10:45 AM - 11:00 AM
[AGE28-06] Numerical analysis of unsaturated water flow in soils with generalized multimodal hydraulic property models using HYDRUS-1D
Keywords:hydraulic property model, generalized multimodal model, HYDRUS-1D
The analytical models of soil water retention curve, θ(h), and unsaturated hydraulic conductivity (HCF), K(h), are important for numerical analyses of unsaturated water flow. The water retention curve θ(h) (WRF) gives the relation between the soil water pressure head h and the volumetric water content, q while K(h) gives the relation between the unsaturated hydraulic conductivity and h. Many WRF models, which estimate θ(h) based on the unimodal pore size distribution, have been proposed (Brooks and Corey, 1964: van Genuchten, 1980: Kosugi, 1996). HCF can be then modeled using the Mualem model (Mualem, 1976) from the WRF.
While the unimodal models can be applied to various soil types, they are known to perform poorly with well-structured aggregated soils, such as Andisols, in which intra- and inter-aggregate pore systems are well structured, and with sandy soils as K(h) is extremely underestimated in the dry range. To overcome this shortcomings, modeling with a linear combination of unimodal models, referred to as multimodal model, was proposed (Priesack and Durner, 2006; Seki et al., 2022). While the multimodal models are expected to flexibly fit a variety of soils, their effect on unsaturated water flow is not fully understood as they have not yet been implemented into common unsaturated water flow programs.
The objectives of this study were, first, to implement generalized multimodal soil hydraulic models into a HYDRUS-1D (Šimůnek et al., 2008). Second, parameter estimation based on the evaporation method was performed using the modified HYDRUS-1D (Šimůnek et al., 1998) to assess the impact of using the multimodal model on numerical analysis of unsaturated water flow. As for the general inverse analysis, software named PEST (Doherty, 2015) was utilized.
In the modified HYDRUS-1D, up to four linearly combined Broods and Corey (BC) model, van Genuchten (VG) model, and Kosugi (KO) model can be used.
Parameters of the generalized multimodal model were estimated from the data of the evaporation methods used Kumamoto Andisol and Tottori dune sand (Sakai and Toride, 2007) using the modified HYDRUS and were compared with the unimodal model, VG. Note that two same models combining BC, VG or KO, respectively, were used here (BC-BC, VG-VG or KO-KO). For Andisol, the generalized multimodal model represented the staircase-like water retention curves better than VG. In dune sands, VG-VG and KO-KO, except BC-BC (fail in inverse analysis due to instability of HYDRUS simulation), overcame the extreme underestimation of K(h) in the dry range, which was a problem with VG, and also represented the rapid decrease in the pressure head with evaporation near the surface.
While the unimodal models can be applied to various soil types, they are known to perform poorly with well-structured aggregated soils, such as Andisols, in which intra- and inter-aggregate pore systems are well structured, and with sandy soils as K(h) is extremely underestimated in the dry range. To overcome this shortcomings, modeling with a linear combination of unimodal models, referred to as multimodal model, was proposed (Priesack and Durner, 2006; Seki et al., 2022). While the multimodal models are expected to flexibly fit a variety of soils, their effect on unsaturated water flow is not fully understood as they have not yet been implemented into common unsaturated water flow programs.
The objectives of this study were, first, to implement generalized multimodal soil hydraulic models into a HYDRUS-1D (Šimůnek et al., 2008). Second, parameter estimation based on the evaporation method was performed using the modified HYDRUS-1D (Šimůnek et al., 1998) to assess the impact of using the multimodal model on numerical analysis of unsaturated water flow. As for the general inverse analysis, software named PEST (Doherty, 2015) was utilized.
In the modified HYDRUS-1D, up to four linearly combined Broods and Corey (BC) model, van Genuchten (VG) model, and Kosugi (KO) model can be used.
Parameters of the generalized multimodal model were estimated from the data of the evaporation methods used Kumamoto Andisol and Tottori dune sand (Sakai and Toride, 2007) using the modified HYDRUS and were compared with the unimodal model, VG. Note that two same models combining BC, VG or KO, respectively, were used here (BC-BC, VG-VG or KO-KO). For Andisol, the generalized multimodal model represented the staircase-like water retention curves better than VG. In dune sands, VG-VG and KO-KO, except BC-BC (fail in inverse analysis due to instability of HYDRUS simulation), overcame the extreme underestimation of K(h) in the dry range, which was a problem with VG, and also represented the rapid decrease in the pressure head with evaporation near the surface.