Estimation of van Genuchten Equation Parameters in Laboratory and through Inverse Modeling with Hydrus-1D

Soil water retention curve (SWRC) becomes important because it guides when and how much to irrigate, optimizing the use of water; can be obtained in the field or laboratory, being commonly determined in the laboratory with porous plate apparatus, and the determination is compromised by issues such as time and labor. In this context, inverse modeling emerges, which allows to obtain a variable going from the effect to the cause, using Hydrus-1D. Hence, this study aims to obtain van Genuchten equation parameters through inverse modeling with Hydrus-1D and make the respective comparisons and inferences. Matric potential data were obtained over time in an instantaneous profile-type experiment. Six sets of three tensiometers each were installed surrounding the center of the experimental plot, at depths of 0.20, 0.35 and 0.50 m. Target depth was 0.35 m, where the roots of most crops are concentrated, and the other tensiometers were used to obtain the potential gradient. Matric potential data were used to feed Hydrus-1D and obtain the van Genuchten equation parameters. Laboratory curves were obtained using porous plate apparatus, with four replicates. It was concluded that, in general, the Hydrus-1D model estimates van Genuchten equation parameters and, consequently, the SWCC of an Argissolo more consistently with field conditions than those obtained in the laboratory; and, provided it is fed with field data, the Hydrus-1D simulates well the behavior of matric potential and moisture over time, reducing the time and labor in the procedures to obtain van Genuchten equation parameters in the laboratory.


Introduction
Knowledge on soil physical attributes, like hydraulic properties, is important for the agricultural sustainability, guiding strategies that lead to maximum crop yield (Imhoff et al., 2016).In this context, the soil water retention curve (SWRC), given by the relationship between water content and the matric potential with which water is retained in the soil-allows to monitor soil moisture and, therefore, define when and how much to irrigate (Lucas et al., 2011).
SWRC can be obtained through various methods, in the field and laboratory.However, it is usually determined under laboratory conditions using the porous plate apparatus, proposed by Richards and Fireman (1943), in which the water content retained in the sample under the applied pressure originates the curve (Menezes et al., 2018).SWRC shape is commonly described by an empirical equation and the model of van Genuchten (1980), with five parameters, is the most used for this purpose, because it fits to a wide variety of soils (Xiang-Wei et al., 2010).
Obtaining soil hydraulic parameters, such as SWRC, either in the field or laboratory, is often demanding, in both time and labor, which makes such determination unviable in some cases (Singh et al., 2010).In this context, inverse modeling emerges, which is nothing more than obtaining certain variable through the solution of an inverse mathematical problem.In other words, it is possible to mathematically obtaining unmeasurable parameters of a system from mensurable ones since they have a physical relationship (Hasanoğlu & Romanov, 2017).--------------     In experiment to verify differences in hydraulic attributes in SWCC and in laboratory, Basile et al. (2003) found that, for all studied cases, there was discrepancy between water contents at ϕ m = 0 obtained in the field and laboratory, and water retention values were always higher for laboratory curves in the interval between ϕ m = 0 and ϕ m = 1 m.For these authors, higher water contents at ϕ m = 0 for soil samples in cylinders must be attributed to the easy air displacement through the sample under laboratory condition.
According to Figure 6, the differences between the curves are not limited to the wettest portion; the driest portion is also visibly different, especially due to the trend of the inverse model, as previously stated, to overestimate moisture contents in comparison to the laboratory procedure.In this case, for the textural class loamy sand, in which residual moisture content must be low, Hydrus-1D was clearly not efficient to simulate soil moisture at lower matric potentials, probably because the input data were limited to the wet part of the soil.
Considering the results observed in the present study, it is worth pointing out the perception of researchers on field and laboratory methods to estimate soil hydraulic attributes.It is true that many papers in the past were dedicated to the discussion on the validity of soil hydraulic properties obtained in the laboratory for the inference on the hydrological behavior in the field and, as a result, less expensive and less time-consuming techniques have been researched (Basile et al., 2003).However, it is important to highlight that these protocols are not always guaranteed to reliably reproduce what occurs in the field, which explains the stimulus to other approaches-for instance, inverse modeling-in the attempt at better perception on the actual soil hydraulic attributes.

Conclusions
In general, the Hydrus-1D model estimates van Genuchten equation parameters and, consequently, the soil-water characteristic curve of an Argissolo more consistently with the field conditions than those obtained in the laboratory.
Hydrus-1D simulates well the behavior of matric potential and soil moisture over time, reducing the time and labor of the procedure to obtain van Genuchten equation parameters in the laboratory. Figure Figure 5 simulation performan was very l