Estimate of the soil water retention curve from the sorptivity and β parameter calculated from an upward infiltration experiment

The water retention curve (theta(h)), which defines the relationship between the volumetric water content (theta) and the matric potential (h), is of paramount importance to characterize the hydraulic behaviour of soils. Because current methods to estimate theta(h) are, in general, tedious and time consuming, alternative procedures to determine theta(h) are needed. Using an upward infiltration curve, the main objective of this work is to present a method to determine the parameters of the van Genuchten (1980) water retention curve (alpha and n) from the sorptivity (S) and the beta parameter defined in the 1D infiltration equation proposed by Haverkamp et al. (1994). The first specific objective is to present an equation, based on the Haverkamp et al. (1994) analysis, which allows describing an upward infiltration process. Secondary, assuming a known saturated hydraulic conductivity, K-s, calculated on a finite soil column by the Darcy's law, a numerical procedure to calculate S and beta by the inverse analysis of an exfiltration curve is presented. Finally, the alpha and n values are numerically calculated from K-s, S and beta. To accomplish the first specific objective, cumulative upward infiltration curves simulated with HYDRUS-1D for sand, loam, silt and clay soils were compared to those calculated with the proposed equation, after applying the corresponding beta and S calculated from the theoretical K-s, alpha and n. The same curves were used to: (i) study the influence of the exfiltration time on S and beta estimations, (ii) evaluate the limits of the inverse analysis, and (iii) validate the feasibility of the method to estimate alpha and n. Next, the theta(h) parameters estimated with the numerical method on experimental soils were compared to those obtained with pressure cells. The results showed that the upward infiltration curve could be correctly described by the modified Haverkamp et al. (1994) equation. While S was only affected by early-time exfiltration data, the beta parameter had a significant influence on the long-time exfiltration curve, which accuracy increased with time. The 1D infiltration model was only suitable for beta < 1.7 (sand, loam and silt). After omitting the clay soil, an excellent relationship (R-2 = 0.99, p < 0.005) was observed between the theoretical alpha and n values of the synthetic soils and those estimated from the inverse analysis. Consistent results, with a significant relationship (p < 0.001) between the n values estimated with the pressure cell and the upward infiltration analysis, were also obtained on the experimental soils. (C) 2016 Elsevier B.V. All rights reserved.