Spatial Variability Of Soil Properties Research Articles

Predicting wetland soil properties using machine learning, geophysics, and soil measurement data

Abstract Purpose Machine learning models can improve the prediction of spatial variation of wetland soil properties, such as soil moisture content (SMC) and soil organic matter (SOM). Their performance, however, relies on the quantity of data used to train the model, limiting their use with insufficient data. In this study, we assessed the use of synthetic data constrained by limited field data for training an eXtreme Gradient Boosting (XGBoost) algorithm used to predict the distribution of soil properties based on geophysical measurements constrained by soil samples. Materials and methods A spatial distribution of soil apparent electrical conductivity (ECa) and laboratory measurements of SOM and SMC from twenty-two core samples were acquired at the St. Michael restored wetland near Defiance, Ohio. The correlations between ECa, SOM, and SMC were explored for predicting the spatial distribution of SOM and SMC. We used a Beta Variational AutoEncoder (β-VAE) approach to synthetically generate over 70,000 training data from the original twenty-two data from soil cores. The training data samples were taken from the latent space. The XGBoost algorithm was then trained on the β-VAE generated data and used to predict the spatial distribution of SOM and SMC at the site. We also validated the accuracy of the XGBoost predictions using an original holdout model validation technique. Results and discussions The generated synthetic data using the β-VAE include both soil attributes and ECa, which are larger and more diverse than the original training set with an absolute mean reconstructed error for SMC and SOM ranging from 0.018 to 0.022 and 0.026 to 0.041, respectively. This indicates that the β-VAE successfully generated a realistic synthetic dataset and overcame the technical barrier of using limited datasets. In addition, using generated data to expand the original training data helps the XGBoost model make more accurate predictions compared to training on the original data. The XGBoost prediction performance yielded average Lin’s concordance correlation coefficient (LCCC) values of 0.82 and 0.85 for SOM and SMC and a ratio of performance to deviation (RPD) values of 1.92 and 2.22 respectively, indicating a good performance. Conclusions This study validated the use of β-VAE to successfully generate synthetic wetland soil datasets with attributes of the original field data that can be effectively used to train the machine learning XGBoost model. The proposed framework offers an efficient solution for mapping the spatial variability of soil property in data-scarce wetland soil environments.

A Study of the Multivariate Spatial Variability of Soil Properties, and their Association with Vine Vigor Growing on a Clayish Soil

A Study of the Multivariate Spatial Variability of Soil Properties, and their Association with Vine Vigor Growing on a Clayish Soil

Coupled heat and water transfer in heterogeneous and deformable soils: Numerical model using mixed finite element method

We present a generic model framework for coupled heat and water transfer (CHWT) in deformable (non-rigid) soils with spatial variations of soil properties. The model backbone is a mixed finite element method (FEM), which solves the Philip and de Vries (1957) CHWT model and achieves conservation of mass and energy on both local and global scales. Spatial variations occur in soil hydraulic and thermal properties due to transient water content and temperature distributions. Based on the mixed FEM scheme, a gradient measure and a clustering model (“k-means”) are proposed to trace the regions with large spatial variations of soil properties, and an adaptive mesh refinement technique is developed to improve the spatial resolution and simulation accuracy. Deformation perturbates local soil topography and alters transient soil water and temperature regimes in the deformed regions. A quasi-static deformation model is presented, and the deformation effects are incorporated into the mixed FEM scheme. When external load exists, soil deformation is simulated with an updated Lagrangian formulation, and the local water content and temperature variations due to soil volume changes are also updated in the CHWT model. Numerical examples, including thermally induced soil water transfer and water infiltration, illustrate the model ability to provide plausible CHWT results, especially the refined solutions near the wetting fronts and the water content and temperature distributions when the soil is deformable. In conclusion, the proposed model framework provides an effective pipeline to incorporate and process the spatial variations of soil properties and soil deformation in CHWT simulations.

Time series prediction of reservoir bank landslide failure probability considering the spatial variability of soil properties

Historically, landslides have been the primary type of geological disaster worldwide. Generally, the stability of reservoir banks is primarily affected by rainfall and reservoir water level fluctuations. Moreover, the stability of reservoir banks changes with the long-term dynamics of external disaster-causing factors. Thus, assessing the time-varying reliability of reservoir landslides remains a challenge. In this paper, a machine learning (ML) based approach is proposed to analyze the long-term reliability of reservoir bank landslides in spatially variable soils through time series prediction. This study systematically investigated the prediction performances of three ML algorithms, i.e. multilayer perceptron (MLP), convolutional neural network (CNN), and long short-term memory (LSTM). Additionally, the effects of the data quantity and data ratio on the predictive power of deep learning models are considered. The results show that all three ML models can accurately depict the changes in the time-varying failure probability of reservoir landslides. The CNN model outperforms both the MLP and LSTM models in predicting the failure probability. Furthermore, selecting the right data ratio can improve the prediction accuracy of the failure probability obtained by ML models.

Interpolation of non-stationary geo-data using Kriging with sparse representation of covariance function

Interpolation of spatially varying measurements, e.g., spatial variation of a geotechnical property along depths, is often needed in many disciplines, including geotechnical engineering. Among various interpolation methods, geostatistical methods, such as Kriging, are popular since they are able to not only provide the best estimates but also quantify the interpolated uncertainty. However, Kriging relies on an assumption of data stationarity, and a transformation from non-stationary data to stationary data by, e.g., detrending, is required when using Kriging to interpolate non-stationary data. Detrending, however, is a tricky and long-lasting challenge in geo-data analyses. To tackle this challenge, this study develops a novel Kriging method for direct interpolation of spatially non-stationary data without detrending. A sparse representation of covariance function is proposed for Kriging interpolation of spatially non-stationary data, bypassing the tricky detrending process. The proposed Kriging method is particularly appealing for non-stationary geotechnical site investigation data often encountered in geotechnical practice (e.g., spatial variation of soil properties over different soil layers along depth). Equations are derived for the proposed Kriging method, and it is illustrated and validated using numerical examples. Results show that the proposed Kriging method with sparse representation of covariance function is directly applicable to spatially non-stationary data and has the advantages of improved accuracy and reduced interpolation uncertainty when interpolating spatially varying, non-stationary data.

A comparison of ground motions predicted through one-dimensional site response analyses and three-dimensional wave propagation simulations at regional scales

One-dimensional (1D) site response analysis (SRA), which considers vertically propagating seismic waves from the bedrock to the surface, has been a common technique among geotechnical engineers to examine site-specific ground shaking. However, observations from past earthquakes and analytical studies indicate that idealizations ingrained in 1D SRA may be too severe to capture the ground truth, such as the omissions of spatial variability of soil properties, surface topography, and basin and directivity effects. Physics-based three-dimensional ground motion simulations (GMSs) can incorporate these factors and yield more reliable predictions. In this study, we utilize ground motions from 57 physics-based broadband (from 0 to 8–12 Hz) GMS for a region of Istanbul. A total of 2912 sites with experimentally measured soil profiles that are distributed over the 30 km-by-12.5 km area are also modeled as soil columns and analyzed through 1D SRA. The ground responses from 1D SRA and three-dimensional (3D) GMS are then compared for all 57 earthquake scenarios. These systematic comparisons are then used for examining model features that are correlated with variations in the ratios of various ground motion intensity measures (IMs) and for developing regression-based formulas that can be used for determining simple factors for the considered region to correctly scale (up or down) the site-specific ground motion intensities obtained from 1D SRA, including peak ground acceleration (PGA), peak ground velocity (PGV), and spectral acceleration ( Sa) values.

Investigation on wave-induced responses of heterogeneous seabed around a buried pipeline based on nonstationary random field model

The instability of the seabed poses a threat to the structural integrity and functionality of submarine pipelines. This numerical study investigates wave-induced responses in a heterogeneous seabed around a buried pipeline, considering spatial variability in soil properties through nonstationary random field models. Three scenarios of spatial variability for soil parameters changing with depth are simulated using both stationary and nonstationary random field models via the Karhunen-Loève expansion method. The discrete random fields are validated in terms of correlation structures and statistical characteristics. The stochastic finite element method, incorporating random fields and a poroelastic seabed model through a LiveLink code, is employed to examine probabilistic outcomes of dynamic soil responses and liquefaction zone distributions within a probabilistic framework. Contrastive analyses are conducted for seabed models with different random field scenarios, in addition to deterministic analyses. The results of the stochastic analysis show that the increasing trend of soil strength with depth has significant effects on the seabed responses and transient liquefaction characteristics around a pipeline.

Evaluating Intra-Field Spatial Variability for Nutrient Management Zone Delineation through Geospatial Techniques and Multivariate Analysis

This research study underscores the importance of effectively managing soil nutrients in a site-specific manner to enhance crop productivity while considering the spatial variability of the soil. The objective is to identify subfields with similar soil characteristics, referred to as management zones (MZs), to promote sustainable land utilization. This study was conducted in two central pivot fields located in Southern Idaho, USA, where barley and sugar beets were grown. Soil samples were collected from each field in a grid pattern and analyzed for various chemical properties. These properties included soil pH, organic matter, cation exchange capacity, excess lime, electrical conductivity, total inorganic nitrogen, phosphorus, potassium, calcium, magnesium, zinc, iron, manganese, copper, and boron. Descriptive statistics and normality assessments were performed, and the coefficient of variation was calculated to assess the heterogeneity of soil properties, revealing significant variability. To determine the spatial variability of soil properties, ordinary kriging was used revealing diverse spatial patterns for each location and soil variable examined with moderate to strong spatial dependence. To develop the MZs, a combination of principal component analysis and fuzzy k-means clustering was utilized, and specific parameters that represented the overall variability of soil properties in each field were identified. Based on the identified parameters, two clusters were created in each field. The first management zone (MZ1) exhibited lower values of soil pH, excess lime content, and electrical conductivity compared to the MZ2. Consequently, higher crop productivity was observed in MZ1 in both fields. The biomass yields of barley and sugar beets in MZ1 surpassed those in MZ2. This study highlights the effectiveness of the methodology employed to delineate MZs, which can be instrumental in precise soil nutrient management and maximizing crop productivity.

SEISMIC MITIGATION EFFECT FOR LARGE-SPACE UNDERGROUND STRUCTURES CONSIDERING SPATIALLY VARYING SOIL PROPERTIES

The seismic response of the large-space underground structure (LSUS) is significantly influenced by the physical properties of the surrounding soil media, while the soil owns a strong spatial variability. This study proposes a seismic response analysis process of the soil-LSUS interaction system is proposed, which can consider the characteristic of the spatially distributed soil properties. The proposed process begins with establishing the spatially random field model of the soil properties using the improved latent space method. Then, the model is calibrated based on the real data and Bayesian approach, and the realization of the random field is accomplished. Further, the soil-LSUS interaction finite element (FE) model is established, which incorporating the soil physical properties generated from the random field. Finally, the nonlinear time-history analysis of the soil-LSUS interaction FE model is conducted. As an illustration of the proposed process, a typical LSUS located in Guangzhou is selected as an example, and the seismic mitigation measure which the lead-filled steel tube damper (LFSTD) is installed between the intermediate column and the top beam is adopted for the LSUS. The influence of the spatial variability of soil properties on the seismic mitigation effect of the LSUS is investigated. Results indicate that the spatial variability of the soil properties can cause a minor influence on the force and deformation of the intermediate column and the energy dissipation ratio between the LFSTD and structure, while it can bring a significant influence on the maximum deformation and force and the shape of the hysteresis loop of the LFSTD.

Assessing the occurrence of soil improvement and its relationship to the dominant life form in the high mountains of Central Spain

Soil in mountainous regions is vital to the health and preservation of these unique and diverse ecosystems. In dry and semi-arid regions, vegetation patches play a crucial role in soil nutrient heterogeneity through continuous feedback with soils, acting as barriers to collect runoff water, sediments, and nutrients from bare soil regions. Soil amelioration, an enhancement of soil biogeochemical processes, results in the formation of “fertility islands,” the extent of which is contingent on the plant species in question, as well as nutrient dynamics and water availability. Here, we collected soil from across a two-peak altitudinal gradient in the Sierra de Guadarrama high-mountains, each peak featuring a different dominant vegetation type (herbs vs. cushion-like) to compare soil nutrients and properties between bare soil and vegetation-covered patches (microhabitats). Soil improvement was assessed in the microhabitats using the Relative Interaction Index (RII). Fertility islands were shown to be prevalent in high-mountain ecosystems, as soil quality and properties were higher beneath vegetation-covered regions than bare soils. There was a difference in RIIs between the transects, with greater soil improvement in the cushion-dominated transect than the herb-dominated one. Changes in nutrient levels were unrelated to patch successional stage, indicaing that plant generations may not shape the spatial variability of soil attributes. Instead, species variety or the presence of dominant clonal species increased soil nutrients and aggregate stability, highlighting the importance of root shape and high biomass in nutrient retention and soil reinforcement. Finally, our findings imply that the poor, shallow soils in the examined peaks, in comparison to other mountains, may account for the poor facilitative interactions. Competition for the scarce resources at these peaks may intensify as climate warms. Thus, while these plants may grow with minimum assistance under current climate circumstances, their associations may be especially vulnerable to climate change.

Analysis of tunneling-induced ground movements in spatially variable soil under the influence of existing building

Tunnel construction invariably results in ground movement, which is significantly impacted by the spatial variability of soil properties and proximity of existing buildings. However, few studies have examined the coupled effect of these two factors on tunneling-induced ground movement. This study establishes a stochastic finite element model that incorporates both the spatial variability of soil properties and influence of existing buildings. The model is used to perform a series of random analyses, studying the effects of story number, eccentricity, and building length on ground settlements induced by tunneling in soil with spatial variability of Young’s modulus. A probabilistic analysis is conducted to investigate the uncertainties associated with ground settlements. The results of the analysis indicate that the presence of the building enhances the influence of the spatial variability of soil, leading to an increase in the dispersion of ground settlement. For high-rise buildings, the reduction of building eccentricity and length will increase the uncertainty of surface settlement. Neglecting the influence of the existing building in the probability analysis of surface settlement can result in an overestimation of the failure probability.

Mapping soil property variability in oil palm (Elaeis guineensis jacq.) plantationsof Krishna basin of Andhra Pradesh through geospatial technologies

Sixty seven soil samples were collected from oil palm growing regions of NTR District of Andhra Pradesh and were analysed for different soil properties. The data was subjected to descriptive analysis to find out the variability and then to geostatistical analysis to design best fit models having minimal error for developing prediction maps. Highest variation was observed in exch. Mg and the lowest in pH. The mean values of soil pH, EC (dS/m), OC (g/kg), Olsen-P (mg/kg), NH4OAc-K (mg/kg), Exch. Ca (mg/kg), Exch. Mg (mg/kg), CaCl2-S (mg/kg) and HWB (mg/kg) were 7.32 ± 0.08, 0.25 ± 0.02, 0.87 ± 0.03, 101.47 ± 6.95, 566.14 ± 42.97, 4.72 ± 0.24, 2.46 ± 0.27, 60.86 ± 2.6 and 5.98 ± 0.25 respectively in the surface layer (0-20 cm) of the soil. Geostatistical analysis revealed wider spatial variability in surface soil properties and they had circular, spherical, stable and exponential best fit semi variogram models for evaluating dependency. The wide spatial variability of soil properties warrants site specific nutrient management for higher oil palm production. Nutrient distribution maps could be developed through kriging interpolation to interpolate the nutrient levels in unsampled areas.

Random analysis method for nonlinear interaction between shield tunnel and spatially variable soil

The spatial variability of soil properties is likely to have a remarkable influence on the longitudinal responses of shield tunnels. This paper proposes a one-dimensional random analysis method (RAM) for the interaction between the shield tunnel with circumferential joints and the spatially variable soil with nonlinear elastoplastic behaviors. A deterministic short beam-spring model with longitudinal variation in the initial subgrade reaction coefficient and the ultimate soil resistance is developed and solved by the finite difference method. The random field of the soil properties discretized by the Karhunen-Loeve expansion method is used to establish the RAM. A hypothetical example is further presented to demonstrate the applicability of the proposed method. The results imply that the soil nonlinearity may lead to a negative effect on the reliability control of the tunnel structure and should be considered to obtain more reasonable probabilistic evaluations of the tunnel deformations. The increasing joint rotational stiffness or shearing stiffness results in the lower variabilities of the maximum tunnel settlement, and reduces the probabilities of the maximum joint opening or joint dislocation exceeding the allowable value, respectively.

Delineation of Productive Zones in Eastern China Based on Multiple Soil Properties

Accurate soil management has long been the focus of research in agroecology. Crop productivity can be enhanced while reducing environmental threats from excessive fertilization by fully comprehending the spatial variability of soil properties and delineating management zones (MZs). A field investigation was carried out at experimental sites outside Hefei City’s administrative districts in China to study the spatial variability of soil properties and the delineation of MZs. A total of 9601 soil samples were collected in the study area. A variety of soil properties were analyzed, including the pH, organic matter, total nitrogen, alkali-hydrolyzable nitrogen, available phosphorus, available potassium, slowly released potassium, available sulfur, available boron, available copper, available zinc, available iron, and available manganese. The coefficient of variation for various soil properties exhibited a wide range, spanning from 12.2% to 100.5%. The geostatistical results show that most soil properties have moderate to strong spatial autocorrelation, and the ordinary kriging method is used to map the distribution of soil properties. The principal component analysis method was used to reduce the dimension of 13 soil properties to 4 principal components, and the fuzzy c-means clustering method was used to delineate MZs. The calculation results of the fuzzy performance index and normalized classification entropy show that the optimum number of MZs is five. In the study area, the western part exhibits the highest soil fertility, primarily attributed to its elevated organic matter content. Additionally, organic matter emerges as a key factor influencing sustainable agricultural production in this region. These results form the basis for soil managing areas outside the administrative districts of Hefei City.

Probabilistic analysis of ground settlement induced by tunnel excavation in multilayered soil considering spatial variability

Accurately evaluating and predicting the ground settlement during tunnel excavation are crucial for ensuring the stability of tunnel structures. However, the traditional methods always use representative (or average) values for deterministic analysis and rarely consider the spatial variability of soil properties. In this study, a probabilistic analysis of ground settlement under uncertain soil properties was performed based on random fields combined with the finite-difference method. The random fields of Young’s modulus and friction angle of multilayered soils were generated based on the Cholesky decomposition in a manner consistent with a specified numerical mesh. The effects of spatial variability and cross-correlation in random fields on the ground settlement were investigated in detail. The results demonstrate that spatially variable soils significantly influence the probability of failure, magnitude, and longitudinal profile of ground settlement in the vertical direction. It can be found that the allowable ground settlement of the limit state function also dominated the degree of the vertical scale of fluctuation on stability. Neglecting the cross-correlation between the two random fields results in an overestimation of the probability of failure compared with an independent analysis. Subsequently, the failure mechanisms of ground settlement in spatially random soils were discussed.

Investigation of topographic amplification on ground motions considering spatial variability of soil properties

Investigation of topographic amplification on ground motions considering spatial variability of soil properties

Mechanism of seismic liquefaction for heterogeneous soil

In addition to inducing uncertainty in the predicted response, natural spatial variability of soil properties affects the mechanical response of geotechnical structures. When a failure surface is involved in the response, this surface can deviate from its theoretical location to pass through weaker zones of material. For the case of seismically induced soil liquefaction, it has been found that a larger amount of excess pore water pressure is generated in a soil exhibiting small-scale variability of its properties than in the corresponding uniform soil having geomechanical properties equal to the average properties of the heterogeneous soil. An explanation for this important phenomenon is provided in this paper based on the results of centrifuge experiments and numerical simulations of heterogeneous and homogeneous saturated soil deposits subjected to seismic loads. It is demonstrated, based on a detailed analysis of the results, that partial drainage during the earthquake, consisting of pore water inflow from loose soil zones that liquefy first toward surrounding dense areas, may trigger softening of dilative soils. The water inflow effects in terms of volumetric strains leading to reduced cyclic resistance of dense sands are compared with results of specially designed cyclic triaxial tests reported by other researchers.

Spatial variation in soil aggregate stability and erodibility at different slope positions in four hilly regions of northeast China

The spatial heterogeneity of erosion can affect soil properties in agricultural areas, which can in turn alter the soil aggregate stability and erodibility. However, it remains elusive how aggregate stability and soil erodibility vary at different positions in the Mollisol region of northeast China, and how soil properties drive these variations. This study selected different slope positions across a latitudinal gradient in four hilly regions from cold temperate (Jiusan, JS; Keshan, KS) to middle temperate (Hailun, HL; Binxian, BX) areas in northeast China. The spatial variations of soil properties, geometric mean diameter (GMD), mean weight diameter (MWD), soil erodibility (K), and soil structural stability index (SSI) were calculated at topsoils (0–20 cm) and subsoils (20–40 cm). Total nitrogen (TN) and soil organic carbon (SOC) decreased with increasing latitudes, and gradually declined and finally rose with decreasing slope positions. MWD, GMD, and SSI were the highest in JS, and first obviously decreased from the top position to lower position, and then increased at the bottom position. The K followed an order of BX > HL > KS > JS, and decreased in the bottom position compared with that in the middle and lower positions across the four regions, which could be ascribed to the new aggregate formation promoted by fine soil particle deposition at the slope bottom. Redundancy analysis and variance decomposition analysis indicated that SOC, TN, and bulk density had higher vector values, and region (interpretation rate = 57.97 %) had the greatest influence on aggregate stability and soil erodibility. Available potassium (AK), saturated conductivity (KS), SOC, TN and soil texture were the main factors driving the variation of MWD and K. These findings highlight the spatial variations in soil aggregate stability and erodibility in Mollisol regions, providing a scientific basis for soil restoration in eroded farmland.

Efficient slope reliability analysis under soil spatial variability using maximum entropy distribution with fractional moments

Spatial variability of soil properties imposes a challenge for practical analysis and design in geotechnical engineering. The latter is particularly true for slope stability assessment, where the effects of uncertainty are synthesized in the so-called probability of failure. This probability quantifies the reliability of a slope and its numerical calculation is usually quite involved from a numerical viewpoint. In view of this issue, this paper proposes an approach for failure probability assessment based on Latinized partially stratified sampling and maximum entropy distribution with fractional moments. The spatial variability of geotechnical properties is represented by means of random fields and the Karhunen-Loève expansion. Then, failure probabilities are estimated employing maximum entropy distribution with fractional moments. The application of the proposed approach is examined with two examples: a case study of an undrained slope and a case study of a slope with cross-correlated random fields of strength parameters under a drained slope. The results show that the proposed approach has excellent accuracy and high efficiency, and it can be applied straightforwardly to similar geotechnical engineering problems.

Variation and dynamics of soil properties of ordinary chernozem in the Saratov region

Based on data from an agrochemical survey using GIS technologies at test site No. 2 in 2009-2016 within the framework of the general scheme of soil-ecological monitoring of the Saratov region, an assessment was made of the dynamics and spatial variation of soil properties of ordinary chernozem, depending on the steepness and exposure of the slope. Soil and agrochemical characteristics in conditions of a pronounced relief demonstrate a certain dependence on the location and are very variable over time. Soil sampling on an area of more than 2000 hectares using navigation equipment made it possible to obtain agrochemical cartograms in GIS and analyze them using zonal statistics, taking into account a digital elevation model. The spatial variability of soil properties is often lower in leveled areas or short transition sections of the slope, while the maximum variability is noted on steeper slopes. The highest values of indicators are typical for flat watershed surfaces, which are the most fertile, however, the closest relationship between humus and other agrochemical indicators was noted just on steeper slopes. The variability of fertility indicators on differently directed slopes was within 28–36 % for P2O5, 20 % for K2O, and 4–6 % for pH. The greatest differences in the variability of the polar slopes along the survey tours are noted for the content of humus in the soil. In 2009, the coefficient of variation was 19 and 30 %, and in 2016, 11 and 24 % on the slope of the northern and southern exposure, respectively. A retrospective analysis showed that the highest rates of decline were noted for the content of humus in the soil and the reaction of the environment on gently sloping slopes (by 2 and 1.5 times, respectively) relative to the average change in the massif. For mobile forms of phosphorus and potassium, the maximum decrease was noted on gentle slopes.