n x p] data matrix, n representing the soil samples and the p the reflectance values. Multiple linear regression was then performed to relate the principal components to the dependent variable. The first 20 principal components were used as the regression variables in the calibration stage. For each individual soil property, stepwise regression was performed with these 20 components to select the most significant model. For example, 12, 11, and 13 principal components were used to develop calibration models for OC, air-dry gravimetric water content, and clay, respectively. The calibration equation obtained using PCA can be summarised as follows:

(1) Soil variable = [C.sub.0] + [C.sub.1]PC1 + [C.sub.2]PC2 +....+ [C.sub.20]PC20

where [C.sub.0] is a constant; [C.sub.1], [C.sub.2], ... [C.sub.20] are regression coefficients; and PC1, PC2, ... PC20 are the principal components. The coefficients that were not significant were dropped from the models.

The calculation of principal components and multiple linear regressions were performed using the JMP version 5 software (SAS 2002). No data pretreatment was used in this method and all calculations were done with the raw spectra having 1065 data points within the UV-VIS-NIR regions.

Prediction quality

The ability of the UV-VIS-NIR technique to predict a soil property was evaluated using statistical parameters commonly used for the NIR technique. One example is the coefficient of determination of measured and predicted values of soil samples ([r.sup.2]), which measures the proportion of the total variation accounted for by the model. The remaining variation is attributed to random error. The standard error of calibration (SEC) is the standard deviation of the difference between the measured and the estimated values for samples in the calibration set, whereas the standard error of prediction (SEP) is the standard deviation of the difference between the measured and the estimated values for samples in the validation set. SEC and SEP were calculated from the following equation:

(2) SEC or SEP = [[[summation of] ((y-x))]/n-1].sup.1/2]

where y is the predicted value by UV-VIS-NIR technique, x is the measured value by standard laboratory methods, and n is the total number of samples. The best calibration is the one with the highest coefficient of determination and lowest SEP.

Two other statistical parameters used to evaluate calibrations were the RPD and RER. The RPD is the ratio of the standard deviation (s.d.) of the measured values of soil properties in the predictioa set to the SEP:

(3) RPD = s.d/SEP

The RER is the ratio of the range of the measured values of soil properties in the prediction set to the SEP:

(4) RER = range/SEP

Outlier detection

Outlier prediction is important during the calibration modelling and monitoring phases. In this study, an outlier was defined as the sample having a difference between measured and predicted values larger than 3 times SEC or SEP and was subsequently excluded from the model. For the calibration set, the number of samples detected as outliers was 1 for pH, air-dry gravimetric water content, and exchangeable Mg; 2 for free Fe: and 3 for OC. No outliers were detected for clay, sand, silt, CEC and exchangeable Ca at the calibration stage. For the validation set the number of outliers was 1 for OC, air-dry gravimetric water content, free iron, clay, sand, and exchangeable Ca. No outliers were found for pH, silt, CEC, and exchangeable Mg at the validation stage. The results (values of SEC, SEE RPD, and RER) presented here were recalculated excluding the outlier samples for each model. Reinoviog the outliers for EC, exchangeable Na and exchangeable K at both calibration and validation stages did not improve their predictions. A summary of various steps used in the soil analysis using UV-VIS-NIR technique is presented in Fig. 3.


Results and discussion

Soil properties

The reflectance technique requires a calibration set of samples, which will represent the samples used for validation. Soil properties estimated by UV-VIS-NIR method and their values analysed by laboratory method are presented in the Table 1. Although soil samples were selected from the visual observation of the PC 1 and PC2 score plot, it can be seen that file range for samples in the validation set was within the sample range for the calibration set for most of the soil properties, except pH, silt, and exchangeable Mg. For above soil properties, the maximum values for the validation set were larger than the maximum values for the calibration set.

Spectral properties

Reflectance spectra of 6 soil samples, covering the spectral range of studied soil samples, are shown in Fig. 4. These reflectance spectra followed the same basic shape as observed by other researchers, with prominent absorption bands around 1400, 1900, and 2200 nm (Shepherd and Walsh 2002). These bands are associated with clay minerals, for example OH features of free water at 1400 and 1900 nm, and lattice OH features at 1400 and 2200 run (Hunt 1980). Soil physical and chemical properties showed correlation with specific wavelengths within UV-VIS-NIR regions. The correlation coefficient (r) at each available wavelength (in this study 1065) for various soil properties was plotted against wavelength (Fig. 5). The highest correlation for free Fe (r -0.44) was found in the UV region (382.4 nm), for OC (r = -0.40) in the VIS region (587, 585.9 nm), and for clay (r = -0.81) in the NIR region (1912, 1901 nm). For the rest of the soil properties, the highest r values were found mostly in the NIR region, which indicates that NIR region is more informative than the other wavelength regions for predicting pH, air-dry gravimetric water content, CEC, and exchangeable Ca and Mg.


Prediction of soil properties

The ability of the UV-VIS-NIR technique to predict soil properties by PCA method is presented in Table 2. For the calibration set, [r.sup.2] values ranged between 0.61 and 0.82 (which explained 61-82% of variation of the appropriate dependent variables) for pH (0.73), OC (0.61), air-dry gravimetric water content (0.82), free iron (0.78), clay (0.82), sand (0.72), CEC (0.75), exchangeable Ca (0.701), and exchangeable Mg (0.73). When predicting unknown samples, i.e. the validation stage, [r.sup.2] values ranged between 0.52 and 0.85 for pH (0.71), OC (0.76), air-dry gravimetric water content (0.85), free iron (0.52), clay (0.72), sand (0.53), CEC (0.64), exchangeable Ca (0.67), and exchangeable Mg (0.63), with an acceptable SEP for all these properties. Calibration [r.sup.2] <0.50 was considered unacceptable in this study, suggesting that prediction for EC, silt, exchangeable K, and exchangeable Na was poor. Figures 6 and 7 show the models for calibration and validation sets of several soil properties.


Usually, in most reflectance spectroscopy techniques, the best calibration is considered the one with the highest coefficient of determination ([r.sup.2]), and the lowest standard error of cross validation (SECV). Using cross validations on a dataset generally gives an over-optimistic indication of the actual performance of the models. When new and totally independent samples are predicted with a 'young' calibration, it is very rare to get an SEP at the same level as the SECV (Dardenne et al. 2000). At the same time, researchers have used different calibration strategies for their proposed methods; for example, Chang et al. (2001) were able to get [r.sup.2] values of 0.8 for soil moisture, sand, silt, CEC, and exchangeable Ca. The calibration strategy in their study was to select 30 similar soil samples to predict one unknown for each of the soil properties, which means that each test sample had a different calibration set. There is only a small amount of literature where a separate validation set has been used to test the performance of the calibration models. The predictions for several soil properties in this study can be compared with the prediction obtained by Ben-Dor and Banin (1995), who investigated 91 soil samples for simultaneous prediction of clay, specific surface area, CEC, air-dry hygroscopic moisture, carbonate content, and organic matter using separate validation sets of samples to test the calibration accuracy. The prediction of clay in our study follows a similar pattern to that obtained by those authors. In their study, calibration [r.sup.2] for clay was 0.76 with SEC 8.6 and validation [r.sup.2] was 0.56 with SEP 10.3. In our study, results obtained by the PCA model for clay were a calibration [r.sup.2] of 0.82 (SEC 7.8) and a validation [r.sup.2] of 0.72 with a SEP of 8.9.

McCarty et al. (2002) used a separate validation set (n = 60) to test the calibration model for total C, inorganic C, and OC by the MIR and NIR techniques. With the NIR technique, the [r.sup.2] value obtained for OC was 0.82 for the validation set with residual mean squared deviation of 5.5. Dunn et al. (2002) selected 90 soil samples as a validation set and obtained an [r.sup.2] value of 0.66 with an SEP of 0.25 for OC. In this study the validation [r.sup.2] for OC was 0.76 with an SEP of 0.44. The precision obtained for OC in our study was within the precision obtained by the previous authors, although the range for OC was different in each study and it is difficult to compare calibrations with different ranges for a soil property using only [r.sup.2] and SEP. Batten (1998) suggested that for comparing the SEE the range and standard deviation of the population should be considered. Therefore, in this study RPD and RER were calculated for PCA models for evaluation of the accuracy of the UV VIS NIR method. In agricultural applications, where RPD and RER were developed, an RPD >3 is considered acceptable and an RPD >5 is considered excellent. RER should be >10. In environmental applications, where samples are much more variable, 'acceptable' values for RPD and RER have not been established (Malley et al. 1999). While no critical levels of RPD mid PER have been set for the NIR analysis of soil, acceptable values depend on the intended application of the predicted values (Dunn et al. 2002).

Chang et al. (2001) reported that the NIR reflectance spectroscopy technique had the ability to predict various properties of soil and they used 3 categories based on RPD in the ranges >2.0, 1.4 2.0, and <1.4 to indicate decreasing reliability of prediction using this technique. The RPD values obtained in the present study were within 1.5-2.0 for pH (1.8), OC (1.7), air-dry gravimetric water content (2.0), clay (1.9), sand (1.5), CEC (1.6), exchangeable Ca (1.7), and exchangeable Mg (1.7). The RPD for free Fe was 1.3, which suggests a less reliable prediction for unknown soil samples. The laboratory method for determination of free iron is complicated and involves several steps including digestion, centrifugation, and atomic absorption analysis. The results with low accuracy obtained by UV-VIS-NIR reflectance spectroscopy can be acceptable due to the simplicity of this method.

Malley et al. (1999) used the NIR technique for the analysis of soil nutrients and they were able to get RER values between 5 and 27 for pH, Ca, Mg, Na, K, and Fe, while using only one type of soil (n = 28) for both the calibration and validation stages. Dunn et al. (2002) were able to get RER values with range 7.5 14.8 for pH, OC, CEC, and exchangeable Ca, Mg, K, and Na using 550 soil samples from 3 Soil Orders of the Australian Soil Classification System. Comparison of SEE RPD, and RER values determined here with those of other investigators could be misleading, because SEP values are dependent on the variation in the chemical analysis data and the total number of soil samples used in the study, and the values for RPD and RER again depend on the value of SEE In our study we considered acceptable RER values to be >6 due to the variation in the soil properties found within 161 samples. For example, the range of CEC was 27-336 m[mol.sub.c]/kg in sample population (n = 550) studied by Dunn et al. (2002), whereas it was 16-312 m[mol.sub.c]/kg for 161 samples in our study, which led us to conclude that increasing sample number in the calibration and validation datasets can improve the precision. The increased number of samples will reduce SEP and increase the value of RPD and RER. All these factors affect prediction, which can produce lower [r.sup.2] and SEP compared with other NIR investigations on soil properties.

Figure 8 represents validated [r.sup.2] against RPD and RER. Plot (a) shows a higher level of accuracy for air-dry gravimetric water content and less accuracy for free Fe, and plot (b) indicates a lower level of accuracy for sand by this method. From the above results, it can be concluded that, using the UV-VIS-NIR technique without data pretreatment and with exclusion of minimum outliers (1-3 for different models), simultaneous prediction of pH, OC, air-dry gravimetric water content, clay, CEC, exchangeable Ca, and exchangeable Mg is possible with RPD values between 1.6 and 2.0. The method requires further testing with larger calibration and validation datasets. However, for rapt& nondestructive measurements of a large number of samples within a relatively heterogeneous population, the UV-VIS-NIR equations presented in this study are likely to be useful.


Predictions were also compared with excluding the UV (250-400 nm) and the UV-VIS (250 700 nm) portions of the wavelength range (Table 3) for pH, OC, air-dry gravimetric water content, free iron, clay, CEC, exchangeable Ca, and exchangeable Mg. There was no significant improvement in SEP values for soil properties when excluding either the UV or UV-VIS portion of the soil spectra. Predictions for OC and CEC were improved within the VIS-NIR regions. Overall predictions were better with the whole spectral range (UV-VIS-NIR) than the NIR range alone. More research is required to fully evaluate the importance of the UV-VIS wavelength range in simultaneous prediction of several soil properties.


The results of this study show that simultaneous prediction of pH, air-dry gravimetric water content, OC, clay, CEC, exchangeable Ca, and exchangeable Mg is possible using the UV-VIS-NIR technique. Using this technique, one can estimate 60-80 soil samples/day without any chemical pretreatment. The sample preparation is easy, requiring only air drying and grinding, and operation of the spectrophotometer does not need highly skilled personnel. The UV VIS-NIR reflectance spectroscopy technique is a fast, cheap, environmental friendly method for the simultaneous characterisation of soil properties. The estimates are less precise than those obtained by routine chemical methods. Given the relative speed and cost of this approach and the large local variation of soil properties, we suggest that the ability to analyse a large number of samples at finer sampling intervals using the spectroscopic technique will outweigh the loss in analytical precision.
Table 1. Statistical summary of soil properties for the samples
used for calibration and validation sets

n, number of samples; s.d., standard deviation

Soil properties Calibration set (n = 121)

 Mean Min. Max. s.d.

pH(1:5 [H.sub.2]O) 7.3 3.7 9.2 1.19
EC(1:5 [H.sub.2]O) (mS/cm) 0.18 <0.01 1.49 0.23
Organic carbon (%) 0.99 0.06 4.95 0.81
Air-dry gravimetric water
 content (g/g) 0.04 <0.01 0.11 0.03
Free iron (%) 0.87 0.05 4.24 0.72
Clay (%) 38.8 1.7 71.7 18.7
Sand (0%) 43.5 8.3 98.3 23.1
Silt (%) 17.6 <0.01 40.0 8.8
CEC ([mmol.sub.c]/kg) 162.1 25.3 311.5 77.3
Fxch. cations ([mmol.sub.c]/kg)
Ca 35.2 <0.01 94.6 24.5
Mg 31.4 0.4 90.8 19.9
K 8.7 1.3 36.9 6.23
Na 17.5 0.2 108.6 21.8

Soil properties Validation set (n = 40)

 Mean Min. Max. s.d.

pH(1:5 [H.sub.2]O) 6.8 5.0 9.7 1.12
EC(1:5 [H.sub.2]O) (mS/cm) 0.16 0.01 0.79 0.16
Organic carbon (%) 1.17 0.15 3.93 0.82
Air-dry gravimetric water
 content (g/g) 0.03 <0.01 0.08 0.02
Free iron (%) 0.68 0.04 2.85 0.60
Clay (%) 30.5 6.7 66.7 16.8
Sand (0%) 51.1 15.0 90.0 21.8
Silt (%) 18.4 3.3 43.3 9.6
CEC ([mmol.sub.c]/kg) 127.7 15.6 247.8 68.9
Fxch. cations ([mmol.sub.c]/kg)
Ca 25.3 0.1 80.9 22.7
Mg 25.9 3.l 94.8 20.3
K 8.0 0.6 25.5 5.4
Na 14.7 0.2 72.8 17.7

Table 2. Prediction within calibration and validation sets using the
principal component regression model

n, number of samples; [r.sup.2], coefficient of determination for
measured and predicted values; SEC, standard error of calibration;
SEP, standard error of performance; RPD, ratio of standard deviation
of validation set to SEP; RER, range of the validation set to the SEP

Soil property Calib. Valid.
 [r.sup.2] SEC [r.sup.2]

pH(1:5 [H.sub.2]O) 0.73 0.62 0.71
EC(1:5 [H.sub.2]O) (mS/cm) 0.32 0.11 0.10
Organic carbon (%) 0.61 0.42 0.76
Air-dry gravimetric water
 content (g/g) 0.82 0.01 0.85
Free iron (%) 0.78 0.31 0.52
Clay (%) 0.82 7.8 0.72
Sand (0%) 0.72 12.2 0.53
Silt (%) 0.34 7.1 0.05
CEC ([mmol.sub.c]/kg) 0.75 38.0 0.64
Fxch. cations ([mmol.sub.c]/kg)
Ca 0.70 13.4 0.67
Mg 0.73 10.0 0.63
K 0.29 5.3 0.00
Na 0.46 19.4 0.34

Soil property SEP RPD RER

pH(1:5 [H.sub.2]O) 0.61 1.8 7.7
EC(1:5 [H.sub.2]O) (mS/cm) 0.13 1.0 4.5
Organic carbon (%) 0.44 1.7 8.6
Air-dry gravimetric water
 content (g/g) 0.01 2.0 8.5
Free iron (%) 0.46 1.3 6.1
Clay (%) 8.9 1.9 6.7
Sand (0%) 14.5 1.5 4.7
Silt (%) 9.8 0.9 4.0
CEC ([mmol.sub.c]/kg) 43.3 1.6 5.4
Fxch. cations ([mmol.sub.c]/kg)
Ca 13.4 1.7 5.0
Mg 12.3 1.7 7.5
K 6.5 0.8 4.0
Na 14.6 1.2 5.0

Table 3. Use of different wavelength regions in prediction of several
soil properties using principal component analysis

[r.sup.2], coefficient of determination for VIS-NIR(400-2NIR(nm)SEP
standard error of performance

Soil property UV--VIS--NIR VIS--NIR
 (250-2500 nm) (400-2500 nm)

 Valid. Valid.
 [r.sup.2] SEP [r.sup.2]

pH(1:5 [H.sub.2]O) 0.71 0.61 0.63
Organic carbon (%) 0.76 0.44 0.81
Air-dry gravimetric water
 content (g/g) 0.85 0.01 0.76
Free iron (%) 0.52 0.46 0.48
Clay (%) 0.72 8.9 0.73
CEC ([mmol.sub.c]/kg) 0.64 43.3 0.68
Ca ([mmol.sub.c]/kg) 0.67 13.4 0.68
Mg ([mmol.sub.c]/kg) 0.63 12.3 0.63

Soil property VIS--NIR NIR
 (400-2500 nm) (700-2500 nm)

 SEP Valid. SEP

pH(1:5 [H.sub.2]O) 0.68 0.70 0.62
Organic carbon (%) 0.35 0.68 0.45
Air-dry gravimetric water
 content (g/g) 0.01 0.80 0.01
Free iron (%) 0.49 0.49 0.48
Clay (%) 8.7 0.75 8.7
CEC ([mmol.sub.c]/kg) 39.2 0.67 40.7
Ca ([mmol.sub.c]/kg) 13.1 0.72 12.8
Mg ([mmol.sub.c]/kg) 12.4 0.59 13.1


Kamrunnahar wishes to thank the Faculty of Agriculture, Food and Natural Resources, The University of Sydney, for the award of AH Thurburn Postgraduate Scholarship.


Banen GD (1998) Plant analysis using near infrared reflectance spectroscopy: the potential and the limitations. Australian Journal of Experimental Agriculture 38, 697-706.

Baumgardner MF, Silva LE Biehl LL. Stoner ER (1985) Reflectance properties of soils. Advances in Agronomy, 38, 1-44.

Ben-Dor E, Banin A (1995) Near-infrared analysis as a rapid method to simultaneously evaluate several soil properties. Soil Science Society of America Journal 59, 364-372.

Chang CW, Laird DA, Mausbach MJ, Maurice J, Hurburgh JR (2001) Near-infrared reflectance spectroscopy--principal components regression analyses of soil properties. Soil Science Society of America Journal 65, 480-490.

Dalal RC, Henry RJ (1986) Simultaneous determination of moisture, organic carbon and total nitrogen by near infrared reflectance spectrophotometry. Soil Science Society of America Journal 50, 120-123.

Dardenne E Sinnaeve G, Baeteu V (2000) Multivariate calibration and chemometrics for near infrared spectroscopy: which method? Journal of Near Infrared Spectroscopy 8, 229-237.

Dunn BW, Beecher HG, Batten GD, Ciavarella S (2002) The potential of near-infrared reflectance spectroscopy for soil analysis--a case study from the Riverine Plain of south-eastern Australia. Australian Journal of Experimental Agriculture 42, 607-614. doi: 10.1071/EA01172.

Gee GW, Bander JW (1986) Particle size analysis. In 'Methods of soil analysis'. Part 1, 2nd edn (Ed. A Klute) pp. 383-411. (American Society of Agronomy: Madison, WI)

Hunt GR (1980) 'Spectroscopic properties of rocks and minerals. In 'Handbook of physical properties of rocks'. (Ed. CR Stewart) pp. 295-385. (CRC Press: Boca Ratun, FL)

Isbell RF (1996) 'The Australlau Soil Classification.' (CSJRO Publishing: Melbourne)

Janik LJ, Merry RH, Skjemstad JO (1998) Can mid infrared diffuse reflectance analysis replace soil extractions? Australian Journal of Experimental Agriculture 38, 681-696.

Krishnan P. Alexander JD, Butler BJ, Hummel JW (1980) Reflectance technique for predicting soil organic matter. Soil Science Society of America Journal 44, 1282-1285.

McCarty GW, Reeves JB, Reeves VB, Follett RF, Kimble JM (2002) Mid-infrared and near-infrared diffuse reflectance spectroscopy for soil carbon measurement. Soil Science Society of America Journal 66, 640-646.

McCleod S (1973) Studies on wet oxidation procedures for the determination of organic carbon in soil. In 'Notes on soil techniques'. pp. 73-79. (CSIRO Division of Soils: Melbourne)

Malley DF, Yesmin L, Wray D, Edwards S (1999) Application of near-infrared spectroscopy in analysis of soil mineral nutrients. Communications in Soil Science and Plant Analysis 30, 999-1012.

Martens H, Naes T (1987) 'Multivariate calibration by data compression.' (John Wiley and Sons: Chichester, UK)

Mehra OP, Jackson ML (1960) Iron oxide removal from soils and clays by a dithlonite-citrate system buffered with sodium bicarbonate. Clays and Clay Minerals 7, 317-327.

Morra MJ, Hall MH, Freeborn LL (1991) Carbon and nitrogen analysis of soil fractions using near-infrared reflectance spectroscopy. Soil Science Society of America Journal 55, 288-291.

Obukhov AI, Orlov DS (1964) Spectral reflectivity of the major soil groups and possibility of using diffuse reflection in soil investigations. Soviet Soil Science 2, 174-184.

Rayment GE, Higginson FR (1992) 'Australian laboratory handbook of soil and water chemical methods.' (Inkata Press: Melbourne)

SAS (2002) 'Statistics and graphics Guide: JMP Version 5.0.' (SAS Institute: Cary, NC)

Shepherd KD, Walsh MG (2002) Development of reflectance spectral libraries for characterization of soil properties. Soil Science Society of America Journal 66, 988-998.

Strens RGJ, Wood BJ (1979) Diffuse reflectance spectra and optical properties of some iron and titanium oxides and hydroxides. Mineralogical Magazine 43,347-354.

Viscarra Rossel RA, McBratney AB (1998) Soil chemical analytical accuracy and costs: implications from precision agriculture. Australian Journal of Experimental Agriculture 38, 765-775.

Manuscript received 24 October 2002, accepted 14 May 2003

Kamrunnahar Islam, Batwant Singh, Alex McBratney

Faculty of Agriculture, Food and Natural Resources, The University of Sydney, Sydney, NSW 2006, Australia
COPYRIGHT 2003 CSIRO Publishing
No portion of this article can be reproduced without the express written permission from the copyright holder.
Copyright 2003 Gale, Cengage Learning. All rights reserved.

Article Details
Printer friendly Cite/link Email Feedback
Author:Islam, Kamrunnahar; Singh, Balwant; McBratney, Alex
Publication:Australian Journal of Soil Research
Geographic Code:8AUST
Date:Nov 1, 2003
Previous Article:Size distributions and minimum Stokes diameters of soil particles.
Next Article:Age of loess deposits in the Central Tablelands of New South Wales.

Related Articles
More than skin deep: new, noninvasive technologies are being used to determine produce quality. How will these tools enhance retail marketing...
Artificial neural network analysis of laboratory and in situ spectra for the estimation of macronutrients in soils of Lop Buri (Thailand).
Multi-spectral vision technology for precision farming: the future of site-specific crop management. (Focus on Biological Engineering).
The use of remote sensing data to extract information from agricultural land with emphasis on soil salinity.
Multi-spectral vision technology for precision farming; the future of site-specific crop management.
Ultra-violet, visible, near-infrared, and mid-infrared diffuse reflectance spectroscopic techniques to predict several soil properties.
The use of diffuse reflectance spectroscopy for in situ carbon and nitrogen analysis of pastoral soils.
Evaluating near infrared spectroscopy for field prediction of soil properties.
Potentially mineralisable nitrogen: relationship to crop production and spatial mapping using infrared reflectance spectroscopy.

Terms of use | Privacy policy | Copyright © 2021 Farlex, Inc. | Feedback | For webmasters |