Premium
Hydraulic Conductivity Function from Water Flow Similarity in Idealized‐ and Natural‐Structure Pores
Author(s) -
Arya Lalit M.,
Heitman J.L.
Publication year - 2010
Publication title -
soil science society of america journal
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.836
H-Index - 168
eISSN - 1435-0661
pISSN - 0361-5995
DOI - 10.2136/sssaj2009.0204
Subject(s) - hydraulic conductivity , soil water , saturation (graph theory) , soil science , permeability (electromagnetism) , degree of saturation , water flow , materials science , mineralogy , chemistry , geology , mathematics , membrane , biochemistry , combinatorics
Our objective was to develop and evaluate a simple, primarily physically based hydraulic conductivity model for natural‐structure soils. The proposed model of the hydraulic conductivity function, K (θ), is based on particle‐size distribution (PSD), flow in idealized capillary tubes of uniform diameter, and measured saturated hydraulic conductivity ( K s ). We assume that hydraulic conductivity at a given saturation is actually saturated conductivity of pores that remain filled at that saturation, i.e., insignificant contribution from unfilled pores. The PSD data are used to generate pore‐size distribution using the Arya–Paris model, based on soil water characteristics. Pores are first treated as idealized, and flow for each domain is calculated using the Hagen–Poiseuille flow equation. We further assume that the ratio of flow in any pore domain to total flow at saturation for idealized pores applies equally to natural‐structure pores. Using this equality, the total flow at saturation, obtained from measured K s , is distributed among the natural pore domains. This approach is advantageous in that flow parameters remain constant for all soils. Using the model, we calculated K (θ) for 29 soils with a wide range of physical properties. Agreement between predicted and experimental data was excellent to good for 21 soils ( r 2 > 0.9 for 14 soils, >0.8 for 7 soils). Poorer agreement in the remaining soils was attributed to uncertainty in input and experimental K (θ) data. A 1:1 comparison of log‐transformed predicted and experimental data for the 29 soils (719 data pairs) showed significant scatter( r 2 = 0.735, RMSE = 1.04), which is consistent with similar comparisons in the literature.