Comparison of different methods to estimate the soil sorptivity from an upward infiltration curve

Abstract

The soil sorptivity, S, which is defined as a measure of the capacity of a porous medium to absorb or desorb liquid by capillarity, is commonly estimated under laboratory conditions from upward infiltration measurements. The objective of this work is to compare different methods to estimate S from a single upward infiltration curve obtained from both theoretical and experimental soils. An additional analysis of the influence of synthetic infiltration noise on the estimation of S was also performed on the theoretical soils. Five different methods were compared: Short Time model for horizontal infiltration (ST), the Cumulative Linearization method (CL) and the Differentiated Linearization (DL) linear regressions models, Short-time (SIM) methods that use the simplified Haverkamp et al. (1994) model, and Complete-time (CIM) upward infiltration method that uses the quasi-analytical Haverkamp et al. (1994) function. Since finite soil columns were considered, the saturated hydraulic conductivity needed to estimate S with the Haverkamp et al. (1994) model was calculated from an overpressure step at the end of the water absorption process, using the Darcy's law. The methods were contrasted on four theoretical and six sieved experimental soils, ranging from sand to clay textures. Although all methods showed acceptable estimates of S on clean theoretical upward infiltration curves, the ST, SIM and CIM were the methods that gave significant (p<0.001) regression analysis on noisied infiltration curves, and only SIM and CIM presented a relative error<1%. From these results we can conclude that although acceptable approaches of S were obtained with the simplest ST method, the CIM procedure was the most accurate method to estimate S in both clean and noised theoretical and experimental upward infiltration curves.