Visible And Near-infrared Diffuse Reflectance Spectroscopy Research Articles

Temporal evaluation of soil chemical quality using VNIR and XRF spectroscopies

New soil sensing technologies, such as visible and near-infrared diffuse reflectance spectroscopy (VNIR) and X-ray fluorescence spectroscopy (XRF), emerged as alternatives to low-cost, rapid, and environmental-friendly soil analyses. Studies involving the prediction of integrated soil quality indices using individual and combined soil sensing techniques have not been conducted yet on tropical soils, especially regarding the temporal stability of models' performance. This study evaluated tropical soils to give insight into the potential of using individual and combined VNIR and XRF sensors to predict the Soil Chemical Quality Index (SCQI) and its sub-functions (Nutrient Availability, Acidity/Al Toxicity, and Nutrient Storage and Cycling). Furthermore, the temporal stability of these models' performance and a spiking strategy to mitigate performance loss throughout crop seasons were assessed. We conducted this study using 924 soil samples collected over three years (2018, 2019, and 2022) in a field managed with an integrated crop-livestock system (200 ha) in a Brazilian tropical area. We determined the SCQI and its sub-functions for each sample from 15 soil quality indicators. Soil samples were scanned with VNIR (350–2500 nm) and XRF (0–35 kV) spectrometers. VNIR spectroscopy predicted the SCQI, Nutrient Availability, and Nutrient Storage and Cycling with satisfactory accuracy (models classified as good, good, and excellent, respectively), outperforming the XRF technique and the fusion approach. The prediction performance of SCQI and its sub-functions does not show temporal stability. However, the performance loss can be mitigated using the spiking strategy, i.e., using new data from subsequent seasons to recalibrate the global model.

Dimensionality Reduction Statistical Models for Soil Attribute Prediction Based on Raw Spectral Data

To obtain a better performance when modeling soil spectral data for attribute prediction, researchers frequently resort to data pretreatment, aiming to reduce noise and highlight the spectral features. Even with the awareness of the existence of dimensionality reduction statistical approaches that can cope with data sparse dimensionality, few studies have explored its applicability in soil sensing. Therefore, this study’s objective was to assess the predictive performance of two dimensionality reduction statistical models that are not widespread in the proximal soil sensing community: principal components regression (PCR) and least absolute shrinkage and selection operator (lasso). Here, these two approaches were compared with multiple linear regressions (MLR). All of the modelling strategies were applied without employing pretreatment techniques for soil attribute determination using X-ray fluorescence spectroscopy (XRF) and visible and near-infrared diffuse reflectance spectroscopy (Vis-NIR) data. In addition, the achieved results were compared against the ones reported in the literature that applied pretreatment techniques. The study was carried out with 102 soil samples from two distinct fields. Predictive models were developed for nine chemical and physical soil attributes, using lasso, PCR and MLR. Both Vis-NIR and XRF raw spectral data presented a great performance for soil attribute prediction when modelled with PCR and the lasso method. In general, similar results were found comparing the root mean squared error (RMSE) and coefficient of determination (R2) from the literature that applied pretreatment techniques and this study. For example, considering base saturation (V%), for Vis-NIR combined with PCR, in this study, RMSE and R2 values of 10.60 and 0.79 were found compared with 10.38 and 0.80, respectively, in the literature. In addition, looking at potassium (K), XRF associated with lasso yielded an RMSE value of 0.60 and R2 of 0.92, and in the literature, RMSE and R2 of 0.53 and 0.95, respectively, were found. The major discrepancy was observed for phosphorus (P) and organic matter (OM) prediction applying PCR in the XRF data, which showed R2 of 0.33 (for P) and 0.52 (for OM) without using pretreatment techniques in this study, and R2 of 0.01 (for P) and 0.74 (for OM) when using preprocessing techniques in the literature. These results indicate that data pretreatment can be disposable for predicting some soil attributes when using Vis-NIR and XRF raw data modeled with dimensionality reduction statistical models. Despite this, there is no consensus on the best way to calibrate data, as this seems to be attribute and area specific.

Prediction of soil carbon and nitrogen contents using visible and near infrared diffuse reflectance spectroscopy in varying salt-affected soils in Sine Saloum (Senegal)

Soil organic carbon (C) and nitrogen (N) contents have an essential role in soil fertility, but they may be affected by salinity, which is especially responsible for land degradation in arid and semiarid regions. The objective of this work was to study the ability of visible and near infrared diffuse reflectance spectroscopy (VNIRS) to predict soil C and N contents and electrical conductivity (EC, a proxy for soil salinity) in variably salt-affected topsoils of the Sine Saloum region (Senegal). Different calibration procedures and spectral pretreatments were compared, and variable log-transformation usefulness was evaluated for prediction optimization.Predictions involved three calibration procedures: global partial least squares regression (PLSR), which used all calibration samples similarly; locally weighted (local) PLSR, with target samples predicted individually by giving higher weight to closest calibration spectra; and global PLSR per salinity class, after spectral discrimination of these classes. Predictions were performed with possible spectrum pretreatments (e.g., derivatization) and variable decimal log-transformation.The study was performed on 311topsoil samples (0–25 cm depth), either unsalted to slightly salty (Salt-, EC ≤ 2mScm−1; 262samples) or medium to highly salty (Salt+, EC > 2mScm−1; 49samples). Soil salinity was accurately discriminated using spectra: in validation, 100% and 95% of Salt- and Salt+ samples were correctly assigned on average, respectively. Best C and N content predictions were achieved after log-transformation using calibration by class (R2VAL = 0.87) and local calibration (R2VAL = 0.77), respectively; best EC prediction was achieved without log-transformation using global calibration (R2VAL = 0.90). This suggested C and N content predictions were affected by salinity; logC and logN distributions were almost symmetrical, hence log-transformation usefulness, while logEC distribution was very asymmetrical. No pretreatment yielded systematically good predictions; nevertheless, first-order derivative using 31-point gap often yielded good predictions, and second-order derivatives poor results.

Studying sand component of soil texture using the spectroscopic method

The mineral portion of soil is identified by its texture which refers to the relative amounts of sand, silt, and clay in the soil. The sand portion is closely related to soil quality which should be highlighted for environmental modeling operations and digital soil mapping projects. However, the identification, mapping, and monitoring of sand content at a large scale using standard laboratory analysis is laborious, time-consuming, and costly. This study aimed to predict sand content utilizing the proximal soil sensing technology. Thus, following supplementary data layers (geology, pedology, land use, etc.) and stratified randomized sampling method, 128 soil samples were collected from the depth of 20 cm of the soil. The study was conducted using the partial least squares regression (PLSR) method with leave-one-out cross-validation procedure and some preprocessing such as spectra reduction method, Savitzky-Golay smoothing, and first-order derivative algorithms. The definitive calibration model was validated using a standalone validation subset with two and four latent vectors and RP (0.83 and 0.82), R2P (0.68 and 0.67), RMSEP (8.68 and 8.83), RPDP (1.78 and 1.75), and RPIQP (2.45 and 2.41), respectively, that spotted as the most appropriate model for the sand content prediction in the study areas. Eventually, the visible and near-infrared diffuse reflectance spectroscopy (VNIR-DRS) potentiality was proved for sand content estimation. Significantly, it is feasible to upscale the sand prediction process utilizing the principal resulted in model and key spectral domains via airborne/satellite hyperspectral data, which emphatically shows the lab diffuse reflectance spectroscopy (LDRS) importance as a starting point for characterizing the informative optical wavelengths.

Prediction of bioaccessible lead in urban and suburban soils with Vis-NIR diffuse reflectance spectroscopy

The successful use of visible and near-infrared (Vis-NIR) reflectance spectroscopy analysis requires selecting an optimal procedure of data acquisition and an accurate modeling approach. In this study, Vis-NIR with 350–2500 nm wavelengths were applied to detect different forms of lead (Pb) through the spectrally active soil constituents combining principal component regression (PCR) and Partial least-square regression (PLSR) for the Vis-NIR model calibration. Three clouds with different soil spectral properties were divided by the Linear discriminant analysis (LDA) in categories of Pb contamination risks: “low,” “health,” “ecological,” ranging from 200 to 750 mg kg−1. Farm soils were used for calibration (n = 26), and more polluted garden soils (n = 36) from New York City were used for validation. Total and bioaccessible Pb concentrations were examined with PLSR models and compared with Support Vector Machine (SVM) Regression and Boosting Regression Tree (BRT) models. Performances of all models' predictions were qualitatively evaluated by the Root Mean Square Error (RMSE), Residual Prediction Deviation (RPD), and coefficient of determination (R2). For total Pb, the best predictive models were obtained with BRT (R2 = 0.82 and RMSE 341.80 mg kg−1) followed by SVM (validation, R2 = 0.77 and RMSE 337.96 mg kg−1), and lastly by PLSR (validation, R2 = 0.74 and RMSE 499.04 mg kg−1). The PLSR technique is the most accurate calibration model for bioaccessible Pb with an R2 value of 0.91 and RMSE of 68.27 mg kg−1. The regression analysis indicated that bioaccessible Pb is strongly influenced by organic content, and to a lesser extent, by Fe concentrations. Although PLSR obtained lower accuracy, the model selected many characteristic bands and, thus, provided accurate approach for Pb pollution monitoring.

Application of VIS-NIR spectroscopy for estimation of soil organic carbon using different spectral preprocessing techniques and multivariate methods in the middle Indo-Gangetic plains of India

Soil organic carbon (SOC) is the main source of soil nutrients, which are essential for the growth and development of agricultural crops. Precise and quick quantification of SOC is of utmost importance in crop husbandry and soil health/carbon sequestration quantification. In order to evaluate visible and near-infrared diffuse reflectance spectroscopy (VIS-NIR) as an alternative to precise and quick method of quantification of SOC in the Indo-Gangetic plains, 280 soil samples were collected covering Inceptisols, Entisols and Alfisols and their spectra recorded. Six preprocessing techniques ((reflectance, absorbance, multiplicative scatter correction (MSC), standard normal variate (SNV), Savitzky–Golay smoothing first derivative (SG-FD) and Savitzky–Golay smoothing second derivative (SG-SD)) and four multivariate methods (partial least-squares regression (PLSR), random forest (RF), support vector regression (SVR) and multivariate adaptive regression splines (MARS)) were evaluated to predict SOC from VIS-NIR spectra. The considerable prediction accuracy and robustness were achieved using the PLSR model (RV2 = 0.73, RMSEV = 0.07, and RPDV = 1.90), RF model (RV2 = 0.69, RMSEV = 0.07, and RPDV = 1.74), SVR model (RV2 = 0.57, RMSEV = 0.08), and RPDV = 1.50), and MARS model (RV2 = 0.63, RMSEV = 0.10, and RPDV = 1.05). Findings from this study identified the reliability of SOC determinations by examining how preprocessing techniques and multivariate methods affect spectral analyses.

Application of Vis-NIR spectroscopy for determination the content of organic matter in saline-alkali soils.

Visible and near-infrared diffuse reflectance spectroscopy (Vis-NIR) has been recognized as a fast method to evaluate the content of soil organic matter (SOM) in various types of soil. The accuracy of Vis-NIR is comparable to conventional laboratory methods for estimating SOM. However, very few studies have applied Vis-NIR to estimate SOM in saline-alkali soil. This study aimed to investigate the efficiency of spectral data for evaluating SOM in saline-alkali soil. Soil samples (n=291) were collected from the five major saline-alkali soil regions in Shaanxi. SOM was measured using standard methods and the samples were scanned using ASD Fieldspec4 at wavelength of 350-2500nm to obtain spectral data. Twenty-six pre-processing methods were tested and partial least squares regression (PLSR) was used to estimate SOM. The best preprocessing was R+SG+SNV+FD. The calibration results were Pc=15, Rc2=0.92, RMSEC=1.11, SEC=1.12, Slope=0.92, Offset=0.45; the validation results were Rv2=0.97, RPD=5.21, RMSEP=0.38, SEP=0.38, Slope=0.97, Offset=0.17. Therefore, this main objective of the study was to propose an effective approach based on Vis-NIR spectroscopy and Chemometrics for predicting saline-alkali SOM contents in the center of Shaanxi, China.

Estimation of andic properties from Vis-NIR diffuse reflectance spectroscopy for volcanic soil classification

Volcanic soils show peculiar characteristics related to the presence of poorly ordered crystalline minerals. These minerals are highly sensitive to specific spectral bands in the visible and near-infrared (Vis-NIR) regions and could be used to establish important relationships between andic properties and soil classification. In order to overcome the expense of traditional soil laboratory analysis and the limitations of pedotransfer functions, this research was designed to test i) the possibility of using diffuse reflectance spectra (DRS) for a supervised chemometric classification of soils with different degrees of andicity, ii) the possibility of using diffuse reflectance spectra (DRS) to estimate the andic properties needed for soil classification (iron and aluminum forms, allophane, phosphate retention, vitric content etc.), and iii) the different multivariate statistical approaches for the analysis of spectra. Diffuse reflectance spectra were measured between 200 and 2500 nm in the laboratory with a Jasco spectrophotometer equipped with an integrated sphere.Chemometric analysis was carried out by discriminant analysis that correctly divided 86% of the horizon samples in 5 classes, representing different levels of expression of the andic properties. Chemometric calibration was obtained by pre-processing Vis-NIR spectra with the application of partial least-squares regression (PLSR). Andic soil attributes were then predicted using calibrated models through PLSR methods and compared with the findings from the Supported Vector Machine (SVM). SVM generally outperformed PLSR on predicting andic properties by Vis-NIR. The most accurate predictive models were obtained for Al and Fe dithionite extracted, both with PLSR (ratio of performance to deviation, RPD 1.9) and SVM (1.7 < RPD < 1.9). The total amount of i) non-crystalline Al and Fe forms (oxalate extractable, Alo, Feo), ii) Alo + 1/2Feo (1.7 < RPD < 1.8), iii) non-crystalline Si (Sio) (RPD 1.5), and iv) Al associated with organic matter (sodium pyrophosphate extractable, Alp; RPD 1.5), Alo-Alp (RPD 1.5), allowed a first soil Andosol qualification, when obtained through SVM. Therefore, the results of this research highlight that Vis-NIR can be used as a preliminary analysis to estimate diagnostic parameters of Andosols. Results indicate that Vis-NIR is a promising technique for volcanic soil classification according to soil properties (especially Fe and Al forms) and horizons, which can be used in lieu of complex chemical and physical analyses involved in routine soil profile classification. Further studies should confirm Vis-NIR ability to discriminate Andosols at more detailed classification levels.

Estimating Soil Salt Content Using Fractional Derivatives and Optional Spectral Indices in the Ebinur Lake Oasis, Northwestern China

Derivative spectroscopy is a powerful mathematical tool that provides more useful information of spectral data than untreated data. Visible and near-infrared diffuse reflectance spectroscopy (VIS-NIR) has shown levels of accuracy comparable to conventional laboratory methods for estimating soil properties. The VIS-NIR spectrum is one of the most important data acquisition technologies for digital soil mapping, precision agriculture, and soil resource surveys. The objective of this study was to develop a soil salt content model using the fractional-order derivatives of field-measured spectral data paired with ground measurements. The results showed the following: There was a significant correlation between the single-band reflectance spectra and the soil salt content, and as the derivative order increased, reflectance values first increased and then decreased, with a peak value, R = 0.525, in the 1.2-order derivative. Strong correlations between the estimation of soil salt content and the derivative spectral data of the difference index (DI), ratio index (RI), and normalized difference index (NDI) were also produced by the 1.2-order derivative. R values for the DI, RI, and NDI were 0.818, 0.8624, and 0.8297, respectively. Selected bands with the highest correlation values from the 1.2-order derivative spectral data of the R1428, DI (R1426 / R2151), RI (R1429 − R2024), and NDI (R1526 − R2470)/(R1526 + R2470)] were used to build a model to estimate soil salt content, which showed an R2 value of 0.53 and low RMSE and SD values. The 15 samples of the data were then used to validate the single-band model, R1428, and the combination models, DI (R1426 / R2151), RI (R1429 − R2024), and NDI [(R1526 − R2470)/(R1526 + R2470)], which demonstrated that the R2 values were greater than 0.8. The RMSE and SD values were low, and the value of the RPD was greater than 1.4. This study demonstrated the potential of using fractional derivatives rather than integer derivatives for soil salt content estimation. By making full use of hyperspectral data, fractional derivatives could enrich data preprocessing methods and unearth information about spectral emissions that are lost by integer derivatives. Although this study is a straightforward application of the fractional derivative method, it provides a reference for the estimation of other parameters with hyperspectral technology.

Visible and Near-Infrared Diffuse Reflectance Spectroscopy for Fast Qualitative and Quantitative Assessment of Nectarine Quality

Visible and near-infrared spectroscopy has been widely used as a non-invasive and rapid-assessment technique for the quality control of agricultural products. In this study, 325 samples of nectarines representing two commercial varieties, cv. ‘Big Top’ and cv. ‘Magique’, were analysed by visible and near-infrared diffuse reflectance spectroscopy (VIS-NIR). The spectral data were pre-treated and analysed to predict the internal quality of the samples and to discriminate between the two varieties. Good prediction of the internal quality of the samples, using partial least-squares regressions, was observed for both (R 2 P of 0.909 and 0.927 and RMSEP of 0.235 and 0.238 for cv. Big Top and Magique, respectively). Discriminant models, using linear discriminant and partial least-squares discriminant analyses, were built to classify the nectarines. Both methods provided good results with rates of 97.44 and 100% of correctly classified samples. The results indicated that visible and near-infrared techniques can be useful and simple methods for quality control and for the correct identification of nectarines in commercial lines as an alternative to the slower and less accurate manual classification.

Prediction of Soil Physical and Chemical Properties by Visible and Near-Infrared Diffuse Reflectance Spectroscopy in the Central Amazon

Visible and near-infrared diffuse reflectance spectroscopy (VIS-NIR) has shown levels of accuracy comparable to conventional laboratory methods for estimating soil properties. Soil chemical and physical properties have been predicted by reflectance spectroscopy successfully on subtropical and temperate soils, whereas soils from tropical agro-forest regions have received less attention, especially those from tropical rainforests. A spectral characterization provides a proficient pathway for soil characterization. The first step in this process is to develop a comprehensive VIS-NIR soil library of multiple key soil properties to be used in future soil surveys. This paper presents the first VIS-NIR soil library for a remote region in the Central Amazon. We evaluated the performance of VIS-NIR for the prediction of soil properties in the Central Amazon, Brazil. Soil properties measured and predicted were: pH, Ca, Mg, Al, H, H+Al, P, organic C (SOC), sum of bases, cation exchange capacity (CEC), percentage of base saturation (V), Al saturation (m), clay, sand, silt, silt/clay (S/C), and degree of flocculation. Soil samples were scanned in the laboratory in the VIS-NIR range (350–2500 nm), and forty-one pre-processing methods were tested to improve predictions. Clay content was predicted with the highest accuracy, followed by SOC. Sand, S/C, H, Al, H+Al, CEC, m and V predictions were reasonably good. The other soil properties were poorly predicted. Among the soil properties predicted well, SOC is one of the critical soil indicators in the global carbon cycle. Besides the soil property of interest, the landscape position, soil order and depth influenced in the model performance. For silt content, pH and S/C, the model performed better in well-drained soils, whereas for SOC best predictions were obtained in poorly drained soils. The association of VIS-NIR spectral data to landforms, vegetation classes, and soil types demonstrate potential for soil characterization.

Consideration of peak parameters derived from continuum-removed spectra to predict extractable nutrients in soils with visible and near-infrared diffuse reflectance spectroscopy (VNIR-DRS)

Visible and near-infrared diffuse reflectance spectroscopy (VNIR-DRS) provides a rapid and inexpensive tool for simultaneous prediction of a variety of different soil properties. Usually, some sophisticated multivariate mathematical or statistical methods are employed in order to extract the required information from the raw spectrum scan. For this purpose especially the partial least squares regression (PLSR) is the most frequently used algorithm. This method generally benefits from complexity, with which the soil spectra are treated. But interestingly, also techniques which focus on only one specific spectral feature, such as simple linear regression dealing with single continuum-removed spectra (CRS) value at selected wavelength, can often provide competitive results too. Such methods rely on known spectral signature of spectrally active soil components. In this study focusing on laboratory soil spectroscopy, we attempted to enhance the potential of CRS by taking into account all possible peaks as derived from CRS and relating their basic parameters, i.e. area, width and depth to soil properties employing the multiple linear regression (MLR) technique. On top of that comparison to PLSR was performed to evaluate the ability of the presented method. Nine measured soil properties on total 97 topsoil samples, were Mehlich 3 extractable elements Ca, Cu, Fe, K, Mg, Mn, P, and Zn and soil pH in CaCl2 extract. In seven cases (Ca, Cu, Fe, Mn, P, Zn and pH), of which three (Ca, Cu and Zn) were predicted reliably accurately (0.50<R2cv<0.80) and the rest four (Fe, Mn, P and pH) only poorly (R2cv<0.50), better results (the differences in R2cv up to 0.1) were obtained with the presented methodology compared against PLSR. For K and Mg, it was clear that K was predicted accurately while the prediction of Mg was not satisfactory, slightly better results (the differences in R2cv were 0.02 and 0.05, respectively) were achieved with PLSR against the presented method. We further concluded that content of clay, soil organic matter (SOM), and soil color were the main driving forces behind the prediction using soil spectroscopy in this particular case. The study indicated that MLR based on CRS peak parameters could be an alternative method in quantitative prediction of different soil properties using VNIR-DRS.

Visible and Near-Infrared Diffuse Reflectance Spectroscopy for Prediction of Soil Properties near a Copper Smelter

Spatial and temporal monitoring of soil properties in smelting regions requires collection of a large number of samples followed by laboratory cumbersome and time-consuming measurements. Visible and near-infrared diffuse reflectance spectroscopy (VNIR-DRS) provides a rapid and inexpensive tool to predict various soil properties simultaneously. This study evaluated the suitability of VNIR-DRS for predicting soil properties, including organic matter (OM), pH, and heavy metals (Cu, Pb, Zn, Cd, and Fe), using a total of 254 samples collected in soil profiles near a large copper smelter in China. Partial least square regression (PLSR) with cross-validation was used to relate soil property data to the reflectance spectral data by applying different preprocessing strategies. The performance of VNIR-DRS calibration models was evaluated using the coefficient of determination in cross-validation (R2cv) and the ratio of standard deviation to the root mean standard error of cross-validation (SD/RMSEcv). The models provided fairly accurate predictions for OM and Fe (R2cv > 0.80, SD/RMSEcv > 2.00), less accurate but acceptable for screening purposes for pH, Cu, Pb, and Cd (0.50 < R2cv < 0.80, 1.40 < SD/RMSEcv < 2.00), and poor accuracy for Zn (R2CV < 0.50, SD/RMSEcv < 1.40). Because soil properties in contaminated areas generally show large variation, a comparative large number of calibrating samples, which are variable enough and uniformly distributed, are necessary to create more accurate and robust VNIR-DRS calibration models. This study indicated that VNIR-DRS technique combined with continuously enriched soil spectral library could be a nondestructive alternative for soil environment monitoring.

Comparison of soil reflectance spectra and calibration models obtained using multiple spectrometers

In the literature of visible and near infrared diffuse reflectance spectroscopy (VNIR-DRS) for soil characterization, the effects of instruments and scanning environments on reflectance spectra and calibration models have not been well documented. To fill this knowledge gap, the goal of this study is to compare soil reflectance spectra and calibration models obtained with different spectrometers operated under different lab environments. Two sets of soil samples were used in this study. The first set (containing 180 samples collected from Quemado, Texas) was scanned by three spectrometers and no effort was provided to control the scanning protocol. The second set (containing 264 samples from Central Texas) was scanned by two spectrometers, and efforts were provided intentionally to control the scanning protocol. Partial least squares regression was applied to develop calibration models for soil organic carbon (OC) content using the first derivative spectra. Three calibration transfer methods (namely, slope and bias correction, direct standardization, and piecewise direct standardization) were used to transfer calibration models from one instrument to another. In the experiment where no scanning control was provided, significant differences were seen in mean soil spectra by different spectrometers. But cross-validation indicated that all three models can predict OC accurately. However, the OC models are quite dissimilar to each other in terms of the regression coefficients at each wavelength, and their application to the spectra measured by other instruments generally yielded poor results. All three calibration transfer methods provided a satisfactory application of the OC model calibrated on the primary instrument to secondary instruments. In the experiment where some controls were provided, mean soil spectra by different spectrometers matched each other well. The two OC models were quite similar, and model application to the spectra measured by the other instrument yielded satisfactory predictions. When scanning control was provided; however, model transfer methods improved the calibration model only marginally. All results indicate that VNIR-DRS calibration models are highly instrument/scanning environment dependent, and their extent of applicability could be highly limited. Provision of controls over the scanning protocol has the potential to remove a great deal of spectral variations that are related to extraneous effects due to multiple instruments/scanning environment. The results of this study have important implications on the future use of VNIR-DRS as a routine method for soil characterization, such as comparisons among VNIR prediction models derived from different soil labs and a global soil spectral library, where multiple instruments have to be involved.

Global soil characterization with VNIR diffuse reflectance spectroscopy

There has been growing interest in the use of diffuse infrared reflectance as a quick, inexpensive tool for soil characterization. In studies reported to date, calibration and validation samples have been collected at either a local or regional scale. For this study, we selected 3768 samples from all 50 U.S. states and two tropical territories and an additional 416 samples from 36 different countries in Africa (125), Asia (104), the Americas (75) and Europe (112). The samples were selected from the National Soil Survey Center archives in Lincoln, NE, USA, with only one sample per pedon and a weighted random sampling to maximize compositional diversity. Applying visible and near-infrared (VNIR) diffuse reflectance spectroscopy (DRS) to air-dry soil (< 2 mm) with auxiliary predictors including sand content or pH, we obtained validation root mean squared deviation (RMSD) estimates of 54 g kg − 1 for clay, 7.9 g kg − 1 for soil organic C (SOC), 5.6 g kg − 1 for inorganic C (IC), 8.9 g kg − 1 for dithionate–citrate extractable Fe (FEd), and 5.5 cmol c kg − 1 for cation exchange capacity (CEC) with NH 4 at pH = 7. For all of these properties, boosted regression trees (BRT) outperformed PLS regression, suggesting that this might be a preferred method for VNIR-DRS soil characterization. Using BRT, we were also able to predict ordinal clay mineralogy levels for montmorillonite and kaolinite, with 88% and 96%, respectively, falling within one ordinal unit of reference X-ray diffraction (XRD) values (0–5 on ordinal scale). Given the amount of information obtained in this study with ∼4 × 10 3 samples, we anticipate that calibrations sufficient for many applications might be obtained with large but obtainable soil-spectral libraries (perhaps 10 4–10 5 samples). The use of auxiliary predictors (potentially from complementary sensors), supplemental local calibration samples and theoretical spectroscopy all have the potential to improve predictions. Our findings suggest that VNIR soil characterization has the potential to replace or augment standard soil characterization techniques where rapid and inexpensive analysis is required.