Hydraulic properties characterization of undisturbed soil cores from upward infiltration measurements

Abstract

Estimation of the α and n van Genuchten (1980) parameters and the saturated hydraulic conductivity, Ks, of undisturbed soil samples is critical for realistic water flow simulations in the vadose zone. Given the Latorre and Moret-Fernández (2019) method, which allows estimating Ks, α and n from the inverse analysis of a single upward infiltration, the objective of this work is to advance on the theoretical understanding of this method and to apply it on undisturbed soil cores. To this end, the influence of the saturated, θs, and residual, θr, water content and initial soil tension, hi, on the α, n and Ks optimization was first studied on synthetic soils. A procedure to simultaneously estimate θs, α, n and Ks, which consisted on optimizing α, n and Ks for a range of θs values, was proposed. To this end, α, n and Ks were optimized leaving each θs as a fixed value. The method was next applied to experimental curves measured from sieved and undisturbed soil cores of 50 mm high and diameter, sampled the last ones from three fields with different management: bare vs. under plant soil; agricultural soil under conservation vs. conventional tillage; and overburden soil from mine vs. topsoil formed by the original soil. Results showed that θs had an important effect on α, n and Ks optimization, where the minimum error within the selected θs interval corresponded to the theoretical θs. An important effect of hi on the optimization was also observed. However, this influence was omitted by using a hi value located in the θr region, where dθdh≈0. Under this assumption, θr only affected the θs optimization, which value varied with θr. A robust relationship (R2 = 0.99; p < 0.0001) was found between theoretical and optimized θs, Ks and α and n calculated for the synthetic soils. Thus, the method allowed estimating four of the five van Genuchten (1980) parameters. An absolute minimum was also observed when the method was applied on experimental infiltration curves. The θs measured gravimetrically on the undisturbed samples was, on average, 8.1% higher than the optimized one. Overall, the method allowed detecting significant differences (p < 0.05) in hydraulic properties between bare and under plant soils and among tillage treatments. The large variability found in the mine’s soils prevented to find significant differences within this scenario.