Proximal Soil Sensors Research Articles

SENSOR SYSTEMS FOR MAPPING SOIL FERTILITY ATTRIBUTES: CHALLENGES, ADVANCES, AND PERSPECTIVES IN BRAZILIAN TROPICAL SOILS

Soil fertility attributes have different scales and forms of spatial and temporal variations in agricultural fields. Adequate spatiotemporal characterization of these attributes is fundamental to the successful development of strategies for variable rate application of fertilizers, enabling the classic benefits of precision agriculture (PA). Studies on Brazilian soil have shown that at least 1 sample ha-1 is required for the reliable mapping of key fertility attributes. However, this sampling density is difficult owing to the operational challenges of sample collection and the cost of laboratory analyses. Given this limitation, soil sensors have emerged as a practical and complementary technique for obtaining information on soil attributes, at high spatial density, without the production of chemical residues and at a reduced cost. Scientists worldwide have devoted their attention to the development and application of sensor systems for this purpose. The concept of proximal soil sensing (PSS) was established in 2011 and involves the application of soil sensors directly on the field. PSS techniques involve different disciplines, such as instrumentation, data science, geostatistics, and predictive modeling. The integration of these different disciplines has allowed successful sensor application for the spatial diagnosis of soil fertility attributes. The present work aimed to present a bibliographic review of the concepts involved and main techniques used in soil sensing to predict fertility attributes. We sought to present a broad view of the challenges, advances, and perspectives of sensor application in Brazilian tropical soils in the context of PA.

Proximal sensing of soil particle sizes using a microscope-based sensor and bag of visual words model

While soil textural information is critical for agronomic decision making and environmental applications, its characterization in the laboratory is laborious and highly time-consuming, often costly and thus, not affordable for a considerable number of samples. This study proposed the use of a microscope-based image acquisition system and Bag of Visual Words (BoVW) model to quantify soil texture ex situ and in situ. A low-cost and portable microscope was used to design and develop a prototype soil surface image acquisition system. A BoVW computer-vision algorithm was used to extract the surficial characteristics pertaining to color and roughness. Partial least squares regression (PLSR) models were applied to correlate BoVW extracted features with the sand, silt, and clay content measured by the hydrometer method. The leave-one-sample-out cross-validation (LOSOCV) showed good agreement between measured and predicted soil separates. Coefficient of determination (R2) between the laboratory measured and image (air-dried soil samples) predicted sand, silt, and clay were 0.77, 0.68, and 0.71 and root mean squared error (RMSE) were 5.92%, 6.01%, and 2.98%, respectively. Similarly, R2 were 0.78, 0.67, and 0.52 and RMSE were 5.81%, 6.06%, and 3.85%, respectively for laboratory measured and in situ moist image predicted sand, silt and clay. These results demonstrated the potential of developing a microscope-based image acquisition system as a proximal soil sensor in conjunction with a BoVW computer-vision algorithm to efficiently characterize soil texture in greater detail, in a more cost effective and time-saving way than the conventional analysis methods.

Clustering Tools for Integration of Satellite Remote Sensing Imagery and Proximal Soil Sensing Data

Remote sensing (RS) and proximal soil sensing (PSS) technologies offer an advanced array of methods for obtaining soil property information and determining soil variability for precision agriculture. A large amount of data collected by these sensors may provide essential information for precision or site-specific management in a production field. Data clustering techniques are crucial for data mining, and high-density data analysis is important for field management. A new clustering technique was introduced and compared with existing clustering tools to determine the relatively homogeneous parts of agricultural fields. A DUALEM-21S sensor, along with high-accuracy topography data, was used to characterize soil variability in three agricultural fields situated in Ontario, Canada. Sentinel-2 data assisted in quantifying bare soil and vegetation indices (VIs). The custom Neighborhood Search Analyst (NSA) data clustering tool was implemented using Python scripts. In this algorithm, part of the variance of each data layer is accounted for by subdividing the field into smaller, relatively homogeneous, areas. The algorithm’s attributes were illustrated using field elevation, shallow and deep apparent electrical conductivity (ECa), and several VIs. The unique feature of this proposed protocol was the successful development of user-friendly and open source options for defining the spatial continuity of each group and for use in the zone delineation process.

Mapping soil organic matter contents at field level with Cubist, Random Forest and kriging

The soil organic matter (SOM) content is strongly related to soil fertility and greenhouse gas emissions. Knowledge of the soil SOM content is therefore necessary for efficient and sustainable management practices. In this study, we compare the performance of five machine learning techniques for the prediction of SOM contents using remote sensing, proximal soil sensors and topographic data as environmental predictor. The methods used were kriging, Cubist, Random Forest and regression-kriging (Cubist- kriging and Random Forest kriging). 285 soil samples were collected from a 10 ha field in Denmark. 75% of the data were used for training and 25% for validation. NDVI (Normalized Difference Vegetation Index), RVI (Ratio Vegetation Index), DVI (Difference Vegetation Index) and SAVI (Soil-Adjusted Vegetation Index) were derived from Sentinel 2 images with 10 m resolution as auxiliary variables. Topographic variables were derived from a digital elevation model (DEM) with 1.6 m resolution. The performance of the methods was compared based on the coefficient of determination (R2) and root mean square error (RMSE). The results showed that kriging achieved the best performance, followed by regression-kriging. The models using only Cubist and Random forest had the poorest performance. The results, therefore, demonstrate that kriging can predict SOM contents without the need of auxiliary variables for fields with high sampling densities.

Simultaneous measurement of multiple soil properties through proximal sensor data fusion: A case study

In this research, proximal soil sensor data fusion was defined as a multifaceted process which integrates geospatially correlated data, or information, from multiple proximal soil sensors to accurately characterize the spatial complexity of soils. This has capability of providing improved understanding of soil heterogeneity for potential applications associated with crop production and natural resource management. To assess the potential of data fusion for the purpose of improving thematic soil mapping over the single sensor approach, data from multiple proximal soil sensors were combined to develop and validate predictive relationships with laboratory-measured soil physical and chemical properties. The work was conducted in an agricultural field with both mineral and organic soils. The integrated data included: topography records obtained using a real-time kinetic (RTK) global navigation satellite system (GNSS) receiver, apparent soil electrical conductivity (ECa) obtained using an electromagnetic induction sensor, and content of several naturally occurring radioisotopes detected using a mobile gamma-ray spectrometer. In addition, the soil profile data were collected using a commercial ruggedized multi-sensor platform carrying a visible and near-infrared (vis-NIR) optical sensor and a galvanic contact soil ECa sensor. The measurements were carried out at predefined field locations covering the entire study area identified from sensor measured a priori information on field elevation, ECa and gamma-ray count. The information was used to predict: soil organic matter (SOM), pH, lime buffer capacity (LBC), as well as concentration of phosphorus (P), potassium (K), calcium (Ca), magnesium (Mg), and aluminum (Al). Partial least squares regressions (PLSRs) were used to predict soil properties from individual sensors and different sensor combinations (sensor data fusion). By integrating the data from all of the proximal soil sensors, SOM, pH, LBC, Ca, Mg, and Al were predicted simultaneously with R2 > 0.5 (RPD > 1.50). Improved predictions were observed for most soil properties based on sensor data fusion than those based on individual sensors. After choosing the optimal sensor combination for each soil property, the predictive capability was compared using different data mining algorithms, including support vector machines (SVM), random forest (RF), multivariate adaptive regression splines (MARS), and regression trees (CART). Improved predictions for SOM, Ca, Mg, and Al were observed using SVM over PLSR. The predictive capability was followed by RF and MARS, with CART. Predictions of pH and LBC were only feasible using MARS and PLSR, respectively. In this field, it was not possible to predict extractable P and K using all tested sensor combinations or algorithms. With large variability in SOM, the field presents a special situation and thus, the result could be specific to the study site. Further research includes an extended number of experimental sites covering different geographic areas around Eastern Canada.

Detailed mapping of soil texture of a paddy growing soil using multivariate geostatistical approaches

This study was conducted to explore performances of multivariate geostatistical techniques, co-kriging and regression kriging in contrast to univariate ordinary kriging, to generate detailed maps of soil texture by using proximally sensed apparent electrical conductivity (ECa) as secondary information. The survey of ECa (n = 21110) was conducted in a paddy tract (2.5 ha) located in Kurunegala, Sri Lanka using DUALEM-1S proximal soil sensor. Twenty-five soil samples were collected on the basis of conditioned Latin hypercube sampling approach. Soil texture was determined using pipette method. Additive log transformed sand and clay values were used to produce soil texture maps using multivariate co-kriging, regression kriging and univariate ordinary kriging. Data ranges of clay (3.3 – 19.5%) and sand (62.5 – 90.1%) showed a considerable variability within the study area. Correlation analysis revealed a strong relationship of clay% with horizontal (r = 0.86) and perpendicular (r = 0.89) coplanar ECa and their geometric mean (r = 0.89). Sand % showed strong negative relationships with horizontal (r = -0.89) and perpendicular (r = -0.89) coplanar ECa and their geometric mean (r = 0.90). Second order polynomial regression models were best fitted for the prediction of clay (R2 = 0.83), and sand (R2 = 0.83) and these relationships were used for regression kriging. Prediction accuracies of geostatistical approaches were investigated by leave-one-out cross validation procedure and estimation of mean error, mean absolute error and root mean square error. Detailed maps resolution = 1 x 1 m) of clay and sand generated using co-kriging, regression kriging and ordinary kriging showed similar spatial patterns. However, multivariate geostatistical techniques produced more accurate detailed soil texture maps. Further, the comparison of mean error, mean absolute error and root mean square error values of three different interpolation techniques indicated that regression kriging produced the most accurate soil texture maps. This study emphasizes the high potential and robustness of regression kriging combined with proximally sensed apparent electrical conductivity to create high accurate detailed texture maps in efficient and cost effective manners.

Assessment of reclaimed soils by unsupervised clustering of proximal sensor data

The application of soil proximal sensors on reclaimed sites presents a novel method for assessing the quality of reclaimed landscapes. This method improves assessment reliability, information management, and environmental assurance. One proximal sensing system that could be used to provide high spatial resolution measurements of soil parameters is an on-the-go optical sensor that collects data at two wavelengths: 660 and 940 nm. Proximal soil sensing data were collected at 27 sites, where organic matter, cation exchange capacity (CEC), and soil water content were collected from 221 soil samples from 0 to 15 cm. The proximal soil sensor data were then automatically clustered using a combination of self-organizing maps and random uniform forests. Overall, the proximal sensor data combined with this data analysis approach created maps with either three or four soil zones. On average, soil zones had statistically significant differences in organic matter, CEC, and water content. This system could be used to map out zones with significant soil variation as part of reclamation monitoring and then used to guide laboratory analytical sampling. Future work should focus on development of on-the-go reflectance spectroscopy systems to provide quantitative soil data with high spatial resolution.

Multi‐sensor fusion for the determination of several soil properties in the Yangtze River Delta, China

SummarySoil organic matter (SOM), total nitrogen (TN), available nitrogen (AN), available phosphorus (AP), available potassium (AK) and pH are key chemical properties for evaluating soil fertility and quality. This study involved the integration of four soil sensors, visible near‐infrared (vis–NIR) spectrometer, mid‐infrared (mid‐IR) spectrometer, portable X‐ray fluorescence (PXRF) analyser and laser‐induced breakdown spectroscopy (LIBS), to achieve rapid measurement of these soil properties. A genetic algorithm and partial least‐squares regression (GA–PLSR) were used to select characteristic bands to reduce data redundancy. We then calibrated models from three aspects: models using partial least‐squares regression (PLSR) based on single sensor data, models using PLSR based on fused sensor data, involving data combined from the four sensors into a new dataset to create a data fusion (DF) model, and models with Bayesian model averaging (BMA) based on prediction results of fused sensor data, involving prediction results combined from the four sensors into a new dataset to form the BMA model. The results showed the following. (i) For the single sensor, the predictive performance decreased as follows: mid‐IR > vis–NIR > LIBS > PXRF. (ii) Compared with the single sensor approach, the DF approach slightly improved or even reduced prediction accuracy and caused a large amount of redundancy. We suggest that this approach is not able to improve predictive ability. (iii) The BMA approach achieved the best prediction for the six soil properties. Our findings suggest that model averaging of vis–NIR, mid‐IR and LIBS could be a reliable and stable approach for the fast measurement of soil properties.Highlights We used four proximal soil sensors to evaluate six key properties for evaluating soil fertility and quality. GA–PLSR was used to select characteristic bands. We compared predictions of six soil properties from single sensor, DF and BMA approaches. BMA predictions were more accurate than predictions from single and fused sensor data.

Mapping The Temporal and Spatial Variability of Soil Moisture Content Using Proximal Soil Sensing

In studies related to soil optical properties, it has been proven that visual and NIR soil spectral response can predict soil moisture content (SMC) using proper data analysis techniques. SMC is one of the most important soil properties influencing most physical, chemical, and biological soil processes. The problem is how to provide reliable, fast and inexpensive information of SMC in the subsurface from numerous soil samples and repeated measurement. The use of spectroscopy technology has emerged as a rapid and low-cost tool for extensive investigation of soil properties. The objective of this research was to develop calibration models based on laboratory Vis-NIR spectroscopy to estimate the SMC at four different growth stages of the soybean crop in Yogyakarta Province. An ASD Field-spectrophotoradiometer was used to measure the reflectance of soil samples. The partial least square regression (PLSR) was performed to establish the relationship between the SMC with Vis-NIR soil reflectance spectra. The selected calibration model was used to predict the new samples of SMC. The temporal and spatial variability of SMC was performed in digital maps. The results revealed that the calibration model was excellent for SMC prediction. Vis-NIR spectroscopy was a reliable tool for the prediction of SMC.

Modelling Soil Water Content in a Tomato Field: Proximal Gamma Ray Spectroscopy and Soil–Crop System Models

Proximal soil sensors are taking hold in the understanding of soil hydrogeological processes involved in precision agriculture. In this context, permanently installed gamma ray spectroscopy stations represent one of the best space–time trade off methods at field scale. This study proved the feasibility and reliability of soil water content monitoring through a seven-month continuous acquisition of terrestrial gamma radiation in a tomato test field. By employing a 1 L sodium iodide detector placed at a height of 2.25 m, we investigated the gamma signal coming from an area having a ~25 m radius and from a depth of approximately 30 cm. Experimental values, inferred after a calibration measurement and corrected for the presence of biomass, were corroborated with gravimetric data acquired under different soil moisture conditions, giving an average absolute discrepancy of about 2%. A quantitative comparison was carried out with data simulated by AquaCrop, CRITeRIA, and IRRINET soil–crop system models. The different goodness of fit obtained in bare soil condition and during the vegetated period highlighted that CRITeRIA showed the best agreement with the experimental data over the entire data-taking period while, in presence of the tomato crop, IRRINET provided the best results.

Characterizing soil particle sizes using wavelet analysis of microscope images

Soil texture (relative proportions of soil particles of varied sizes) is a fundamental soil physical property affecting almost all other soil physical properties and processes of agricultural, environmental and engineering importance. However, characterization of particle sizes in the laboratory presents a range of challenges in terms of the time, labor, difficulty and/or cost involved with the analysis. Continuous wavelet transform (CWT) has been used in characterizing scale-specific variations in spatial or temporal domain as well as in image analysis. The objective of this study was to develop a CWT-based computer vision algorithm to characterize soil particle sizes from digital images captured with a microscope. A cheap portable microscope with 5 MP camera and maximum magnification of 200× was used to develop an image acquisition system. Three images of air-dried, ground (2 mm) soil samples were captured in laboratory conditions for each soil sample (56 + 67 = 123) collected from two agricultural fields (Field26 and Field86) with highly variable soils. Triplicate in-situ images were also collected from 67 locations from Field86 after scrapping off surface residues. The color images were transferred to grey-scale images and the CWT was performed along the 20 equally-spaced rows and columns. The total area under the average global wavelet spectrum represented the total variation in any image. Two fractions of particle sizes; ‘coarse’ (diameter between 2.0 mm and 0.05 mm) and ‘fine’ (diameter < 0.05 mm), respectively, representing sand and sum of silt and clay were calculated based on the area under the curve and compared with lab-measured particle sizes using the hydrometer method. The lab-measured coarse- and fine-fractions showed strong agreement with the predicted (from image) fractions. The regression relationship showed the prediction capability of 87% and 88% for coarse (RMSE 44.7 g kg−1) and fine (RMSE 44.7 g kg−1) fractions, respectively for Field26 samples. A similar prediction was obtained (88% with RMSE 40.2 g kg−1 for coarse and 88% with RMSE 40.3 g kg−1 for fine) for Field86 samples. The efficiency of the wavelet algorithm shows promise in determining the particle sizes from an image and the portable nature of the image acquisition system results in a good proximal soil sensor. In contrast to the laboratory images, a weak prediction (48% for coarse and 56% for fine) was observed for the images taken in-situ mainly due to the quality of the images as they were affected by various field conditions; this requires further research.

Quantifying and predicting spatio-temporal variability of soil CH4 and N2O fluxes from a seemingly homogeneous Australian agricultural field

Soil methane (CH4) and nitrous oxide (N2O) fluxes are difficult to predict from soil temperature and moisture alone, especially compared to carbon dioxide (CO2) fluxes. That difficulty is reflected in high spatial and temporal (spatiotemporal) variability of these two greenhouse gases (GHGs). We used a 16ha field, under homogeneous soils and vegetation, to simultaneously explore spatial and temporal variability of soil CH4 and N2O fluxes. We also measured soil physical and chemical properties in order to explain, and predict, spatial variability of these two gases. Gas fluxes were measured using either a dynamic chamber (spatial variability study) or automated chambers using FTIR (temporal variability study). Soil samples were analysed for 30 chemical parameters (including at least two forms of soil carbon and nitrogen), while two proximal soil sensors were used to collect fine-resolution soil electrical conductivity and gamma radiometric concentration across the site. Fluxes of CH4 and N2O showed distinct spatial patterns, and were uniquely related to soil properties. Spatial variability in both CH4 and N2O fluxes was greater than five months of temporal variability (an increase in 112% and 39% in standard deviations for each gas respectively). If we relied solely on the autochambers for mean field fluxes, we would have underestimated fluxes by 59 and 197%, for CH4 and N2O respectively. CH4 fluxes were more spatially-dependent than those of N2O (semivariance analysis), but both showed greater spatial dependence than previously reported. Nearly 40 and 50% of the mean spatial flux of CH4 and N2O were from 1% of the area. Spatial variability in soil CH4 fluxes was predicted best by electrical conductivity measurements at 0–50cm (r=0.74) and soil C. Soil N2O fluxes, on the other hand, were predicted best by soil N and the gamma radiometric data (r=0.48). Overall, our results clearly show that the large spatial variance of both CH4 and N2O fluxes requires great caution when scaling from chamber-based measurements to the field and beyond. Proximal sensors (as used here) can help map “hot spots” of soil CH4 and N2O fluxes at the field scale.

Application of portable XRF and VNIR sensors for rapid assessment of soil heavy metal pollution.

Rapid heavy metal soil surveys at large scale with high sampling density could not be conducted with traditional laboratory physical and chemical analyses because of the high cost, low efficiency and heavy workload involved. This study explored a rapid approach to assess heavy metals contamination in 301 farmland soils from Fuyang in Zhejiang Province, in the southern Yangtze River Delta, China, using portable proximal soil sensors. Portable X-ray fluorescence spectroscopy (PXRF) was used to determine soil heavy metals total concentrations while soil pH was predicted by portable visible-near infrared spectroscopy (PVNIR). Zn, Cu and Pb were successfully predicted by PXRF (R2 >0.90 and RPD >2.50) while As and Ni were predicted with less accuracy (R2 <0.75 and RPD <1.40). The pH values were well predicted by PVNIR. Classification of heavy metals contamination grades in farmland soils was conducted based on previous results; the Kappa coefficient was 0.87, which showed that the combination of PXRF and PVNIR was an effective and rapid method to determine the degree of pollution with soil heavy metals. This study provides a new approach to assess soil heavy metals pollution; this method will facilitate large-scale surveys of soil heavy metal pollution.

Performance Evaluation of Proximal Sensors for Soil Assessment in Smallholder Farms in Embu County, Kenya.

Four proximal soil sensors were tested at four smallholder farms in Embu County, Kenya: a portable X-ray fluorescence sensor (PXRF), a mobile phone application for soil color determination by photography, a dual-depth electromagnetic induction (EMI) sensor, and a LED-based soil optical reflectance sensor. Measurements were made at 32–43 locations at each site. Topsoil samples were analyzed for plant-available nutrients (N, P, K, Mg, Ca, S, B, Mn, Zn, Cu, and Fe), pH, total nitrogen (TN) and total carbon (TC), soil texture, cation exchange capacity (CEC), and exchangeable aluminum (Al). Multivariate prediction models of each of the lab-analyzed soil properties were parameterized for 576 sensor-variable combinations. Prediction models for K, N, Ca and S, B, Zn, Mn, Fe, TC, Al, and CEC met the setup criteria for functional, robust, and accurate models. The PXRF sensor was the sensor most often included in successful models. We concluded that the combination of a PXRF and a portable soil reflectance sensor is a promising combination of handheld soil sensors for the development of in situ soil assessments as a field-based alternative or complement to laboratory measurements.

Proximal Soil Sensing - A Contribution for Species Habitat Distribution Modelling of Earthworms in Agricultural Soils?

BackgroundEarthworms are important for maintaining soil ecosystem functioning and serve as indicators of soil fertility. However, detection of earthworms is time-consuming, which hinders the assessment of earthworm abundances with high sampling density over entire fields. Recent developments of mobile terrestrial sensor platforms for proximal soil sensing (PSS) provided new tools for collecting dense spatial information of soils using various sensing principles. Yet, the potential of PSS for assessing earthworm habitats is largely unexplored. This study investigates whether PSS data contribute to the spatial prediction of earthworm abundances in species distribution models of agricultural soils.Methodology/Principal FindingsProximal soil sensing data, e.g., soil electrical conductivity (EC), pH, and near infrared absorbance (NIR), were collected in real-time in a field with two management strategies (reduced tillage / conventional tillage) and sandy to loam soils. PSS was related to observations from a long-term (11 years) earthworm observation study conducted at 42 plots. Earthworms were sampled from 0.5 x 0.5 x 0.2 m³ soil blocks and identified to species level. Sensor data were highly correlated with earthworm abundances observed in reduced tillage but less correlated with earthworm abundances observed in conventional tillage. This may indicate that management influences the sensor-earthworm relationship. Generalized additive models and state-space models showed that modelling based on data fusion from EC, pH, and NIR sensors produced better results than modelling without sensor data or data from just a single sensor. Regarding the individual earthworm species, particular sensor combinations were more appropriate than others due to the different habitat requirements of the earthworms. Earthworm species with soil-specific habitat preferences were spatially predicted with higher accuracy by PSS than more ubiquitous species.Conclusions/SignificanceOur findings suggest that PSS contributes to the spatial modelling of earthworm abundances at field scale and that it will support species distribution modelling in the attempt to understand the soil-earthworm relationships in agroecosystems.

Microscope-based computer vision to characterize soil texture and soil organic matter

Characterization and quantification of soil properties are important for the optimum use and management our soil. While Soil texture is an important factor for decision making in agriculture, civil engineering and other industries, soil organic matter (SOM) is the backbone of soil health and quality and also affects a range of other soil properties and processes. Traditional methods for estimating these soil properties are time consuming and laborious. This paper discloses the design and development of a new cost effective in situ computer vision-based sensor system to estimate soil texture and SOM. A small and inexpensive hand-held microscope was used to develop an image acquisition system. Images acquired in laboratory with variable texture and SOM were processed using means of geospatial data analysis based computer vision algorithm. Simple linear regression predictive relationships were developed to estimate soil texture and SOM using various computed parameters of acquired imagery. Predicted sand (coefficients of determination, R2 = 0.63) and SOM (R2 = 0.83) were in good agreement with the laboratory measurement. These correspond to root mean square error of 84.7 g sand kg−1 of soil and 0.11 log (SOM %) for soils exhibiting the range of % sand from 3.4 to 59.3 and % SOM from 5.5 to 72.8, respectively. Low cost and the portability of the acquisition system and the computer vision algorithm developed in this study suggested suitability for its use in both laboratory and field conditions and shows promise as a proximal soil sensor.

Predicting total dissolved salts and soluble ion concentrations in agricultural soils using portable visible near-infrared and mid-infrared spectrometers

Soil salinization is the primary obstacle to sustainable agricultural development in arid regions. Because total dissolved salts and soluble ion content are the primary indicators of the degree of soil salinization, their accurate estimation is essential to the determination of appropriate soil salinization remediation techniques, irrigation regimes, and the agricultural distribution layout. A total of 261 soil samples were collected from agricultural fields in the province of Xinjiang, China. A portable Fourier transform (FT) mid-infrared (MIR) spectrometer (4000–600 cm−1) and a visible near-infrared (VNIR) field spectrometer (350–2500 nm) were used to obtain soil spectra. We subsequently used partial least-square regression (PLSR) and support vector machine (SVM) algorithms to establish models in VNIR, MIR, and VNIR–MIR regions. The main objectives of this study are (i) to investigate the possibility of using spectroscopic techniques to predict total dissolved salts and soluble ion content; (ii) to compare the prediction accuracy of these soil properties in the VNIR, MIR, and VNIR–MIR spectral regions; (3) to compare the prediction accuracy with linear and nonlinear algorithms. Our findings demonstrated that spectroscopic techniques are a promising way to predict total dissolved salts and soluble ion content. Good predictions were obtained for total dissolved salts content, HCO 3 − , SO 4 2 − and Ca2+, satisfactory for Mg2+, Cl−, and Na+, but poor for K+. This work demonstrates the potential of portable VNIR and MIR spectrometers as proximal soil sensors for more efficient soil analysis and acquisition of soil salinity information.

Effects of Soil Surface Roughness on Soil Reflectance Measured in Laboratory and Outdoor Conditions

Various remote sensing methods, including diffuse and contact reflectance spectroscopy (DRS and CRS, respectively) as well as proximal soil sensing (PSS), have been widely used for obtaining information about the soil properties. The influence of soil surface roughness on soil spectra obtained with a FieldSpecPro (ASD) spectroradiometer using three measurement techniques was investigated. The PSS spectra in outdoor (PSS-O) and laboratory (PSS-L) conditions were collected and then compared with the spectra of the same soil recorded using Muglight (ML) adapter. PSS measurements of each soil sample were taken at three surface roughness levels: the lowest roughness level, representing the rolled surface, the second representing a moderately rough (MR) surface of medium-sized clods and aggregates, and the third level with the highest soil surface roughness, where the clods and aggregates, with size beyond their assessed mean size for a given soil. The roughness indices height standard deviation (HSD) and tortuosity ( $\mathbf{T}_\mathbf{3D}$ ) were closely related to the size of the aggregates and clods forming the soil surface. The ML spectra of ground samples were more strongly correlated with the PSS-L spectra than with the PSS-O spectra, and mean coefficient of determination ( $\mathbf{R}^\mathbf{2}$ ) for the spectra obtained at three levels of roughness was 0.99 and 0.95, respectively. With the increase of surface roughness, the differences between the level of ML and both PSS-L and PSS-O spectra increased. The relative difference (RD) calculated for PSS-L and PSS-O spectra in relation to ML spectra was better explained by the $\mathbf{T}_\mathbf{3D}$ ( $\mathbf{R}^{\mathbf{2}} = \mathbf{0.82}$ ), than the HSD ( $\mathbf{R}^{\mathbf{2}} = \mathbf{0.68}$ ).

Baseline estimates of soil organic carbon by proximal sensing: Comparing design-based, model-assisted and model-based inference

For baselining and to assess changes in soil organic carbon (C) we need efficient soil sampling designs and methods for measuring C stocks. Conventional analytical methods are time-consuming, expensive and impractical, particularly for measuring at depth. Here we demonstrate the use of proximal soil sensors for estimating the total soil organic C stocks and their accuracies in the 0–10cm, 0–30cm and 0–100cm layers, and for mapping the stocks in each of the three depth layers across 2837ha of grazing land. Sampling locations were selected by probability sampling, which allowed design-based, model-assisted and model-based estimation of the total organic C stock in the study area. We show that spectroscopic and gamma attenuation sensors can produce accurate measures of soil organic C and bulk density at the sampling locations, in this case every 5cm to a depth of 1m. Interpolated data from a mobile multisensor platform were used as covariates in Cubist to map soil organic C. The Cubist map was subsequently used as a covariate in the model-assisted and model-based estimation of the total organic C stock. The design-based, model-assisted and model-based estimates of the total organic C stocks in the study area were similar. However, the variances of the model-assisted and model-based estimates were smaller compared to those of the design-based method. The model-based method produced the smallest variances for all three depth layers. Maps helped to assess variability in the C stock of the study area. The contribution of the spectroscopic model prediction error to our uncertainty about the total soil organic C stocks was relatively small. We found that in soil under unimproved pastures, remnant vegetation and forests there is good rationale for measuring soil organic C beyond the commonly recommended depth of 0–30cm.

Exploring the predictability of soil texture and organic matter content with a commercial integrated soil profiling tool

SummaryIn soil mapping, combining information from conceptually different proximal soil sensors can increase the accuracy of prediction and robustness of the model when compared with using individual sensors. In this study the predictability of soil texture (clay, silt and sand fractions) and soil organic matter (SOM) content was tested with a commercial integrated soil profiling tool that included sensors for measuring apparent electrical conductivity (ECa), reflectance in the visible and near‐infrared (vis‐NIR) parts of the electromagnetic spectrum and insertion force (IF). The measurements were made at 20 locations on each of two Swedish farms. At every location, sensor measurements were made at 1.5‐cm intervals from the soil surface to a depth of 0.8 m. Soil samples were collected close to the sensor measurement points and analysed for texture and SOM content. Farm‐specific calibrations were developed for texture and SOM with each sensor separately and with combinations of all three sensors. The calibrations were made using both partial least squares regression (PLSR) and simple linear regression. The results for the two farms were quite consistent in terms of rank in prediction performance between the individual sensors and the sensor combinations. The vis‐NIR spectrometer was the best individual sensor for predicting the soil properties tested on both farms, with root mean square error of cross‐validation (RMSECV) of 0.3–0.5% for SOM, about 6% for clay and silt and 10–11% for sand. The inclusion of IF reduced the RMSECV for predictions of SOM content by about 10%. For soil texture, including ECa reduced the RMSECV on average for all particle size fractions by 5–10%. However, the small improvements obtained by combining sensors do not provide strong support for combining vis‐NIR sensor measurements with measurements of ECa and or IF.