Estimate Soil Water Research Articles

The water vapor distribution profile in dry soil layer during evaporating with high spatial resolution monitored by distributed fibre-optic sensing technology

The dry soil layer (DSL) typically forms on the surface of sandy soils during the evaporation process, with water in the DSL moving solely as vapor. Understanding the DSL thickness and its water vapor distribution is crucial for accurately estimating soil water flux, improving evaporation mechanisms, and enhancing ecological conservation efforts. This study employed distributed fibre-optic sensing (DFOS) technology to achieve high spatial resolution measurements (up to 1 mm) of the water vapor distribution within the DSL. We measured soil water vapor profiles in a sand column subjected to evaporation over 24 days using a 60 cm long relative humidity sensing probe and an optical frequency domain reflection (OFDR) interrogator. We effectively captured the downward migration of the evaporating surface from the sand surface and quantified the inhibitory effect of DSL thickness on evaporation using a logarithmic function. Additionally, we identified a linear distribution of water vapor with depth in the DSL. Our findings offer novel insights into the role of DSL in the hydrological behaviour of unsaturated soils and demonstrate the potential of our approach for investigating dynamic changes in in-situ DSLs, particularly in arid and semi-arid regions.

Estimating soil water retention curves over the entire saturation range: A thermal conductivity-based approach

The relationship between soil water content (θ) and suction (h, referring to the absolute value of pressure head), is described by the soil water retention curve (SWRC). Our earlier research (Fu et al. (2021, 2023a) [J. Hydrol.127171]; [J. Hydrol.129898]) recognized underlying correlations between SWRCs and soil thermal conductivity (λ) versus θ curves, and developed methodologies to ascertain the parameters of the van Genuchten (vG) equation using λ(θ) measurements, described by the Ghanbarian & Daigle (GD) equation, and basic soil characteristics. Limitations intrinsic to the van Genuchten equation restrict the GD-vG approach to generate precise estimates only in the wet and medium suction range, specifically h ranging from 0 to 150 mH2O. The validity of these approaches in the dry region remains uncertain. In this study, we associated the Peters-Durner-Iden (PDI) model parameters to those of the GD model. An initial examination was performed on the linearization processes needed to derive the hydraulic continuity water content (θhc) from the capillary water component as characterized by the PDI model and to choose the suction at oven dryness (h0) based on PDI model performance. Subsequently, two piecewise functions and two pedo-transfer functions were formulated to compute the PDI model parameters utilizing soil porosity, particle size distribution, and GD parameters, based on a calibration dataset comprising 25 different soils. The new GD-PDI approach was subsequently assessed with six independent soils and juxtaposed with the previous GD-vG approach. The GD-PDI approach outperformed the GD-vG approach, particularly within the dry range.

Evaluation of Pedotransfer Functions to Estimate Soil Water Retention Curve: A Conceptual Review

The soil water retention curve (SWRC) is a vital soil property used to evaluate the soil’s water holding capacity, a critical factor in various applications such as determining soil water availability for plants, soil conservation and management, climate change adaptation, and mitigation of flood risks. Estimating SWRC directly in the field and laboratory is a time-consuming and laborious process and requires numerous instruments and measurements at a specific location. In this context, various estimation approaches have been developed, including pedotransfer functions (PTFs), over the past three decades to estimate soil water retention and its associated properties. Despite the efficiencies, PTFs and semi-physical approach-based models often have several limitations, particularly in the dry range of the SWRC. PTFs-based modeling has become a key research topic due to readily available soil data and cost-effective methods for deriving essential soil parameters, which enable more efficient decision-making in sustainable land-use management. Therefore, advancement and adjustment are necessary for reliable estimations of the SWRC from readily available data. This article reviews the evaluation of the current and past PTFs for estimating the SWRC. This study aims to evaluate PTF techniques and semi-physical approaches based on soil texture, bulk density, porosity, and other related factors. Additionally, it also assesses the performance and limitations of various common semi-physical models proposed and developed by Arya and Paris, Haverkamp and Parlange, the Modified Kovács model by Aubertin et al., Chang and Cheng, Meskini-Vishkaee et al., Vidler et al., and Zhai et al. This assessment will be effective for researchers in this field and provide valuable insight into the importance of new PTFs for modeling SWRC.

A Study on Hyperspectral Soil Moisture Content Prediction by Incorporating a Hybrid Neural Network into Stacking Ensemble Learning

The accurate prediction of soil moisture content helps to evaluate the quality of farmland. Taking the black soil in the Nanguan District of Changchun City as the research object, this paper proposes a stacking ensemble learning model integrating hybrid neural networks to address the issue that it is difficult to improve the accuracy of inversion soil moisture content by a single model. First, raw hyperspectral data are processed by removing edge noise and standardization. Then, the gray wolf optimization (GWO) algorithm is adopted to optimize a convolutional neural network (CNN), and a gated recurrent unit (GRU) and an attention mechanism are added to construct a hybrid neural network model (GWO–CNN–GRU–Attention). To estimate soil water content, the hybrid neural network model is integrated into the stacking model along with Bagging and Boosting algorithms and the feedforward neural network. Experimental results demonstrate that the GWO–CNN–GRU–Attention model proposed in this paper can better predict soil water content; the stacking method of integrating hybrid neural networks overcomes the limitations of a single model’s instability and inferior accuracy. The relative prediction deviation (RPD), root mean square error (RMSE), and coefficient of determination (R2) on the test set are 4.577, 0.227, and 0.952, respectively. The average R2 and RPD increased by 0.056 and 1.418 in comparison to the base learner algorithm. The study results lay a foundation for the fast detection of soil moisture content in black soil areas and provide a data source for intelligent irrigation in agriculture.

Temperature and water content estimation in soils of the semi‐arid region of Brazil using finite difference and CFD

AbstractDetermining the temperature and water content of soil, at a given instant or along time, is fundamental to understand several soil‐related phenomena and processes. Evaporation, aeration, chemical‐reaction rates and types, biological processes such as germination and growth of seeds, root development, nutrient and water uptake by roots, and decomposition of organic matter by microbes, are all strongly influenced by soil temperature. On the other hand, infiltration of water through the soil surface allows soil to temporarily store water, making it available for uptake by plants and organisms living in soil. Furthermore, soil water content is closely related to physical and chemical properties of soil, such as oxygen content and demand, which impacts root breathing, microbial activity and soil chemical balance. The accurate evaluation of these two parameters and their interconnection is even relevant in semi‐arid regions, where climate conditions are particularly difficult, such as the north‐eastern zone of Brazil. Thus, the use of computational models and coupled approaches are imperative for rigorous descriptions. This work presents a contribution to estimate soil temperature and water content, by solving the heat transfer equation and the Richards equation, respectively, through finite differences. As input, the model uses the experimental material composition of the soil, the time‐dependent temperature profile at the surface and information about the regional rain regime. Three different numerical approaches were implemented: explicit, simple implicit and the Crank–Nicolson method. The calculations for temperature and water content of the soil obtained with these computational models were compared with the results from Computational Fluid Dynamics (CFD). The relative differences between the numerical methods were less than 0.006% by solving the heat transfer equation and less than 2.75% using the Richards equation. The maximum relative differences within the model, including both a constant and a variable water‐content profile, were 3.28%. The results from the computational model using the CFX tool have maximum relative differences of 0.6%, which contributes to verifying the accuracy of the implemented methods.

Developing novel ensemble models for predicting soil hydraulic properties in China’s arid region

Accurately quantifying the mass (water, nutrients, and carbon) and energy exchange processes between the Earth’s atmosphere, biosphere, and lithosphere requires the accurate parameterization of soil hydraulic properties (SHPs) and their spatial heterogeneity. Because direct measurements of SHPs are difficult, time-consuming, and impossible at larger spatial scales, various pedotransfer functions (PTFs) have been developed in the last few decades, providing divergent estimates of SHPs from readily measurable variables. However, existing PTFs are mostly developed for specific regions and may not be suitable for other pedoclimatic conditions. Here, PTFs were developed using multiple machine-learning algorithms to estimate SHPs and examined the weaknesses and strengths of each method in estimating the average and variability of SHPs across China’s arid region. The optimal PTFs were applied to a 1 × 1 km2 regional map of texture and bulk density, thus producing maps of the saturated hydraulic conductivity (Ks), the parameters of the van Genuchten (VG) formulation, field capacity (θfc), wilting point (θwp), plant available water (θpa), and soil macroporosity (ϕm) in the 0–2 m soil profile throughout the region. The results indicate that the ensemble model with the averaging method (EMA) is the most robust for estimating all SHPs. The EMA-PTFs for Ks yielded the best performance for sand textures, followed by sandy loam, loam, silty loam, silty clay loam, loamy sand, and clay loam textures. There were no significant differences in estimating soil water retention curve parameters among the different soil texture classes. Using the same observed data set, we demonstrated that the new EMA-PTFs outperformed those of existing PTFs, such as Rosetta and HYPRES, with RMSE values decreasing by 25–83 % depending on the SHPs. Furthermore, our SHP datasets exhibited significantly deeper soil profiles and higher accuracy than other available regional and global SHP products. The VG retention parameter α shows the greatest variation vertically throughout the 0–2 m soil profile, followed by ln(Ks), θr, θpa, θfc, θwp, ϕm, θs and n. All SHPs exhibit much lower variability in the desert than in other areas, likely due to the homogeneous particle size distribution of desert areas. This study provides more accurate SHP estimates in China’s arid region than the existing SHP products by developing advanced PTFs and highlights the inconsistency in SHPs among different global or regional products; the uncertainties induced by PTFs should thus be considered in future terrestrial biosphere modeling.

Estimating soil water content from thermal images with an artificial neural network

Data of soil water content measured at regular time intervals could be used to set an irrigation schedule for farmland. Changes in soil water content could also provide an early warning of potential slope slippage. An artificial neural network is proposed for detecting soil water content from five variables: soil and water temperature changes after sunlight exposure, plant coverage rates, fine-grained soil content, and effective soil particle size. Temperature changes were detected using thermal images taken using an unmanned aerial vehicle equipped with a thermal camera in the morning and at noon. Plant coverage was estimated using visible light photograph, and the fine-grained soil content, effective particle size, and soil water content were determined from laboratory tests of soil samples. An optimized neural network was established by analyzing the performance of various neural network parameters. The mean square error of the optimized neural network reached a minimum value of 0.10061 in the validation data set at the 12th epoch, the training was stopped to avoid overfitting. The correlation coefficients for the training, validation, testing, and overall data sets were 0.880, 0.772, 0.810, and 0.847 respectively. The target values and the predicted values of the neural network are highly correlated. Thus, the proposed neural network is effective for predicting soil water content. A relative importance analysis revealed that soil and water temperature changes after sunlight exposure had the greatest influence on the neural network’s results.

A framework for estimating soil water characteristic curve and hydraulic conductivity function of permeable reactive media

The unsaturated behavior of permeable reactive barriers (PRB) is a critical component in predicting the removal efficiency through the adsorption of contaminants. This study investigates the framework to estimate the soil water characteristic curve (SWCC) and hydraulic conductivity function (HCF) for iron oxide-coated sand (IOCS) and zeolite, which are common materials used in PRBs. A multistep outflow (MSO) experiment was performed and the results of the MSO experiment were used to optimize associated parameters in Kosugi's SWCC and HCF. In addition, three scenarios of optimization analysis were investigated to evaluate the best-fitting model for estimating SWCC and HCF. The low root mean square error (RMSE) of fitted parameters indicates the Kosugi model well described the observed suction profiles in MSO experiments. In addition, the lowest RMSE and coefficient of variation suggested the inclusion of the additional parameter β provided the best estimation of the three materials (clean sand, IOCS, and zeolite). The physically reasonable estimation of SWCC and HCF of the three materials from the optimized parameters suggests the proposed framework is a reasonable model for the unsaturated behavior of PRBs.

Technical Note: Revisiting the general calibration of cosmic-ray neutron sensors to estimate soil water content

Abstract. Cosmic-ray neutron sensing (CRNS) is becoming increasingly popular for monitoring soil water content (SWC). To retrieve SWC from observed neutron intensities, local measurements of SWC are typically required to calibrate a location-specific parameter, N0, in the corresponding transfer function. In this study, we develop a generalized conversion function that explicitly takes into account the different factors that govern local neutron intensity. Thus, the parameter N0 becomes location independent, i.e. generally applicable. We demonstrate the feasibility of such a “general calibration function” by analysing 75 CRNS sites from four recently published datasets. Given the choice between the two calibration strategies – local or general – users will wonder which one is preferable. To answer this question, we estimated the resulting uncertainty in the SWC by means of error propagation. While the uncertainty in the local calibration depends on both the local reference SWC itself and its error, the uncertainty in the general calibration is mainly governed by the errors in vegetation biomass and soil bulk density. Our results suggest that a local calibration – generally considered best practice – might often not be the best option. In order to support the decision which calibration strategy – local or general – is actually preferable in the user-specific application context, we provide an interactive online tool that assesses the uncertainty in both options (https://cosmic-sense.github.io/local-or-global, last access: 23 February 2024).

Assessing Horton and Kostiakov models focused on estimating soil water infiltration

Water infiltration into the soil plays key role in proper soil and water conservation and management processes. The aim of the current study is to assess water behavior in soil infiltration processes by using the Horton and Kostiakov models. Double-ring infiltrometer method was used to determine water infiltration into the soil. In addition, soil samples were collected at depths of 0-10 cm, 10-20 cm and 20-30 cm, to determine soil moisture. Cumulative infiltration (IA) and instantaneous infiltration rates per time interval (TiR) were calculated. Equations describing the investigated phenomena were adjusted, according to the Horton and Kostiakov models. High water infiltration rate was observed at the beginning of the process, likely due to low moisture level at the top soil layer. Both models tend to underestimate the initial infiltration rate value. However, the Horton model tends to overestimate most of the instantaneous infiltration values up to 25 minutes. The Kostiakov model, on the other hand, tends to overestimate most of the instantaneous infiltration rate values after 80 minutes. Despite the large fluctuation observed in TiR values in the soil, the assessed models have satisfactorily described the infiltration behavior at the experimental site. Based on the heat map analysis, the Kostiakov model was the one capable of determining the infiltration values closest to actual one.

The Least Limiting Water Range to Estimate Soil Water Content Using Random Forest Integrated with GIS and Geostatistical Approaches

In this study, as a result of the determination of the Least limiting water range (LLWR) contents of alluvial lands with different soils, which are distributed in the Bafra Plain where intensive agricultural activities are carried out, the compression and aeration problems in the area are revealed with distribution maps. In addition, the predictability of LLWR was evaluated with the random forest (RF) algorithm, which is one of the machine learning algorithms, and the usability of the prediction values distribution maps was revealed. The LLWR contents of the soils varied in the range of 0.049- 0.273 cm3 cm-3 for surface soils. While there were aeration problems in 6.72%, compaction problems in 20.16%, both aeration and compaction problems in 0.8% of the surface soils examined in the study area, 72.32% were determined under optimal conditions. For the 20-40 cm depth, an aeration problem in the 5.88%, a compaction problem in the 28.57%, and both an aeration and a compaction problem in the 2.52% of the points were detected. In the estimation of LLWR with the random forest algorithm, the RMSE value obtained for 0-20 cm depth was determined as 0.0218 cm3 cm-3, and for 20-40 cm it was determined as 0.0247 cm3 cm-3 . In the distribution maps of the observed and predicted values obtained, the lowest RMSE value was determined by Simple Kriging interpolation methods for 0-20 cm depth and Ordinary Kriging interpolation methods for 20-40 cm. While the distribution of obtained and predicted values in surface soils was similar, variations were found in the distribution of areas with low LLWR below the surface. As a result of the study, it has been revealed that LLWR can be obtained with a low error rate with the RF algorithm, and distribution maps can be created with lower error in surface soils.

Assessing the Potential of UAV-Based Multispectral and Thermal Data to Estimate Soil Water Content Using Geophysical Methods

Knowledge of the soil water content (SWC) is important for many aspects of agriculture and must be monitored to maximize crop yield, efficiently use limited supplies of irrigation water, and ensure optimal nutrient management with minimal environmental impact. Single-location sensors are often used to monitor SWC, but a limited number of point measurements is insufficient to measure SWC across most fields since SWC is typically very heterogeneous. To overcome this difficulty, several researchers have used data acquired from unmanned aerial vehicles (UAVs) to predict the SWC by using machine learning on a limited number of point measurements acquired across a field. While useful, these methods are limited by the relatively small number of SWC measurements that can be acquired with conventional measurement techniques. This study uses UAV-based data and thousands of SWC measurements acquired using geophysical methods at two different depths and before and after precipitation to predict the SWC using the random forest method across a vineyard in the central United States. Both multispectral data (five reflectance bands and eleven vegetation indices calculated from these bands) and thermal UAV-based data were acquired, and the importance of different reflectance data and vegetation indices in the prediction of SWC was analyzed. Results showed that when both thermal and multispectral data were used to estimate SWC, the thermal data contributed the most to prediction accuracy, although multispectral data were also important. Reflectance data contributed as much or more to prediction accuracy than most vegetation indices. SWC measurements that had a larger sample size and greater penetration depth (~30 cm sampling depth) were more accurately predicted than smaller and shallower SWC estimates (~18 cm sampling depth). The timing of SWC estimation was also important; higher accuracy predictions were achieved in wetter soils than in drier soils, and a light precipitation event also improved prediction accuracy.

Gamma radiation for the estimation of mineral soil water content in a boreal forest

Continuous monitoring of water quantities in different soil horizons is necessary to understand the behavior of infiltrated water in the soil. Under certain conditions, using measurements of natural ground gamma radiation can help us estimate soil water content measurements over a 100 m2 surface within a 15 cm depth. A CS725 sensor can provide up to four daily estimates of soil water content by detecting the natural emission of gamma radiation. However, in boreal forest environments, gamma radiation mitigated by the water in the thick humus layers (litter, fermented, and humic horizon) can bias the underlying mineral soil water content measurements. The objective of this research was to evaluate the accuracy of methods that incorporate variables describing the surface humus layer into calculations of the underlying mineral soil water content, by measuring the soil’s natural gamma emission with the CS725. Using raw gamma radiation values obtained by CS725 sensors deployed over various boreal soils, we tested two functions. The first one included variables describing the humus layer and the other excluded these variables (manufacturer’s method). The function that included the descriptive humus layer variables showed superior results compared to the function without. The results of this study suggest that the CS725 sensor can adequately estimate mineral soil water content within ±10% absolute of the reference water content when examined with the following humus variables: humus layer thickness, fractioned composition, bulk density, and linear gamma radiation attenuation.

Point Pedotransfer Functions (PPTF) for Estimating Soil Water Retention Curve by Multiple Linear Regression Methods

The aim of this study was to derive and find the best Point Pedotransfer Functions (PPTF) for estimating SWRC for Iraqi soils using the multiple regression methods. Soil samples were collected from 30 different sites of Iraqi soil at two depths: 0-0.3 m and 0.3-0.6 m, in order to obtain the required database for the derivation of PTF, which includes the soil particle size distribution, carbonate minerals, bulk density, particles density, organic matter, saturated hydraulic conductivity, capillary height and the boundaries of available water. Measured characteristics were considered as explanatory variables (EV). Volumetric water content was measured at suctions 0, -5, -10, -33, -500, -100, -500, -1000, and -1500 kPa and considered as response variables (RV). PPTF was derived using SAS statistical software based on multiple linear regression. Three methods for fitting the PPTF to EV were Maximum R2 Improvement (MR), Backward Elimination (BW) and Forward selection (FW). The prediction accuracy was evaluated through Mean Bias Error (MBE), Mean Absolute Error (MAE), root mean squared error (RMSE) and Coefficient Of Determination (R2). The results showed that all EVs were included in the PPFT derivation by MR method, while in the BW and FW methods the significant level of entery (SLENTRY) of EV was 0.1. It was also found that variables w1, capillary height, available water, bulk density, and porosity were included in the derivation of most of the models using the BW and FW method. Besides no statistically significant differences were found among the values of MBE, MAE, RMSE, and R2 of the PPTF although the number of selected EV are different.

Imaging of the electrical activity in the root zone under limited-water-availability stress: a laboratory study for Vitis vinifera

Abstract. Understanding root signals and their consequences for the whole plant physiology is one of the keys to tackling the water-saving challenge in agriculture. The implementation of water-saving irrigation strategies, such as the partial root zone drying (PRD) method, is part of a comprehensive approach to enhance water use efficiency. To reach this goal tools are needed for the evaluation of the root's and soil water dynamics in time and space. In controlled laboratory conditions, using a rhizotron built for geoelectrical tomography imaging, we monitored the spatio-temporal changes in soil electrical resistivity (ER) for more than a month corresponding to eight alternating water inputs cycles. Electrical resistivity tomography (ERT) was complemented with electrical current imaging (ECI) using plant-stem-induced electrical stimulation. To estimate soil water content in the rhizotron during the experiment, we incorporated Archie's law as a constitutive model. We demonstrated that under mild water stress conditions, it is practically impossible to spatially distinguish the limited-water-availability effects using ECI. We evidenced that the current source density spatial distribution varied during the course of the experiment with the transpiration demand but without any significant relationship to the soil water content changes. On the other hand, ERT showed spatial patterns associated with irrigation and, to a lesser degree, to RWU (root water uptake) and hydraulic redistribution. The interpretation of the geoelectrical imaging with respect to root activity was strengthened and correlated with indirect observations of the plant transpiration using a weight monitoring lysimeter and direct observation of the plant leaf gas exchanges.

Estimation of soil water and organic matter content in medium and low yield fields of Ningxia Yellow River Irrigation area based on hyperspectral information.

Accurately obtaining soil water and organic matter content is of great significance for improving soil qua-lity in croplands with medium to low yield. We explored the estimation effect of fractional order differentiation (FOD) combined with different spectral indices on soil water and organic matter content in medium and low yield croplands of Ningxia Yellow River Irrigation Area. After root mean square transformation of field measured hyperspectral reflectance, we used 0-2 FOD (with a step length of 0.25) to construct difference index (DI), ratio index (RI), product index (PI), sum index (SI), generalized difference index (GDI), and nitrogen planar domain index (NPDI) and to select the optimal spectral index based on the correlation coefficients between six spectral indices with soil water and organic matter contents. We constructed a model for estimating soil water and organic matter content based on partial least squares regression (PLSR) and support vector machine (SVM). The results showed that the correlation between soil water and organic matter content and spectral information was effectively improved after FOD transformation compared with the original spectrum, with maximum increases of 0.1785 and 0.1713, respectively. The soil water content sensitive bands were mainly in the range of 400-630 and 1350-1940 nm, while the sensitive bands of organic matter content were mainly at 460-850, 1530-1910, and 2060-2310 nm. The accuracy of SVM model was significantly higher than that of PLSR, and the soil water content estimation model based on 1.75-order NPDI-SVM reached the highest precision, with a validation determination coefficient (Rp2) of 0.970, root mean square error (RMSE) of 1.615, and relative percent deviation (RPD) of 4.211. The organic matter content estimation model based on 0.5 order DI-SVM had the best performance, with Rp2, RMSE and RPD of 0.983, 0.701 and 5.307, respectively. Our results could provide data and technological support for soil water and nutrient monitoring, quality improvement, and graphics creating in similar area with medium to low yield fields.

Estimating soil–water characteristic curve (SWCC) using machine learning and soil micro-porosity analysis

Estimating soil–water characteristic curve (SWCC) using machine learning and soil micro-porosity analysis

Using microdialysis with a deuterium oxide tracer to estimate water exchange, water content and active surface area of the probe

Microdialysis is a useful tool for measuring in situ fluxes of soil compounds with minimal disturbance of soil structure and function. Fluxes of sampled compounds are commonly calculated per unit of membrane surface area, assuming that the entire membrane surface is capable of exchange – which is unlikely given varying soil moisture and the occlusion of membrane pores by the soil solid phase. We present a method to quantify the degree of connectivity of the microdialysis probe membrane to the surrounding soil by means of water exchange between a microdialysis perfusate and soil solution using deuterium (2H2O; equilibrated to DHO) as an internal standard. We applied the method to a range of probe membrane surface areas and soil moisture conditions to generate empirical models that estimate membrane surface area active in exchange. Our results suggest that even in a saturated sandy soil, active membrane surface areas reach only 40.3% of the probe surface area, perhaps due to occlusion by soil particles. However, when accounting for volumetric water content of the soil, active surface areas approached 80–90% of the area likely in contact with water, indicating that sampling efficiency of water-filled pores may still be high, particularly at slow flow rates. Furthermore, our method enables assessment of local soil water content around the probe. Models estimating soil water content were applied to field measurements of DHO exchange in three soil horizons (Organic, B1, B2) at two boreal sites, and in situ estimates were similar to those from conventional soil moisture methods when models were calibrated with the same soil type. We present DHO exchange as a powerful method for improving microdialysis flux interpretations in future studies, and for exploring small-scale water variability in relatively undisturbed soils.

Timely monitoring of soil water-salt dynamics within cropland by hybrid spectral unmixing and machine learning models

Soil salinization and water scarcity are main restrictive factors for irrigated agriculture development in arid regions. Knowing dynamics of soil water and salt content is an important antecedent in remediating salinized soils and optimizing irrigation management. Previous studies mostly used remote sensing technologies to individually monitor water or salt content dynamics in agricultural areas. Their ability to asses different levels of crop water and salt management has been less explored. Therefore, how to extract effective diagnostic features from remote sensing images derived spectral information is crucial for accurately estimating soil water and salt content. In this study, Linear spectral unmixing method (LSU) was used to obtain the contribution of soil water and salt to each band spectrum (abundance), and endmember spectra from Sentinel-2 images. Calculating spectral indices and selecting optimal spectal combination were individually based on soil water and salt endmember spectra. The estimation models were constructed using six machine learning algorithms: BP Neural Network (BPNN), Support Vector Regression (SVR), Partial Least Squares Regression (PLSR), Random Forest Regression (RFR), Gradient Boost Regression Tree (GBRT), and eXtreme Gradient Boosting tree (XGBoost). The results showed that the spectral indices calculated from endmember spectra were able to effectively characterize the response of crop spectral properties to soil water and salt, which circumvent spectral ambiguity induced by water-salt mixing. NDRE spectral index was a reliable indicator for estimating water and salt content, with determination coefficients (R2) being 0.55 and 0.57, respectively. Compared to other models, LSU-XGBoost model achieved the best performance. This model properly reflected the process of soil water-salt dynamics in farmland during crop growth period. This study provided new methods and ideas for soil water-salt estimation in dry irrigated agricultural areas, and provided decision support for governance of salinized land and optimal management of irrigation.

Estimating soil water content in a thorny forest ecosystem by time‐lapse electrical resistivity tomography (ERT) and HYDRUS 2D/3D simulations

AbstractDetermination of soil volumetric water content in forest ecosystems is particularly challenging due to deep rooting systems and unknown soil vertical and spatial heterogeneity. This research aims to test two undisturbed methods, electrical resistivity tomography (ERT) and HYDRUS 2D/3D, for 2D determination in a thorny forest ecosystem. The experiment consisted of infiltrating 10 L of water lasting 60 min. During infiltration, ERT measured apparent resistivity by time‐lapse measurements, and was measured with an FDR probe (EnviroSCAN) at 33, 63, 83, 97, and 163 cm depth close to the infiltration site. At the end of infiltration, a soil pit was dug, and 100 measurements of were performed with a TDR in a 10 × 10 cm regular grid. Archie law transformed soil resistivity (ERT) into using manual calibration, verified by an independent dataset. The 2D profile obtained by ERT was qualitatively compared with the HYDRUS 2D/3D one. HYDRUS 2D/3D was parametrized with calibrated parameters obtained with HYDRUS 1D using 106 days of obtained with EnviroSCAN. The results of HYDRUS 1D calibration and verification were satisfactory, with RMSE and Nash‐Sutcliffe coefficients ranging from 0.021 to 0.034 cm3 cm−3 and 0.11 to 0.77, respectively. The forward HYDRUS 2D/3D simulation disagrees with EnviroSCAN data for 33 cm depth. However, it follows the trend with near to zero variation of water content at 63 cm depth. Water content determination by ERT was satisfactory with RMSE for calibration and verification of 0.017 and 0.021 cm3 cm−3. HYDRUS 2D/3D and ERT comparisons were not equal, with a shallower wetting front by ERT and a deeper one for HYDRUS. Still, both wetting fronts agree with the wetting depth estimated by EnviroSCAN. We conclude that both methods are an alternative for determination in heterogeneous and deep soils of forest ecosystems.