Vadose Zone Hydrology Research Articles

Machine‐Learning Applications in Vadose Zone Hydrology

Machine‐Learning Applications in Vadose Zone Hydrology

Machine learning applications in vadose zone hydrology: A review

AbstractMachine learning (ML) has been broadly applied for vadose zone applications in recent years. This article provides a comprehensive review of such developments. ML applications for variables corresponding to different complex vadose zone processes are summarized mostly in a prediction context. By analyzing and assessing these applications, we discovered extensive usages of classic ML models with relatively limited applications of deep learning (DL) approaches in general. We also recognized a lack of benchmark datasets for soil property research as well as limited integration of physics‐based vadose zone principles into the ML approaches. To facilitate this interdisciplinary research of ML in vadose zone characterization and processes, a paradigm of knowledge‐guided machine learning is suggested along with other data‐driven and ML model‐based research suggestions to advance future research.

Determination of a pedotransfer function for specific air–water interfacial area in sandy soils: A pore network‐informed multigene genetic programming approach

AbstractUnderstanding specific air–water interfacial area (SAWIA) is essential for characterizing and modeling various phenomena in vadose zone hydrology, such as virus and colloid transport, contaminant dissolution, evaporation, and the hydro‐mechanical behavior of unsaturated soils. Traditional measurement methods, including X‐ray imaging and tracer techniques, often encounter challenges, leading to a scarcity of studies that provide a reliable relationship for SAWIA. Currently, no pedotransfer function in the literature links SAWIA with saturation and suction using readily measurable soil properties such as median grain size and porosity. In this study, we initially developed a pore network model capable of predicting SAWIA by calibrating it with corresponding soil‐water retention curves (SWRCs). We then used these models to compile a comprehensive database of SAWIA for six sandy soils, for which experimental SWRCs were available, covering a range of median grain sizes and porosities. Utilizing this database, we established a pedotransfer function through multigene genetic programming. The accuracy of this function was validated against experimental data not previously used in its training and testing. Our parametric study indicated that increases in either porosity or median particle size led to a decrease in the regions exhibiting higher SAWIA in terms of saturation and suction.

Effects of dynamic capillarity on the shear strength of sandy soils during transient two‐phase flow: Insights from non‐equilibrium triaxial simulations

AbstractModeling two‐phase flow in unsaturated porous media is not only important to vadose zone hydrology but also of great value in diverse disciplines. Common approaches use a simplified relationship between fluid pressure difference and saturation, neglecting the influence of saturation change rates. However, many studies have suggested that the applicability of this approach is limited to situations where the rate of change in saturation is insignificant. Despite several studies highlighting the importance of non‐equilibrium capillarity effects in unsaturated flow modeling, its significance in the mechanical response of the porous medium remains unclear. This study thus aims to address this gap by comparing the simulation results of the traditional static approach and an advanced approach that incorporates dynamic capillarity effects. The comparison is conducted under various flow boundary conditions to assess the magnitude of the differences between the two approaches. The results indicate that as the hydraulic boundary conditions’ absolute values increase, the contrast between the mechanical response of the two simulation scenarios (dynamic and static) becomes more significant. For instance, the dynamic model can predict shear strengths up to 50% higher than the static model. This highlights the importance of considering non‐equilibrium effects while modeling the mechanical behavior of an unsaturated porous medium. Finally, the parametric study of the effect of dynamic coefficient, air entry value, and saturated conductivity reveals the more pronounced effect of the dynamic coefficient on the mechanical response.

Inverse modeling of frequency domain-based one-dimensional soil water flow in layered soils

Soils are heterogeneous at multiple scales. Estimates of spatially varying saturated hydraulic conductivity (Ks) of soils dictate the resolution of modeling water flux based on the Richardson-Richards equation (RRE), but direct measurements of Ks everywhere in a field are practically impossible. As a result, this study develops an inverse approach to estimate spatially varying Ks utilizing soil moisture observations driven by periodic weather variability at different frequencies. Using a numerical solution to vertical, one-dimensional RRE based on Gardner-Kozeny (GK) model in a heterogeneous soil with periodic flux, we investigate the spatial cross-correlation between Ks and observations (the amplitude or phase shift) of soil moisture fluctuation at different frequencies without high computational cost in the conventional time-domain model. These correlation patterns vary with the spatial location and frequency, suggesting that soil moisture fluctuations at different frequencies at one monitoring location carry non-redundant information about the heterogeneous distribution of Ks. Guided by cross-correlation maps, several inverse experiments reveal the effectiveness of multi-frequency data of soil moisture fluctuation for mapping the heterogeneous distribution of Ks. Monte Carlo simulations suggest that multi-frequency data can provide non-redundant and useful information about the estimation of heterogeneous Ks. Synthetic data based on a commonly-applied nonlinear van Genuchten-Mualem (VG) model is then tested in our inversion algorithm. The results indicate that the spatial pattern of the inverted GK Ks is very similar to that of the VG Ks, although these two parameters are inherently not identical since they are used in two different models. In addition, the inverted GK Ks can successfully predict the soil moisture fluctuation of frequencies that were not used in the inversion. Despite having no field experiments to validate this model, we believe this is the first attempt to conduct frequency-domain inverse modeling in vadose zone hydrology using soil moisture fluctuation.

Plant growth‐promoting rhizobacteria mediate soil hydro‐physical properties: An investigation with Bacillus subtilis and its mutants

AbstractPlant growth‐promoting rhizobacteria and other soil bacteria have the potential to improve soil hydro‐physical properties and processes through the production of extracellular polymeric substances (EPS). However, the mechanisms by which EPS mediates changes in soil properties and processes remain incompletely understood, partly due to variations in EPS composition produced under different environmental conditions. In this study, we investigated the influence of different bacterial traits on intrinsic soil properties and processes of evaporation and infiltration using sand treated with the wild‐type Bacillus subtilis variant (UD1022) and its two mutant variants, – and srf. The – mutant suppresses EPS production through alterations in the eps and tasA genes, while the srf mutant lacks the gene for surfactin production. Experimental results confirmed that the solution viscosity of the – mutant was the lowest and the solution surface tension of the srf mutant was the highest among the three tested bacteria strains. The distinct intrinsic properties of EPS produced by these bacterial strains resulted in varied hydro‐physical responses in the treated sand. Key influences included modifications in wettability, hydraulic decoupling (or mixed wettability), and aggregation, which collectively led to reduced evaporation rates and heterogeneous water distribution during infiltration in the bacteria‐treated sands. Our findings advance the understanding of the role bacterial EPS play in vadose zone hydrology and offer insights for the development of sustainable strategies for increasing water retention, supporting crop production in arid regions, and facilitating land restoration.

Rock fragments influence the water retention and hydraulic conductivity of soils

AbstractRock fragments (RFs) influence soil hydraulic properties (SHPs), and knowledge about the SHPs of stony soils is important in vadose zone hydrology. However, experimental evidence on effective SHPs of stony soils is still scarce and mostly restricted to water‐saturated conditions and low volumetric contents of RFs. We examined the influence of RFs on SHPs through a series of measurements. Stony soils were prepared by packing 250‐cm3 cylinders with soils of two textures (sandy loam and silt loam) and with different volumes of RFs (up to 50% v/v) with a diameter of 8–16 mm. Samples were prepared in a way that the background soils (diameter smaller than 2 mm) had identical bulk density. The simplified evaporation method was used to determine the effective SHPs of stony soils. We used the obtained SHP data to evaluate the performance of models, which predict the effective SHPs of stony soils from SHPs of the background soil. The results highlight the systematic dependency of SHPs on volumetric content of RFs. The difference between modeled and measured SHPs was substantial for cases in which the soil contained a high amount of RFs. Accounting for the moisture content of RFs improved the prediction of the effective water retention curve of stony soils compared with a simple scaling that used only the content of RFs. Among the evaluated models for the effective hydraulic conductivity, the model based on the general effective medium theory showed the best performance, particularly for low RF contents.

Western U.S. multistate research project on “water movement in soils”: A retrospective

AbstractTo prioritize agricultural research in the United States and to improve its competitiveness globally, Multistate Research Funds (MRFs) were set aside under the Research and Marketing Act of 1946. To implement the act, Agricultural Experiment Station (AES) Directors in the western United States met regularly to evaluate, fund, and review multistate research projects (MRPs), with membership of AES scientists named by the Technical Committee. This article highlights the history of research collaboration in the soil and vadose zone scientific community that was initiated in the western United States. The scientific interactions that started in 1958 with 10 scientists in western land‐grant universities and the USDA to address “Water Movement in Soil” have grown in membership and scope through successive 5‐yr projects. We highlight the value of such collaboration and the scientific advances in soil science and vadose zone hydrology.

Can Microplastic Pollution Change Soil-Water Dynamics? Results from Controlled Laboratory Experiments

Current knowledge about the potential impacts of microplastics (MP) on vadose zone hydrology is scarce. The primary goal of this study was to address some of the limitations of previous research by developing more reliable and conclusive statistical evidence to better understand whether MP pollution can potentially cause hydrological impacts. We examined the effects of MP shape (type), as well as the magnitude of pollution (MP/soil mass ratio, λ) on water holding capacity (WHC) and bare soil water evaporation (ER) of fine sand, under controlled laboratory conditions. Three different shapes (types) of MP—fiber (polyacrylic), strand (polymethyl methacrylate), and pellet (acrylonitrile butadiene styrene), with six environmentally relevant MP concentration levels (MP/soil mass ratio), all ≤1.5%, were studied. Statistical regressions and non-parametric analyses of variance (i.e., Kruskal–Wallis analysis) indicate that MP pollution has a substantial potential to change WHC and late-stage evaporation, even at relatively low MP concentrations, but has minimal impacts on early stage evaporation of the studied fine sand. The magnitude of the impacts depends on individual MP shape (type) and concentration, connoting those MP impact mechanisms are complex. These findings suggest that the global issue of growing soil–MP pollution should be regarded as a concerning environmental and water resources stressor that could potentially cause widespread environmental change by altering soil-water dynamics at the watershed scale.

A Thermodynamic Formulation of Water Potential in Soil

AbstractIn the mid‐20th century, harnessing of thermodynamics in describing water movement in soil, viz the concept of water potential, marks the emergence of modern soil physics, unsaturated soil mechanics, and vadose zone hydrology. Yet to date, a seamless linkage between thermodynamics and water potential is still missing, leading to several long‐lasting dilemmas regarding soil properties, for example, abnormal soil water density, peculiar film water viscosity and relative permittivity, pore water pressure, and water freezing temperature depression. Here, a thermodynamic framework is established by synthesizing recent advancements in soil‐water interaction. The classical thermodynamic concepts are revisited, highlighting the difference between macroscopic systems commonly treated in conventional theories and the intermolecular scale system subject to external fields. Soil water is conceived as an intermolecular scale open thermodynamic system subject to external fields of gravity, osmosis, and adsorption. The formulated thermodynamic framework is verified by reducing to the conventional definition of matric potential and a recently proposed unitary definition. The accuracy of the framework is further justified in terms of mechanical equilibrium criteria. The framework predicts the existence of spatially varied pore water pressure in soil pores and can serve as the theoretical basis for reconciling the physical origin of abnormal soil behavior such as water density, film water viscosity, relative permittivity, and negative or positive pore water pressure.

Machine learning in vadose zone hydrology: A flashback

AbstractArtificial intelligence (AI) and machine learning (ML) have been recently applied extensively in various disciplines of vadose zone hydrology. However, not much attention has been paid to their database‐dependent accuracy and uncertainty, reproducibility, and delivery, which undermines their applications to real‐world problems. We discuss lessons from the past and emphasize the need for and lack of fundamental protocols (i.e., detailed clarification on data processing, ML models accessibility, and a clear path for reproducing results).

Near‐surface controls on peatland hydrology: Implications for rapid adaptation and enhanced resilience to disturbances

AbstractNorthern peatlands faced compounding disturbances that transformed such critical ecosystems from long‐term carbon sinks into carbon sources. Considerable investment is therefore directed for restoring their carbon sequestration potential through large‐scale rewetting/rehabilitation. However, rapid need to transform their carbon dynamics contrasts with millennial timescales over which peat profiles that control key ecohydrological processes within these ecosystems have developed. This study demonstrates the sensitivity of vadose zone hydrology of northern peatlands to hydrophysical properties of the very near‐surface peat layer and therefore the potential capability of at least some ecohydrological processes to respond rapidly to developments in peat properties as a result of restoration. HYDRUS 1‐D Monte Carlo simulations were undertaken of near‐surface peat layers of various species and depths overlying degraded peat layer during periods of sustained drying. The modelling results showed that shallow additions of newly developed Sphagnum peat, just a couple of centimetres in depth, substantially modified near‐surface hydrology of peat profiles and significantly altered the time taken for reaching important ecohydrological pressure heads. Whilst a degraded peat layer reached threshold pressure head (TP) of −100 cm in 119 h, addition of 2.5‐cm layer of S. magellancium reduced the average time to TP by 18 hours, whilst S. fuscum and amalgamated Sphagnum overlying degraded peat across initial WTDs (5, 10, 15 and 20 cm) increased average time to TP by 304 and 540 h, respectively. This demonstrates that whilst peat hydrophysical properties have developed over millennia, ecohydrological dynamics of these systems rapidly adjusted through restoration approaches in response to disturbances.

Review of Rain or Shine: An Introduction to Soil Physical Properties and Processes. Tyson Ochsner. Published by Oklahoma State University Libraries under the Creative Commons Attribution 4.0 International License. Version 6. Last updated 19 Jan. 2022. DOI: https://doi.org/10.17605/OSF.IO/Z4RBT

Review of <i>Rain or Shine: An Introduction to Soil Physical Properties and Processes</i>. Tyson Ochsner. Published by Oklahoma State University Libraries under the Creative Commons Attribution 4.0 International License. Version 6. Last updated 19 Jan. 2022. DOI: https://doi.org/10.17605/OSF.IO/Z4RBT

Modeling seasonal soil moisture dynamics in gley soils in relation to groundwater table oscillations in eastern Croatia

This study aims to explain complex vadose zone hydrology of fine-textured (gley) agricultural soils influenced by a shallow and dynamic groundwater (GW) levels. The field site was located in the Biđ field (Eastern Croatia), where a detailed soil survey was performed. The simulations included a three-year period (2016–2018) at four locations. Soil hydraulic parameters (SHP) were estimated based on variables determined in the laboratory, while soil water flow was monitored using in-field zero-tension lysimeters. Piezometers were installed and used to monitor daily oscillations of groundwater levels (average depth to GW 2.2 m), while data from nearby Sava River was monitored. Unsaturated flow and water regime assessment was performed using HYDRUS-1D numerical modeling. Additional SHP optimization of van Genuchten-Mualem parameters (α and n) was performed using Shuffled Complex Evolution algorithm (SCE). The autocorrelation analysis was used to detect patterns in the precipitation, GW, and river level time series, while the Mutual Information (MI) was used to estimate the codependence of the processes in unsaturated zone and the main hydrological events. The model successfully (R2 0.72 – 0.94) reproduced measured lysimeters outflows. The outflows from lysimeters were connected to precipitation patterns, transpiration intensity, and soil moisture content influenced by the shallow water table. Comparable MI values obtained for precipitation, GW, and river level suggest a concurrent role of these parameters in the unsaturated flow dynamics. The relationship between upward flux/water storage change into the domain, and transpiration/growth stages, suggests a strong connection between the water fluxes and the root water uptake. Results confirm the importance of GW for the agricultural production due to the major influence on upper soil layer moisture.

Hierarchical prediction of soil water content time series

Quantifying the spatial and temporal dynamics of soil moisture is an important subject in vadose zone hydrology. A 1282-day field study was conducted to provide a hierarchy of data to assess neural network simulations of field soil water content time series θ(t). Volumetric water content was determined, typically once a week, by neutron thermalization at 20-, 40-, 60-, 80- and 100-cm depths. Soil samples were taken at 60 locations between 14- and 114-cm depths to determine soil properties, water retention, and saturated hydraulic conductivity. Prediction of hydraulic parameters from basic and extended soil properties yielded low correlation coefficients. Water content could be predicted reasonably with neural networks from soil properties or hydraulic parameters (0.880 < R < 0.942). Prediction of θ(t) based solely on rainfall data was not accurate. Independent networks could accurately simulate water content from one observed θ(t) based on the Nash-Sutcliffe Model Efficiency and Percent Bias. Once a network is trained for a particular depth, accurate predictions can be made beyond the training period using observations at just one location.

&amp;lt;i&amp;gt;STH-net:&amp;lt;/i&amp;gt; a soil monitoring network for process-based hydrological modelling from the pedon to the hillslope scale

Abstract. The Schäfertal Hillslope site is part of the TERENO Harz/Central German Lowland Observatory, and its soil water dynamics are being monitored intensively as part of an integrated, long-term, multi-scale, and multi-temporal research framework linking hydrological, pedological, atmospheric, and biodiversity-related research to investigate the influences of climate and land use change on the terrestrial system. Here, a new soil monitoring network, indicated as STH-net, has been recently implemented to provide high-resolution data about the most relevant hydrological variables and local soil properties. The monitoring network is spatially optimized, based on previous knowledge from soil mapping and soil moisture monitoring, in order to capture the spatial variability in soil properties and soil water dynamics along a catena across the site as well as in depth. The STH-net comprises eight stations instrumented with time-domain reflectometry (TDR) probes, soil temperature probes, and monitoring wells. Furthermore, a weather station provides data about the meteorological variables. A detailed soil characterization exists for locations where the TDR probes are installed. All data have been measured at a 10 min interval since 1 January 2019. The STH-net is intended to provide scientists with data needed for developing and testing modelling approaches in the context of vadose-zone hydrology at spatial scales ranging from the pedon to the hillslope. The data are available from the EUDAT portal (https://doi.org/10.23728/b2share.82818db7be054f5eb921d386a0bcaa74, Martini et al., 2020).

Predicting water flow in fully and partially saturated porous media: a new fractal-based permeability model

Predicting the permeability of porous media in saturated and partially saturated conditions is of crucial importance in many geo-engineering areas, from water resources to vadose zone hydrology or contaminant transport predictions. Many models have been proposed in the literature to estimate the permeability from properties of the porous media such as porosity, grain size or pore size. In this study, we develop a model of the permeability for porous media saturated by one or two fluid phases with all physically-based parameters using a fractal upscaling technique. The model is related to microstructural properties of porous media such as fractal dimension for pore space, fractal dimension for tortuosity, porosity, maximum radius, ratio of minimum pore radius and maximum pore radius, water saturation and irreducible water saturation. The model is favorably compared to existing and widely used models from the literature. Then, comparison with published experimental data for both unconsolidated and consolidated samples, we show that the proposed model estimate the permeability from the medium properties very well.

Eklenik Sapma Yaklaşımı ile Türetilen Yağış-Seviye İlişkisinden Yeraltısuyu Beslenimi Tahmin Modeli: Gazipaşa (Antalya) Ovası Akiferi Örneği

Groundwater recharge is one of the basic parameters required in water resources potential calculations. There are several methods applied to estimate groundwater recharge, however, every method has its own advantages and disadvantages, due to the fact that the theory of infiltration and the vadose zone hydrology is highly complex. Search for a method based on less complex theory, which requires readily available and representative parameters, and gives reliable results is still of interest. Groundwater recharge estimation based on the assumption that the groundwater level in the saturated zone rises as a result of the cumulative rainfall may well meet these expectations. It requires less data, easy to calculate and gives relatively reliable results. In this paper, following an explanation of the theoretical basis, the method is applied to Antalya-Gazipaşa plain aquifer. The analyses have revealed that the method can produce the deviations in observed groundwater based on the cumulative rainfall departure, with high coefficients of determination. The method was also used to analyse the external effects, such as percolation from river beds and groundwater pumping, on the deviations of groundwater levels from the average, which suggests that the plain aquifer constitutes an open system.

The Recharge Process and Influencing Meteorological Parameters Indicated by Cave Pool Hydrology in the Bare Karst Mountainous Area

Understanding the recharge and runoff processes of the vadose zone is significant for water resource management and utilization in karst mountain areas. Hydrological modeling of the vadose zone in karst caves has provided new methods of evaluating water resources in vadose zones. This paper provides modeling of vadose zone hydrology in a subtropical karst cave. The monitoring was conducted in Yuanyang Cave, Fengshan County, Guangxi Province, Southwest China. By monitoring the water level of a pool recharged by drop water in a cave, a model was established to calculate the natural leakage from the bottom and the infiltrated recharge from the vadose zone above. Combined with meteorological data records, the occurrence of recharge events in the vadose zone was analyzed. The correlation between them was established by multiple linear regression. The results showed that the infiltration ratio of precipitation was 20.88%. Recent rainfall of 4–7 days had shown a greater impact on recharge events than that of 3 days. The effect of evaporation was significant. The regression model in the cave pool was used to understand the hydrological process of the vadose zone, which provided a useful method for water resource management and evaluation in the remote karst mountain area.

Analytical Solution for Field Soil Water Content Profiles

AbstractQuantifying the spatial and temporal dynamics of soil moisture is an important subject in vadose zone hydrology. Although progress has been made to measure soil water content, soil water flow needs to be simulated for many applications. An analytical solution for soil water content, resulting from vertical flow and simplified mathematical conditions and hydraulic properties, was derived with the Cole‐Hopf transformation. The solution serves as a transfer function with the calibrated boundary condition and water content as respective input and output. This transfer function was applied to time series of soil water content, obtained from weekly observations with a neutron probe over a 1282‐day period at five depths and 16 locations. Water retention and saturated hydraulic conductivity were measured on soil samples in the laboratory. The transfer function was calibrated to observed time series of soil water content θ(t) with a median coefficient of determination R2 = 0.980. It was then used to predict θ(t) using constants from the calibration location and hydraulic parameters from the prediction location. There was a fairly good correspondence between predicted and observed θ(t) using calibrations elsewhere (R2 = 0.783). As an alternative, an observed time series θ(t) was scaled by multiplying it by the ratio of experimental saturated water contents for prediction and observation location. This scaling approach led to poorer predictions (R2 = 0.651). The analytical solution provides a flexible approach to quantify soil moisture for a variety of scenarios regarding available data on topography, climate, soil properties, and soil hydraulic parameters.