A. Timmerman’s research while affiliated with KU Leuven and other places

In this paper, the applicability of Richards' equation for water flow and the convection-dispersion equation for solute transport is evaluated to model field-scale flow and transport under natural boundary conditions by using detailed experimental data and inverse optimization. The data consisted of depth-averaged time series of water content, pressure head and resident solute concentration data measured several times a day during 384 d. In a first approach, effective parameters are estimated using the time series for one depth and assuming a homogeneous soil profile. In a second approach, all time series were used simultaneously to estimate the parameters of a multi-layered soil profile. Water flow was described by the Richards' equation and solute transport either by the equilibrium convection-dispersion (CDE) or the physical non-equilibrium convection-dispersion (MIM) equation. To represent the dynamics of the water content and pressure head data, the multi-layered soil profile approach gave better results. Fitted soil hydraulic parameters were comparable with parameters obtained with other methods on the same soil. At larger depths, both the CDE- and MIM-models gave acceptable descriptions of the observed breakthrough data, although the MIM performed somewhat better in the tailing part. Both models underestimated significantly the fast breakthrough. To describe the breakthrough curves at the first depth, only the MIM with a mixing layer close to the soil surface gave acceptable results. Starting from an initial value problem with solutes homogeneously distributed over the mobile and immobile water phase was preferable compared to the incorporation of a small layer with only mobile water near the soil surface.

Temporal and spatial variability of water content in soil results from a complex interaction of different factors such as duration and frequency of rainfall, soil layering, vegetation, and topography. The objectives of this study were (i) to use a resistant median-polishing scheme to quantify the temporal variability of a depth and a horizontal location factor in an additive model, and (ii) to investigate the time stability of those two factors at a detailed temporal scale during different infiltration and redistributions cycles. Time series of water content were measured at 5 depths and 12 locations along a transect of 6 m using Time Domain Reflectometry (TDR). Measurements were repeated every 2-hours for 168 days under natural boundary conditions. At each time step, the mean water content of the soil profile, 5 depth factors and 12 location factors were estimated. The time series of these factors were qualitatively interpreted and related to the atmospheric and prevailing soil conditions. It was found that micro-heterogeneity plays an important role, even at this small plot-scale. The relative contributions of the factors were dependent on the antecedent soil moisture conditions. Also, the ratio of the deterministic variance, i.e., variance explained by the deterministic factors, of water content to the observed variance is variable in time.

To investigate relations between solute transport, soil properties, and experimental conditions, we summarize results from leaching experiments that we carried out in a range of soils, at different scales (column (0.3–1.0 m ID, 1.0 m length) and field plot scale), and using different leaching rates (0.5–30 cm d 1). The lateral mixing regime and longitudinal dispersion were derived from time series of tracer concentrations at several depths in the soil. Field-and column-scale transport were similar in loam and silt loam soils. The mixing regime was related to soil morphological features, such as vertical tongues, stratification, macropores, and a water-repellent layer. The dispersion increased in all soils more than linearly with increasing leaching rate, implying that the dispersivity is not an intrinsic soil characteristic. The change of dispersivity with leaching rate was linked to the unsaturated hydraulic conductivity using a multidomain conceptualization of the pore space.

La expansión de las actividades humanas tiene gran impacto en el medio ambiente, generando cambios que deben ser contrarrestados, como el cambio climático y la dispersión de contaminantes de la industria y la agricultura; a través de estrategias que permitan la conservación del suelo y de las reservas de agua. La implementación de estrategias con enfoque de sistemas, que abarque modelos de simulación matemáticos, permite una mejor toma de decisiones y una mejor comprensión de la complejidad y la interacción de los diferentes procesos que afectan el destino de los nutrientes, contaminantes y sustancias químicas en el medio ambiente - suelo- en los cultivos y en las zonas de movimiento de aguas. El sistema WAVE se presenta en el presente artículo como una herramienta matemática que describe el transporte y transformaciones de la materia y la energía en el suelo, los cultivos y el suelo no saturado, sin embargo no es compatible con los procesos de las aguas subterráneas, drenajes y ríos.

We propose a novel method to characterize the fluid-filled (usually air or water) space in images of porous media at the pore scale. First, an aperture map is created based on a skeleton process, to describe all local sizes in the pore space. Then the pore space is segmented in pores, defined as elementary objects that compose the pore space. Using this segmented image, a pore network is created, which is a graphic representation of the pore space that includes local sizes and direct information about connectivity at the pore scale. As an application of this method for pore space modelling, the equivalent hydraulic conductivity or permeability for a soil sample is computed.

Saturated hydraulic conductivity (Ksat) is one of the most important factors playing a role in many agronomic, engineering and environmental activities. Therefore, many field and laboratory methods have been developed over time to estimate Ksat. Unfortunately, these methods often result in very dissimilar Ksat values since the parameter is very sensitive to sample size, flow geometry, sample collection procedures and various soil physical-hydrological characteristics. At 23 locations on a macroporous sandy loam hillslope, tension disc infiltrometer experiments with vertically installed TDR probes and tensiometers were carried out. Ksat was estimated using the stationary flux and the classical Logsdon and Jaynes analysis. A numerical inversion method implemented in the Hydrus-2D model resulted in estimates of Ksat as well as estimates of the Van Genuchten soil hydraulic parameters. At the same locations single ring pressure infiltrometer measurements were carried out and Ksat was estimated in two different ways. After the single ring pressure infiltrometer experiment, undisturbed soil cores were taken inside the ring and Ksat was estimated in the laboratory using a constant head method. Hence, Ksat estimates from three different measurement techniques, on a similar though not identical scale, can be compared. Geometric means of the Ksat estimates of all measurements techniques show the same order of magnitude. Only the numerical inversion method in Hydrus-2D resulted in a geometric mean of Ksat that was approximately 10 times lower. Low correlations between Ksat estimates of all methods are observed except for the high correlation found between the Ksat estimates from the soil core and the pressure ring infiltrometer measurements. The fact that both methods use ponding as a boundary condition and are predominantly one-dimensional might explain this high correlation. Ksat estimates between all methods were found to be significantly different except for the estimates from the soil cores which were not significantly different from any other method. Although all measurement techniques act on a similar scale there is lack of agreement between Ksat estimates. This indicates that local Ksat estimates depend not only on soil texture and structure but also on factors such as flow geometry and soil disturbance imposed by the measurement technique.

The fate of nitrogen in the soil is of major concern because of the potential hazard for nitrogen, applied in excess of the natural decomposing capacity of the soil, to contaminate shallow and deep aquifers. For the prediction of the nitrogen behavior in soils simulation models are frequently used. In this study the transport and fate of nitrate within the soil profile was analyzed by comparing historic field data with the simulation results of two mathematical models, i.e., the water and agrochemicals in the soil crop and Vadose environment (WAVE) and DRAINMOD-N. After calibration and validation of both models, they were used to simulate the nitrogen transport and transformation of the Hooibeekhoeve experiment, situated in the sandy region of the Kempen, Belgium, for a 30-year (1969–1998) period. In the analysis a continuous cropping with maize was assumed. Comparison between experimental measured and simulated state variables indicate that the nitrate concentrations in the soil and nitrate leaching to drains are controlled by the fertilizer practice, the initial conditions and the rainfall depth, and distribution. Furthermore, the study reveals that the models used give a fair description of the nitrogen dynamics in the profile root zone at field scale. It was concluded that the calibrated models are useful tools to optimize the nitrogen application strategy resulting in an acceptable level of nitrate leaching for a long period as a function of the combination “climate–crop–soil–bottom boundary condition.

The geometry of pore space is one of the key features in understanding transport of water and solutes. For soils where macropores are present over the whole length of the soil profile and simulations at the pore scale are used to determine the macroscopic soil hydraulic properties the quantitative characterization of the soil structure and pore network is extremely important. The application of microfocus X-ray computed tomography (tCT) provides a possi- bility for a non-destructive, three-dimensional morphological characterization of soil structure and the pore network at a microscale level. Two heterogeneous sandy loam soil cores (100 cm3) are scanned using microfocus X-ray computed tomography with a resolution of approximately 100 tm. The samples are chosen on the basis of their different macroscopic hydraulic properties that are determined in an inverse optimiza- tion procedure using multistep outflow and direct hydraulic measurements data. The complete water-filled pore space is derived from tCT images of saturated soil sam- ples. Scanning at two different energies (dual energy approach) gives information on the water-air distribution within the pore network at different successive soil pres- sure head conditions. This information can explain drainage or infiltration behavior of the pore network. Moreover 3D reconstructed images is converted to different bi- nary images by choosing appropriate threshold values to obtain pore networks with different degrees of connectivity. Several other geometrical properties such as pore volume, pore area of the derived pore network are then calculated. Furthermore, pore connectivity parameters such as the Euler Poincaré characteristic are determined. A pore size distribution, which considers the hydraulic diameter of the pores, is based on the techniques of the mathematical morphology, namely erosion and dilatation. This study discusses the derivation of true quantitative images using tCT, which is essential for obtaining good 3D quantitative soil structural information. The obtained results of the characterization of soil structure are compared to the outcome of the multistep outflow experiments.