Near infrared spectroscopy as an easy and precise method to estimate soil texture

doi:10.1016/j.geoderma.2018.10.038

Geoderma

Volume 337, 1 March 2019, Pages 906-913

https://doi.org/10.1016/j.geoderma.2018.10.038 Get rights and content

Highlights

•
10,800 near infrared spectra from soil samples in the German Agricultural Soil Inventory were used.
•
Near-infrared spectroscopy can predict soil texture at national/regional scales with high precision.
•
Log-ratio transformation meets the constraint: ∑Particle size fractions = 100%.
•
Log-ratio transformation is key to improve calibrations for soil texture.

Abstract

Soil texture is beside soil organic matter the most fundamental soil parameter affecting most biological, chemical, physical soil properties. However, conventional laboratory analyses of soil texture are time-consuming and expensive, calling for alternative analytical methods. In this study, we tested the performance of near infrared spectroscopy (NIRS) derived texture estimates for a national-scale soil dataset of the German Agricultural Soil Inventory, using memory-based learning and log-ratio transformation. The developed NIRS models had a root mean square error of prediction of 66.1, 55.5, and 18.5 g kg⁻¹ for sand, silt, and clay content, respectively. The lowest relative error (7.5%) was found for clay content, while the relative error for silt content was 11.2% and for sand content 11.1%. Ratio of performance to deviation varied between 4.8 and 6.2 in all cases, indicating excellent model fit. The key to excellent model performance was log-ratio transformation, which allowed all three particle size fractions to be modeled simultaneously while meeting the constraint that all size fractions should add up to 100%. Our NIRS based soil texture estimates outperformed the texture-by-feel method applied to the same set of soils (Vos et al. 2016). NIRS is thus a suitable low-cost analytical method for texture analysis that can be used for large-scale datasets.

Introduction

Soil texture is among the most important and fundamental soil parameters, as it determines soil physical, chemical, and biological properties. Structure, aggregation, water retention, infiltration capacity, nutrient sorption, root penetration resistance, microbial activity, soil carbon turnover, susceptibility to erosion and compaction and, ultimately, soil suitability for agricultural and forest production are all closely linked to soil texture (Phogat et al., 2015). Therefore, soil texture is a parameter on which accurate information is required in soil science and in practice. However, such information is difficult and time-consuming to obtain, e.g., standard texture analysis involves sieving and sedimentation of suspended soil in solution. For the purposes of regional- to continental-scale soil monitoring frameworks and detailed field-scale agricultural maps, texture analysis requires disproportionately large laboratory resources. An alternative method to standard texture analytics, texture-by-feel, that makes it easy and fast to estimate soil texture has been proposed (Vos et al., 2016). Texture-by-feel is widely performed by soil scientists to characterize the texture of soil in the field. Vos et al. (2016) showed that this simple method of testing texture with the fingers is unexpectedly precise and can be used to replace time-consuming laboratory texture analysis for a wide range of purposes and studies. However, the method is associated with a relative error of 14, 30, and 36% in estimation of clay, silt, and sand content, respectively.

Near infrared spectroscopy (NIRS) has gained increasing attention in soil analysis during the past 15 years and has been applied to estimate soil physical and chemical properties (Jaconi et al., 2017; Shepherd and Walsh, 2007; Viscarra-Rossel and Webster, 2012). It is simple to apply and can also be used in situ in the field to estimate soil properties from absorption spectra within the range 780–3000 nm. Thus, NIRS appears to be an appealing technique in balancing low cost and reasonable accuracy, and in extrapolating datasets to regional scale and beyond. However, NIRS is often only calibrated for field to local scale, while application on regional to national scale is challenging due to the high spatial variability in the soil solid phase.

Compositional data are relative information, as they are usually parts of a whole and should always add up to 100% (Pawlowsky-Glahn and Egozcue, 2006). By definition, particle size fractions can be classified as compositional data. However, NIRS-based estimates of the sand, silt, and clay fractions in soil often do not add up to 100%, due to calibration uncertainty and because they are estimated independently.

Aitchison (1982) found a suitable sample space to develop calibrations for compositional data. By applying log-ratio transformation, he was able to calibrate fractions simultaneously and also meet the constraint that all fractions should add up to 100%. Log-ratios of components provide a natural mean of studying compositional data by using information about relative fractions, not absolute fractions. This method has been widely used in mapping soil texture (Lark and Bishop, 2006; Odeh et al., 2003), but has not been applied previously to build NIRS calibrations for soil texture.

The objectives of this study were: (i) to assess the performance of NIRS-derived texture estimates subjected to log-ratio transformation for a national-scale soil dataset managed by the German Agricultural Soil Inventory and (ii) to compare NIRS-derived texture estimates with texture-by-feel estimates reported by Vos et al. (2016) for the same set of soils.

Section snippets

Study area and soil samples

The study was carried out using 10,802 soil samples obtained in the German Agricultural Soil Inventory, which represents 2236 sites (soil profiles). The spatial distribution of sites (grid 8 km × 8 km) covered all parts of Germany, but only soils under agriculture. Soil samples were taken from five to seven different depth increments between 0 cm and 200 cm depth (for further details, see Jaconi et al., 2017). If there were different soil horizons within a depth increment, the sample was split

Accuracy and precision of NIRS-derived soil texture estimates

We were able to estimate soil particle size distribution of the regional-scale dataset from the German Agricultural Soil Inventory with high precision using the MBL algorithm with log-ratio transformation. The smallest relative error was obtained for the clay fraction (RMSEP = 18.5 g kg⁻¹, RE = 7.5%), while the silt and sand content had a similar error (RE = 11%) (Table 2). The RPD and RPIQ values also indicated that the NIRS models were excellent for all three soil particle size classes, with

Conclusions

The NIRS models we developed for estimating soil particle size classes can be classified as excellent in all cases. Estimation of clay content outperformed estimation of other particle size classes, which reflects the direct influence of clay on absorption properties of soil in the NIR range. Moreover, the NIRS-derived texture estimates were much better than those obtained previously with texture-by-feel, an alternative rapid method for estimating soil texture in the field (Vos et al., 2016).

Acknowledgments

The authors gratefully acknowledge Silke Weis, Arne Heidkamp, Roland Prietz, Anna Jacobs, and the team at the German Agricultural Soil Inventory for their efforts in collecting and analyzing the soil samples. The authors also thank the project “CNPq – Science without borders” (process no.: 245735/2012-7), for providing a scholarship to A.J., and the German Federal Ministry of Food and Agriculture, for funding the German Agricultural Soil Inventory at the Thünen Institute of Climate-Smart

References (33)

V. Bellon-Maurel et al.
Critical review of chemometric indicators commonly used for assessing the quality of the prediction of soil attributes by NIR spectroscopy
TrAC Trends Anal. Chem.
(2010)
E. Ben-Dor et al.
Imaging Spectrometry for Soil Applications
(2008)
D.J. Brown et al.
Global soil characterization with VNIR diffuse reflectance spectroscopy
Geoderma
(2006)
L. Ramirez-Lopez et al.
The spectrum-based learner: a new local approach for modeling soil vis–NIR spectra of complex datasets
Geoderma
(2013)
J.B. Sankey et al.
Comparing local vs. global visible and near-infrared (VisNIR) diffuse reflectance spectroscopy (DRS) calibrations for the prediction of soil clay, organic C and inorganic C
Geoderma
(2008)
B. Stenberg et al.
Visible and near infrared spectroscopy in soil science
Adv. Agron.
(2010)
R.A. Viscarra Rossel et al.
In situ measurements of soil colour, mineral composition and clay content by vis–NIR spectroscopy
Geoderma
(2009)
C. Vos et al.
Field-based soil-texture estimates could replace laboratory analysis
Geoderma
(2016)
J. Aitchison
The statistical analysis of compositional data
J. R. Stat. Soc.
(1982)
B.G. Barthès et al.
Determination of total carbon and nitrogen content in a range of tropical soils using near infrared spectroscopy: influence of replication and sample grinding and drying
J. Near Infrared Spectrosc.
(2006)

C.-W. Chang et al.

Near-infrared reflectance spectroscopy–principal component regression analyses of soil properties

Soil Sci. Soc. Am. J.

(2001)

D. Cozzolino et al.

The potential of near-infrared reflectance spectroscopy to analyse soil chemical and physical characteristics

J. Agric. Sci.

(2003)

L. Deiss et al.

Soil texture and organic carbon fractions predicted from near-infrared spectroscopy and geostatistics

Soil Sci. Soc. Am. J.

(2017)

C. Du et al.

Evaluation of soil fertility using infrared spectroscopy: a review

Environ. Chem. Lett.

(2009)

V. Genot et al.

Near infrared reflectance spectroscopy for estimating soil characteristics valuable in the diagnosis of soil fertility

J. Near Infrared Spectrosc.

(2011)

A. Gholizadeh et al.

A memory-based learning approach as compared to other data mining algorithms for the prediction of soil texture using diffuse reflectance spectra

Remote Sens.

(2016)

Cited by (52)

Removal of environmental influences for estimating soil texture fractions based on ZY1 satellite hyperspectral images
2024, Catena
Soil texture is one of the important factors affecting the physical and chemical properties of soil, and is generally divided into sand, silt, and clay. Understanding its spatial distribution is crucial for agricultural and environmental decision-making. Estimation of soil texture fractions (STFs) can be achieved through visible, near-infrared, and shortwave infrared (VNIR-SWIR) spectra, making hyperspectral imaging technology particularly promising for low-cost and large-scale monitoring of STFs. However, the process of image acquisition is inevitably affected by soil environmental factors. In this study, a soil texture fractions estimation method for hyperspectral images was proposed based on Orthogonal Signal Correction (OSC). Firstly, the OSC algorithm was applied to eliminate the impact of soil environmental factors on the image spectra. Then, the spectral feature was enhanced by Fractional Order Derivative (FOD). Subsequently, a Partial Least Squares Regression (PLSR) model was established based on the optimal bands selected by Competitive Adaptive Reweighted Sampling (CARS). A total of 107 soil samples were obtained from the agricultural area in Lishu County, Jilin Province, China. The hyperspectral images of the ZY1-02D and ZY1-02E satellites were acquired simultaneously in this area. The results showed that the OSC algorithm effectively reduced the influence of soil environmental factors on image spectra, and significantly improved the accuracy of STFs estimation. After applying OSC to the original image spectra, the estimation accuracies (R²) for sand, silt, and clay content were improved from 0.52, 0.51, 0.50 to 0.81, 0.65, and 0.70 respectively. Compared to External Parameter Orthogonalization (EPO) correction, the model accuracy and stability were higher with OSC correction. The selected results from CARS indicated that the characteristic spectral bands of soil texture are concentrated around 500 nm, 750 nm, 900 nm, and 2200 nm. Further analysis indicates that the distribution trends of the soil texture fractions in this study were consistent with previous studies, demonstrating the good applicability of the estimation method.
Enhancing soil texture classification with multivariate scattering correction and residual neural networks using visible near-infrared spectra
2024, Journal of Environmental Management
Soil texture is one of the most important indicators of soil physical properties, which has traditionally been measured through laborious procedures. Approaches utilizing visible near-infrared spectroscopy, with their advantages in efficiency, eco-friendliness and non-destruction, are emerging as potent alternatives. Nevertheless, these approaches often suffer from limitations in classification accuracy, and the substantial impact of spectral preprocessing, model integration, and sample matrix effect is commonly disregarded. Here a novel 11-class soil texture classification strategy that address this challenge by combining Multiplicative Scatter Correction (MSC) with Residual Network (ResNet) models was presented, resulting in exceptional classification accuracy. Utilizing the LUCAS dataset, collected by the Land Use and Cover Area frame Statistical Survey project, we thoroughly evaluated eight spectral preprocessing methods. Our findings underscored the superior performance of MSC in reducing spatial complexity within spectral data, showcasing its crucial role in enhancing model precision. Through comparisons of three 1D CNN models and two ResNet models integrated with MSC, we established the superior performance of the MSC-incorporated ResNet model, achieving an overall accuracy of 98.97 % and five soil textures even reached 100.00 %. The ResNet model demonstrated a marked superiority in classifying datasets with similar features, as observed by the confusion matrix analysis. Moreover, we investigated the potential benefit of pre-categorization based on land cover type of the soil samples in enhancing the accuracy of soil texture classification models, achieving overall classification accuracies exceeding 99.39 % for woodland, grassland, and farmland with the 2-layer ResNet model. The proposed work provides a pioneering and efficient strategy for rapid and precise soil texture identification via visible near-infrared spectroscopy, demonstrating unparalleled accuracy compared to existing methods, thus significantly enhancing the practical application prospects in soil, agricultural and environmental science.
Water content estimation of recycled building materials based on near-infrared spectroscopy
2024, Construction and Building Materials
Recycling of construction and demolition waste (CDW) is one key component of the circular economy. Reusing construction materials opens the opportunity to preserve finite natural resources and their corresponding landscapes, and also reduces the emission of greenhouse gases.
Unfortunately, recycling of CDW is challenging when the new product ought to be used on the same quality level as the demolished one and down-cycling should be avoided. Therefore, the CDW needs to be classified and impurities must be removed. The water content quantification of the recycled aggregates is important because the amount of existing water needs to be known, e.g., when calculating a mixture for recycling concrete. New developments in the classification and the sorting of CDW make use of the information from the spectral range in the visual and infrared region. Therefore, the approach of this study is to examine whether this information can be used to estimate the water content of various CDW materials. Based on reflection measurements in the near-infrared range between 1000 nm and 2300 nm a mathematical relationship between the spectral measurements and the water content is established. Three estimates for the water content are set up. The mean absorption, the water index, and the absorption difference are later tested on CDW material samples covering concrete, autoclaved aerated concrete, lightweight concrete, brick, roof tile, calcium silicate brick and gypsum.
Spectral prediction of soil salinity and alkalinity indicators using visible, near-, and mid-infrared spectroscopy
2023, Journal of Environmental Management
Drought and the impacts of climate change have led to an escalation in soil salinity and alkalinity across various regions worldwide, including Iran. The Chahardowli Plain in western Iran, in particular, has witnessed a significant intensification of this phenomenon over the past decade. Consequently, modeling of soil attributes that serve as indicators of soil salinity and alkalinity became a priority in this region. To date, only a limited number of studies have been conducted to assess indicators of salinity and alkalinity through spectrometry across diverse spectral ranges. The spectral ranges encompassing mid-infrared (mid-IR), visible, and near-infrared (vis-NIR) spectroscopy were employed to estimate soil properties including sodium adsorption ratio (SAR), exchangeable sodium ratio (ESR), exchangeable sodium percentage (ESP), pH, and electrical conductivity (EC). Five distinct models were employed: Partial Least Squares Regression (PLSR), bootstrapping aggregation PLSR (BgPLSR), Memory-Based Learning (MBL), Random Forest (RF), and Cubist. The calibration and assessment of model performance were carried out using several key metrics including Ratio of Performance to Deviation (RPD) and the coefficient of determination (R²). Analysis of the outcomes indicates that the accuracy and precision of the mid-IR spectra surpassed that of vis-NIR spectra, except for pH, which exhibited a superior RPD compared to other properties. Notably, in the prediction of pH utilizing vis-NIR reflectance spectra, the BgPLSR model exhibited the highest accuracy and precision, boasting an RPD value of 2.56. In the domain of EC prediction, the PLSR model yielded an RPD of 2.64. For SAR, the MBL model achieved an RPD of 2.70, while ESR prediction benefited from the MBL model with an impressive RPD of 4.36. Likewise, the MBL model demonstrated remarkable precision and accuracy in ESP prediction, garnering an RPD of 4.41. The MBL model's efficacy in forecasting with limited datasets was notably pronounced among the models considered. This study underscores the valuable role of spectral predictions in facilitating the work of soil surveyors in gauging salinity and alkalinity indicators. It is recommended that the integration of spectrometry-based salinity and alkalinity predictions be incorporated into forthcoming soil mapping endeavors within semi-arid and arid regions.
Quantifying soil properties relevant to soil organic carbon biogeochemical cycles by infrared spectroscopy: The importance of compositional data analysis
2023, Soil and Tillage Research
Oxyhydroxides, soil texture and soil organic carbon (SOC) fractions are key parameters determining organic carbon cycling in soils. Standard laboratory methods to determine these soil properties are, however, time–consuming and expensive. Visible near infrared (Vis–NIR) and mid infrared (MIR) spectroscopy have been recognized as a promising alternative, but previous studies have not explicitly considered the above–mentioned soil properties as compositional data. The fractional components in compositional data are interrelated but their sum should be unity – these features should be represented in the spectral modeling process to minimize the prediction bias. In this study, two unique datasets were used to test these premises. The first one consisted of 655 samples collected from agricultural terraces and lynchets across Europe, which were scanned to acquire MIR spectra, while in the second one 4516 samples from private gardens across Flanders, Belgium were used to acquire Vis–NIR spectra. Memory–based learning models were optimized using both raw data (conventional method) and transformed data of soil properties by additive log–ratio (alr), centered log–ratio (clr), and isometric log–ratio (ilr) transformation methods. Results showed that the log–ratio transformation methods produced predictions as accurate as the conventional method, whilst also added two significant benefits: (1) they ensured the predicted fractions added up to 100% and (2) they reduced the number of samples with extreme prediction errors. We found that for 11 out of 18 investigated soil properties, the three log–ratio transformation methods provided similar model performance, whilst ilr outperformed clr for the prediction of silt and sand content of garden soils, for coarse particulate SOC (>250 µm) and microaggregate–associated SOC (250–53 µm) of terrace soils. For the remaining three properties (Al oxyhydroxides) alr outperformed ilr. Fair to excellent predictive models (RPD from 1.4 to 4.3; R² from 0.50 to 0.95) were achieved for soil oxyhydroxides (Fe, Al, Mn) and soil texture from MIR spectra. Our approach also enabled accurate predictions of silt and sand content of garden soils using Vis–NIR spectra (RPD = 1.9; R² = 0.72), although accuracy for clay was lower (RPD = 1.3; R² = 0.49). This study demonstrates that combining soil infrared spectroscopy with a compositional data analysis is a promising technique that enables cost-effective and reliable quantification of soil properties relevant to SOC stability, thus offering a practical opportunity to assess the role of SOC in global C cycling.
Prediction of soil properties based on characteristic wavelengths with optimal spectral resolution by using Vis-NIR spectroscopy
2023, Spectrochimica Acta - Part A: Molecular and Biomolecular Spectroscopy
Visible and near-infrared (Vis-NIR) spectroscopy technique has been recognized as a cost-effective, rapid, non-destructive alternative to traditional soil physicochemical analysis to estimate soil properties over the past few decades. Most efforts are devoted to the selection of characteristic wavelengths to eliminate the uninformative variables while ignoring the impact of the spectral resolution of these wavelengths on the prediction accuracy of soil properties. Therefore, the originality of this study is to identify the characteristic wavelengths with the optimal spectral resolution to achieve a better prediction performance. A 'two-step' wavelength selection method was proposed to select the characteristic wavelengths. Then, we simulated 1 nm–100 nm spectral resolution based on the spectral database measured by a portable ASD spectroradiometer and adopted the artificial bee colony (ABC) algorithm to further improve the prediction ability by configuring the most appropriate spectral resolution for each characteristic wavelength. The soil databases for this study consisted of 112 soil samples collected from Songnen Plain area in northeast China, and partial least squares regression (PLSR) was used to establish relations between pretreatment spectra and soil properties, including soil organic matter (SOM), available phosphorus (AP), and available potassium (AK). The independent validation results of this strategy effectively favored the prediction accuracy of SOM ( ${RMSE}_{p} = 2.758, {R^{2}}_{p} = 0.857, {RPIQ}_{p} = 4.579$ ), AP ( ${RMSE}_{p} = 0.062, {R^{2}}_{p} = 0.790, {RPIQ}_{p} = 3.552$ ), and AK ( ${RMSE}_{p} = 0.366, {R^{2}}_{p} = 0.758, {RPIQ}_{p} = 2.525$ ) compared with the PLSR models developed with full-spectra. In general, the method presented in this study suggested a framework for selecting characteristic wavelengths with optimal spectral solutions to predict SOM, AP, AK, and perhaps some other soil properties. The results of this paper also will provide guidance for the development of the low-cost specialized spectroscopic instruments for soil properties measurement.

View all citing articles on Scopus

View full text

Near infrared spectroscopy as an easy and precise method to estimate soil texture

Highlights

Abstract

Introduction

Section snippets

Study area and soil samples

Accuracy and precision of NIRS-derived soil texture estimates

Conclusions

Acknowledgments

TrAC Trends Anal. Chem.

Geoderma

Geoderma

Geoderma

Adv. Agron.

Geoderma

Geoderma

The statistical analysis of compositional data

J. R. Stat. Soc.

Determination of total carbon and nitrogen content in a range of tropical soils using near infrared spectroscopy: influence of replication and sample grinding and drying

J. Near Infrared Spectrosc.

Near-infrared reflectance spectroscopy–principal component regression analyses of soil properties

Soil Sci. Soc. Am. J.

The potential of near-infrared reflectance spectroscopy to analyse soil chemical and physical characteristics

J. Agric. Sci.

Soil texture and organic carbon fractions predicted from near-infrared spectroscopy and geostatistics

Soil Sci. Soc. Am. J.

Evaluation of soil fertility using infrared spectroscopy: a review

Environ. Chem. Lett.

Near infrared reflectance spectroscopy for estimating soil characteristics valuable in the diagnosis of soil fertility

J. Near Infrared Spectrosc.

A memory-based learning approach as compared to other data mining algorithms for the prediction of soil texture using diffuse reflectance spectra

Remote Sens.