Prediction of soil organic matter using a spatially constrained local partial least squares regression and the Chinese vis–NIR spectral library

We need to determine the best use of soil vis–NIR spectral libraries that are being developed at regional, national and global scales to predict soil properties from new spectral readings. To reduce the complexity of a calibration dataset derived from the Chinese vis–NIR soil spectral library (CSSL), we tested a local regression method that combined geographical sub‐setting with a local partial least squares regression (local‐PLSR) that uses a limited number of similar vis–NIR spectra (k‐nearest neighbours). The central idea of the local regression, and of other local statistical approaches, is to derive a local prediction model by identifying samples in the calibration dataset that are similar, in spectral variable space, to the samples used for prediction. Here, to derive our local regressions we used Euclidean distance in spectral space between the calibration dataset and prediction samples, and we also used soil geographical zoning to account for similarities in soil‐forming conditions. We tested this approach with the CSSL, which comprised 2732 soil samples collected from 20 provinces in the People's Republic of China to predict soil organic matter (SOM). Results showed that the prediction accuracy of our spatially constrained local‐PLSR method (R² = 0.74, RPIQ = 2.6) was better than that from local‐PLSR (R² = 0.69, RPIQ = 2.3) and PLSR alone (R² = 0.50, RPIQ = 1.5). The coupling of a local‐PLSR regression with soil geographical zoning can improve the accuracy of local SOM predictions using large, complex soil spectral libraries. The approach might be embedded into vis–NIR sensors for laboratory analysis or field estimation.     

基于偏最小二乘回归的土壤有机质含量高光谱估算

为实现基于光谱分析土壤有机质含量的快速测定,该文以江汉平原公安县的土壤为研究对象,进行室内理化分析、光谱测量与处理等一系列工作,在土壤原始光谱反射率(raw spectral reflectance,R)的基础上,提取了其倒数之对数(inverse-log reflectance,LR)、一阶微分(first order differential reflectance,FDR)和连续统去除(continuum removal,CR)3种光谱指标,分析4种不同形式的光谱指标与有机质含量的相关性,对相关系数进行P=0.01水平上的显著性检验来确定显著性波段的范围,并基于全波段(400~2 400 nm)和显著性波段运用偏最小二乘回归(partial least squares regression,PLSR)建立了该区域土壤有机质高光谱的预测模型,通过模型精度的比较确定最优模型。结果表明,进行CR变换后,光谱曲线的特征吸收带更加明显,相关系数在可见光波段范围内有所提高;基于全波段的PLSR建模效果要优于显著性波段,其中以CR的预测精度最为突出,其模型的决定系数R2和相对分析误差RPD分别为0.84、2.58;显著性波段的PLSR模型与全波段对比在模型精度方面虽有一定差距,但从模型的复杂程度来比较,具有模型简单、运算量小、变量更少的特点;最后,综合比较了全波段和显著性波段4种光谱指标的反演精度,发现CR-PLSR模型的建模和预测的效果比R-PLSR、LR-PLSR、FDR-PLSR模型都要显著。该研究可为将CR-PLSR高光谱反演模型用于该区域土肥信息的遥感监测提供参考。     

基于行走式近红外反射光谱和偏最小二乘法的冰碛土中土壤碳和粘粒含量制图研究

Characterizing spatial variability of soil attributes, using traditional soil sampling and laboratory analysis, is cost prohibitive. The potential benefit of managing soils on a site-specific basis is well established. High variations in glacial till soil render detailed soil mapping difficult with limited number of soil samples. To overcome this problem, this paper demonstrates the feasibility of soil carbon and clay mapping using the newly developed on-the-go near-infrared reflectance spectroscopy (NIRS). Compared with the geostatistics method, the partial least squares regression (PLSR), with NIRS measurements, could yield a more detailed map for both soil carbon and clay. Further, by using independent validation dataset, the accuracy of predicting could be improved significantly for soil clay content and only slightly for soil carbon content. Owing to the complexity of field conditions, more work on data processing and calibration modeling might be necessary for using on-the-go NIRS measurements.     

Carbon isotopes of water‐extractable organic carbon in a depth profile of forest soil imply a dynamic relationship with soil carbon

Although considerable research has been conducted on the importance of recent litter compared with older soil organic matter as sources of dissolved organic carbon (DOC) in forest soils, a more thorough evaluation of this mechanism is necessary. We studied water‐extractable organic carbon (WEOC) in a soil profile under a cool‐temperate beech forest by analysing the isotopic composition (¹³C and ¹⁴C) of WEOC and its fractions after separation on a DAX‐8 resin. With depth, WEOC became more enriched in ¹³C, which reflects the increasing proportion of the hydrophilic, isotopically heavier fraction. The ¹⁴C content in WEOC and its fractions decreased with depth, paralleling the ¹⁴C trend in soil organic matter (SOM). These results indicate a dynamic equilibrium of WEOC and soil organic carbon. The dominant process maintaining the WEOC pool in the mineral soil appears to be the microbial release of water‐soluble compounds from the SOM, which alters in time‐scales of decades to centuries.     

Cold and hot water–extractable organic matter as indicators of litter decomposition in forest soils

Mild extractions were used as indicators of easily decomposable organic matter (OM). However, the chemical composition of extracted OM often remained unclear. Therefore, the composition of cold and hot water–extractable OM was investigated in the O horizons (Oi, Oe, Oa) of a 170 y old beech stand (Fagus sylvatica) in the Ore Mtns., SE Germany. To simulate litter decomposition, the O horizon samples were incubated for 1 week under defined conditions. Cold‐ and hot‐water extracts were analyzed and chemically characterized by pyrolysis–field ionization mass spectrometry (Py‐FIMS). The C and N concentrations were always lower in the cold‐(C: 2.69 to 3.95 g kg^–1; N: 0.14 to 0.29 g kg^–1) than in the hot‐water extracts (C: 13.77 to 15.51 g kg^–1; N: 0.34 to 0.83 g kg^–1). The C : N ratios of both extracts increased with increasing depth. Incubation increased the concentrations of C and N in all water extracts, while C : N ratios of extracts decreased. The molecular‐chemical composition of cold and hot water–extracted OM revealed distinct differences. Generally, cold water–extracted OM was thermally more stable than hot water–extracted OM. The mass spectra of the hot water–extracted organic matter revealed more intensive signals of carbohydrates, phenols, and lignin monomers. Additionally, the n‐C₂₈ fatty acid and the n‐C₃₈–to–n‐C₅₂ alkyl monoesters clearly distinguished the hot‐ from the cold‐water extract. A principle‐component analysis visualized (1) alterations in the molecular‐chemical composition of cold‐ and hot‐water extracts due to previous incubation of the solid O horizon samples and (2) a decomposition from the Oi to the Oh horizon. This provides evidence that the macromorphological litter decomposition was reflected by the chemical composition of water extracts, and that Py‐FIMS is well‐suited to explain at the molecular level why OM decomposability is correlated with water‐extracted C.     

基于可见光-近红外漫反射光谱的红壤有机质预测及其最优波段选择

有机质是土壤肥力、土壤质量的重要指标,同时也是土壤碳库的重要组成部分。传统的土壤有机质测试方法费时、繁琐,难以满足快速监测土壤有机质含量的需求。近年来,具有无损、快速、简便等优点的可见-近红外光谱技术的应用为土壤有机质的快速监测提供了有效途径。     

Kinetics of water‐extractable zinc release from seaweed (Fucus serratus) as soil amendment

Soil fertilization with trace‐metal rich organic fertilizers such as Fucus serratus seaweed may be an effective way to combat micronutrient deficiency. In this study the kinetics of zinc release from Fucus serratus seaweed was investigated in a packed soil column leaching experiment over 1,776 h. The release of zinc from control (soil only) and treatment (soil + seaweed; equivalent zinc application rate of 1.42 kg ha^?1) columns, measured by ICP‐MS, demonstrated two distinct release stages. The cumulative zinc release data for each phase were fitted to five kinetic models: zero order, first order, Elovich, power function and parabolic diffusion. In the first stage (0–400 hours) the release of zinc from both control and treatment was best described by a parabolic rate law, indicating release of zinc from a soluble soil reservoir. In the second stage (400–1,776 h) zinc release followed a zero order rate law indicative of slow release from an essentially insoluble reservoir. The modelled difference between the amount of zinc released from treatment and control columns in stage 1 (230 ± 11 µg) represented the total amount of zinc added via seaweed. The parabolic rate constant for seaweed zinc release was 12.09 µg g^?1 h^?0.5. In summary, the addition of F. serratus to soil is a viable source of labile zinc and a low cost agronomic option for mitigating zinc deficiency in soils.     

Sample preparation and selection for qualitative and quantitative analyses of soil organic carbon with mid‐infrared reflectance spectroscopy

Mid‐infrared spectroscopy (MIRS) is a well‐established analytical tool for qualitative and quantitative analysis of soil samples. However, effects of soil sample grinding procedures on the prediction accuracy of MIR models and on qualitative spectral information have not been well investigated and, in consequence, not standardized up to now. Further, the effects of soil sample selection on the accuracy of MIR prediction models has not been quantified yet. This study investigated these effects by using 180 well‐characterized soil samples that were ground for different times (0, 2 or 4 minutes) and then used for MIR measurements. To study the impact of sample preparation, soil spectra were subjected to principal component analyses (PCA), multiple regression and partial least square (PLS) analysis. The results indicate that the prediction accuracy of MIR models for soil organic carbon (SOC) and pH and the qualitative spectral information were better overall for lightly ground (2 minutes) soil samples compared with intensively (4 minutes) or unground soil samples. Whereas the grinding procedure did not show any effect on spectra of clay minerals, spectral information for quartz and for SOC was modified. Even though it is difficult to recommend a global standardized soil sample grinding procedure for MIR measurements because of different mill types available within laboratories, we highly recommend using an internally standardized grinding procedure. Moreover, we show that neither land use nor soil sampling depth influences the prediction of the SOC content. However, sand and clay content substantially affect the score vectors used by the PLS algorithm to predict the SOC content. Thus, we recommend using soil samples similar in texture for more precise SOC calibration models for MIR spectroscopy.     

Application of diffuse reflectance FT-IR spectroscopy and partial least-squares regression to predict NMR properties of soil organic matter

Carbon 13 nuclear magnetic resonance spectroscopy (¹³C NMR) is a powerful technique for studying the structure and turnover of soil organic matter, but is time consuming and expensive. It is therefore worth seeking swifter and cheaper methods. Diffuse reflectance FT‐IR spectroscopy (DRIFT), along with partial least squares (PLS) algorithms, provides statistical models to quantify soil properties, such as contents of C, N and clay. I have applied DRIFT?PLS to quantify soil organic C species, as measured by solid state ¹³C NMR spectroscopy, for several bulk soils and physical soil fractions. Calibration and prediction models for organic C and for particular NMR regions, namely alkyl C, O?alkyl C and carboxyl C, attained R² values of between 0.94 and 0.98 (calibration) and 0.70–0.93 (cross‐validation). The prediction of unknown soil samples, after pre‐selection by statistical indices, confirmed the applicability of DRIFT?PLS. The prediction of aromatic C failed, probably because of superimposition of aromatic bands by signals from minerals. Results from fractions of particulate organic matter suggest that the chemical homogeneity of the material hampers the quantification of its constituting C species by DRIFT?PLS. For alkyl C, prediction of carbon species by DRIFT?PLS was better than direct peak‐area quantification in the IR spectra, but advantageous in parts only compared with a linear model correlating C species with soil C contents. In conclusion, DRIFT?PLS calibrated with NMR data provides quantitative information on the composition of soil organic matter and can therefore complement structural studies by its application to large numbers of samples. However, it cannot replace the information provided by more specific methods. The actual potential of DRIFT?PLS lies in its capacity to predict unknown samples, which is helpful for classification and identification of environmental outliers or benchmarks.     

Determination of extractable sulfate and total sulfur from plant and soil material by an autoanalyzer

Abstract

The method has been modified from Sinclair^3,4 and tested for extractable sulfate and total sulfur on plant, soil, and water samples. Inorganic sulfur is extracted from plant and soil material by using dilute acidic extractants, and total sulfur is estimated from dry ashed or wet ashed material whereby various sulfur forms are oxidized to sulfate‐sulfur. The sulfate is precipitated in sample solutions as barium sulfate and determined turbidimetrically by AutoAnalyzer.

The method is rapid, precise, and sensitive enough to be used on a routine basis.     

Impacts of land‐use intensity on soil organic carbon content,soil structure and water‐holding capacity

The impact of land‐use intensity is evaluated through changes in the soil properties in different areas of the traditional central Spanish landscape. Soil organic carbon (SOC) content, bulk density, aggregate stability and water‐holding capacity (WHC) in the topsoil of active and abandoned vineyards, livestock routes (LR) and young Quercus afforested areas were analysed. These different types of land use can be interpreted as having a gradient of progressively less impact on soil functions or conservation. As soil use intensity declines, there is an increase in SOC content (from 0.2 to 0.6%), WHC (from 0.2 to 0.3 g H₂O per g soil) and aggregate stability (from 4 to 33 drop impacts). Soils beneath vines have lost their upper horizon (15 cm depth) because of centuries‐old tillage management of vineyards. Except for an increase in bulk density (from 1.2 to 1.4 g/cm³), there were no differences in soil characteristics 4 yr after the abandonment of vine management. LR can be considered sustainable uses of land, which preserve or improve soil characteristics, as there were no significant differences between topsoil from LR and that from a 40‐yr‐old Quercus afforested area. SOC content, one of the main indicators for soil conservation, is considered very low in every case analysed, even in the more conservative uses of land. These data can be useful in understanding the slow rate of recovery of soils, even after long‐term cessation of agricultural land use.     

Evaluating a low‐cost portable NIR spectrometer for the prediction of soil organic and total carbon using different calibration models

This study aims to assess the performance of a low‐cost, micro‐electromechanical system‐based, near infrared spectrometer for soil organic carbon (OC) and total carbon (TC) estimation. TC was measured on 151 soil profiles up to the depth of 1 m in NSW, Australia, and from which a subset of 24 soil profiles were measured for OC. Two commercial spectrometers including the AgriSpec^TM (ASD) and NeoSpectra^TM (Neospectra) with spectral wavelength ranges of 350–2,500 and 1,300–2,500 nm, respectively, were used to scan the soil samples, according to the standard contact probe protocol. Savitzky–Golay smoothing filter and standard normal variate (SNV) transformation were performed on the spectral data for noise reduction and baseline correction. Three calibration models, including Cubist tree model, partial least squares regression (PLSR) and support vector machine (SVM), were assessed for the prediction of soil OC and TC using spectral data. A 10‐fold cross‐validation analysis was performed for evaluation of the models and devices accuracies. Results showed that Cubist model predicts OC and TC more accurately than PLSR and SVM. For OC prediction, Cubist showed R² = 0.89 (RMSE = 0.12%) and R² = 0.78 (RMSE = 0.16%) using ASD and NeoSpectra, respectively. For TC prediction, Cubist produced R² = 0.75 (RMSE = 0.45%) and R² = 0.70 (RMSE = 0.50%) using ASD and NeoSpectra, respectively. ASD performed better than NeoSpectra. However, the low‐cost NeoSpectra predictions were comparable to the ASD. These finding can be helpful for more efficient future spectroscopic prediction of soil OC and TC with less costly devices.     

Assessment of soil organic carbon at local scale with spiked NIR calibrations: effects of selection and extra‐weighting on the spiking subset

Spiking is a useful approach to improve the accuracy of regional or national calibrations when they are used to predict at local scales. To do this, a small subset of local samples (spiking subset) is added to recalibrate the initial calibration. If the spiking subset is small in comparison with the size of the initial calibration set, then it could have little noticeable effect and a small improvement can be expected. For these reasons, we hypothesized that the accuracy of the spiked calibrations can be improved when the spiking subset is extra‐weighted. We also hypothesized that the spiking subset selection and the initial calibration size could affect the accuracy of the recalibrated models. To test these hypotheses, we evaluated different strategies to select the best spiking subset, with and without extra‐weighting, to spike three different‐sized initial calibrations. These calibrations were used to predict the soil organic carbon (SOC) content in samples from four target sites. Our results confirmed that spiking improved the prediction accuracy of the initial calibrations, with any differences depending on the spiking subset used. The best results were obtained when the spiking subset contained local samples evenly distributed in the spectral space, regardless of the initial calibration's characteristics. The accuracy was improved significantly when the spiking subset was extra‐weighted. For medium‐ and large‐sized initial calibrations, the improvement from extra‐weighting was larger than that caused by the increase in spiking subset size. Similar accuracies were obtained using small‐ and large‐sized calibrations, suggesting that incipient spectral libraries could be useful if the spiking subset is properly selected and extra‐weighted. When small‐sized spiking subsets were used, the predictions were more accurate than those obtained with ‘geographically‐local’ models. Overall, our results indicate that we can minimize the efforts needed to use near‐infrared (NIR) spectroscopy effectively for SOC assessment at local scales.     

Calcite and gypsum solubility products in water‐saturated salt‐affected soil samples at 25°C and at least up to 14 dS m−1

Calcite and gypsum are salts of major ions characterized by poor solubility compared with other salts that may precipitate in soils. Knowledge of calcite and gypsum solubility products in water‐saturated soil samples substantially contributes to a better assessment of processes involved in soil salinity. The new SALSOLCHEMIS code for chemical equilibrium assessment was parameterized with published analytical data for aqueous synthetic calcite and gypsum‐saturated solutions. Once parameterized, SALSOLCHEMIS was applied to calculations of the ionic activity products of calcium carbonate and calcium sulphate in 133 water‐saturated soil samples from an irrigated salt‐affected agricultural area in a semi‐arid Mediterranean climate. During parameterization, sufficiently constant values for the ionic activity products of calcium carbonate and calcium sulphate were obtained only when the following were used in SALSOLCHEMIS: (i) the equations of Sposito & Traina for the free ion activity coefficient calculation, (ii) the assumption of the non‐existence of the Ca (HCO ₃)⁺ and CaCO₃^o ion pairs and (iii) a paradigm of total ion activity coefficients. The value of 4.62 can be assumed to be a reliable gypsum solubility product (pKs) in simple aqueous and soil solutions, while a value of 8.43 can only be assumed as a reliable calcite solubility product (pKs) in simple aqueous solutions. The saturated pastes and saturation extracts were found to be calcite over‐saturated, with the former significantly being less so (p IAP = 8.29) than the latter (p IAP = 8.22). The calcite over‐saturation of saturated pastes increased with the soil organic matter content. Nevertheless, the inhibition of calcite precipitation is caused by the soluble organic matter from a dissolved organic carbon threshold value that lies between 7 and 12 mm . The hypothesis of thermodynamic equilibrium is more adequate for the saturated pastes than for the saturation extracts.     

Use of mid‐infrared spectroscopy in the diffuse‐reflectance mode for the prediction of the composition of organic matter in soil and litter

Mid‐infrared spectroscopy (MIRS) is assumed to be superior to near‐infrared spectroscopy (NIRS) for the prediction of soil constituents, but its usefulness is still not sufficiently explored. The objective of this study was to evaluate the ability of MIRS to predict the chemical and biological properties of organic matter in soils and litter. Reflectance spectra of the mid‐infrared region including part of the near‐infrared region (7000–400 cm^–1) were recorded for 56 soil and litter samples from agricultural and forest sites. Spectra were used to predict general and biological characteristics of the samples as well as the C composition which was measured by ¹³C CPMAS‐NMR spectroscopy. A partial least‐square method and cross‐validation were used to develop equations for the different constituents over selected spectra ranges after several mathematical treatments of the spectra. Mid‐infrared spectroscopy predicted well the C : N ratio: the modeling efficiency EF was 0.95, the regression coefficient (a) of a linear regression (measured against predicted values) was 1.0, and the correlation coefficient (r) was 0.98. Satisfactorily (EF ≥ 0.70, 0.8 ≤ a ≤ 1.2, r ≥ 0.80) assessed were the contents of C, N, and lignin, the production of dissolved organic carbon, and the contents of carbonyl C, aromatic C, O‐alkyl C, and alkyl C. However, the N mineralization rate, the microbial biomass and the alkyl–to–aromatic C ratio were predicted less satisfactorily (EF < 0.70). Limiting the sample set to mineral soils did generally not result in improved predictions. The good and satisfactory predictions reported above indicate a marked usefulness of MIRS in the assessment of chemical characteristics of soils and litter, but the accuracies of the MIRS predictions in the diffuse‐reflectance mode were generally not superior to those of NIRS.     

Nodulation and root growth increase in lower soil layers of water‐limited faba bean intercropped with wheat

Below‐ground niche complementarity in legume–cereal intercrops may improve resource use efficiency and root adaptability to environmental constraints. However, the effect of water limitation on legume rooting and nodulation patterns in intercropping is poorly understood. To advance our knowledge of mechanisms involved in water‐limitation response, faba bean (Vicia faba L.) and wheat (Triticum aestivum L.) were grown as mono‐ and intercrops in soil‐filled plexiglass rhizoboxes under water sufficiency (80% of water‐holding capacity) and water limitation (30% of water‐holding capacity). We examined whether intercropping facilitates below‐ground niche complementarity under water limitation via interspecific root stratification coupled with modified nodulation patterns. While no significant treatment effects were measured in intercropped wheat growth parameters, water limitation induced a decrease in shoot and root biomass of monocropped wheat. Likewise, shoot biomass and height, and root length of monocropped faba bean significantly decreased under water limitation. Conversely, water limitation stimulated root biomass of intercropped faba bean in the lower soil layer (15–30 cm soil depth). Similarly, total nodule number of faba bean roots as well as nodule number in the lower soil layer increased under intercropping regardless of water availability. Under water limitation, intercropping also led to a significant increased nodule biomass (48%) in the lower soil layer as compared to monocropping. The enhanced nodulation in the lower soil layer and the associated increase in root and shoot growth provides evidence for a shift in niche occupancy when intercropped with wheat, which improves water‐limited faba bean performance.     

Proximal soil sensing of soil texture and organic matter with a prototype portable mid‐infrared spectrometer

Recent advances in semiconductor technologies have given rise to the development of mid‐infrared (mid‐IR) spectrometers that are compact, relatively inexpensive, robust and suitable for in situ proximal soil sensing. The objectives of this research were to evaluate a prototype portable mid‐IR spectrometer for direct measurements of soil reflectance and to model the spectra to predict sand, clay and soil organic matter (SOM) contents under a range of field soil water conditions. Soil samples were collected from 23 locations at different depths in four agricultural fields to represent a range of soil textures, from sands to clay loams. The particle size distribution and SOM content of 48 soil samples were measured in the laboratory by conventional analytical methods. In addition to air‐dry soil, each sample was wetted with two different amounts of water before the spectroscopic measurements were made. The prototype spectrometer was used to measure reflectance (R) in the range between 1811 and 898 cm^?1 (approximately 5522 to 11 136 nm). The spectroscopic measurements were recorded randomly and in triplicate, resulting in a total of 432 reflectance spectra (48 samples × three soil water contents × three replicates). The spectra were transformed to log₁₀ (1/R) and mean centred for the multivariate statistical analyses. The 48 samples were split randomly into a calibration set (70%) and a validation set (30%). A partial least squares regression (PLSR) was used to develop spectroscopic calibrations to predict sand, clay and SOM contents. Results show that the portable spectrometer can be used with PLSR to predict clay and sand contents of either wet or dry soil samples with a root mean square error (RMSE) of around 10%. Predictions of SOM content resulted in RMSE values that ranged between 0.76 and 2.24%.     

Improving the prediction performance of a large tropical vis‐NIR spectroscopic soil library from Brazil by clustering into smaller subsets or use of data mining calibration techniques

Effective agricultural planning requires basic soil information. In recent decades visible near‐infrared diffuse reflectance spectroscopy (vis‐NIR) has been shown to be a viable alternative for rapidly analysing soil properties. We studied 7172 samples of seven different soil types collected from several regions of Brazil and varying in organic matter (OM) (0.2–10.3%) and clay content (0.2–99.0%). The aim was to explore the possibility of enhancing the performance of vis‐NIR data in predicting organic matter and clay content in this library by dividing it into smaller sub‐libraries on the basis of their vis‐NIR spectra. We used partial least square regression (PLSR) models on the sub‐libraries and compared the results with PLSR and two non‐linear calibration techniques, boosted regression trees (BT) and support vector machines (SVM) applied to the whole library. The whole library calibrations for clay performed well (ME (modelling efficiency) > 0.82; RMSE (root mean squared error) < 10.9%), reflecting the influence of the direct spectral responses of this property in the vis‐NIR range. Calibrations for OM were reasonably good, especially in view of the very small variation in this property (ME > 0.60; RMSE < 0.55%). The best results were, however, found when dividing the large library into smaller subsets by using variation in the mean‐normalized or first derivative spectra. This divided the global data set into clusters that were more uniform in mineralogy, regardless of geographical origin, and improved predictive performance. The best clustering method improved the RMSE in the validation to 8.6% clay and 0.47% OM, which corresponds to a 21% and 15% reduction, respectively, as compared with whole library PLSR. For the whole library, SVM performed almost equally well, reducing RMSE to 8.9% clay and 0.48% OM.