Approaches for estimating unsaturated soil hydraulic conductivities at various bulk densities with the extended Mualem-van Genuchten model

Abstract

<p>The Mualem-van Genuchten model has been widely used for estimating unsaturated soil hydraulic conductivity (<em>K</em>u) from measured saturated hydraulic conductivity (<em>K</em>s) and fitted water retention curve (WRC) parameters. Soil bulk density (ρb) variations affect the accuracy of <em>K</em>u estimates. In this study, we extend the Mualem-van Genuchten model to account for the ρb effect with ρb-related WRC and <em>K</em>s models. We apply two functions (A and B) that relate the van Genuchten WRC model to ρb and two models (1 and 2) that estimate <em>K</em>s with various ρb. By combining the ρb-related WRC functions and <em>K</em>s models, we develop four integrated approaches (i.e., A1, A2, B1, and B2) for estimating <em>K</em>u at various ρb. <em>K</em>umeasurements made on five soils with various textures and ρb are used to evaluate the accuracy of the four approaches. The results show that all approaches produce reasonable <em>K</em>u estimates, with average root mean square errors (RMSEs) less than 0.35 (expressed in dimensionless unit because logarithmic <em>K</em>u values are used). Approach A2, with an average RMSE of 0.25, agrees better with <em>K</em>u measurements than does Approach A1 that has an average RMSE of 0.28. This is because Model 2 accounts for the WRC shape effect near saturation. Approaches A1 and A2 give more accurate <em>K</em>u estimates than do Approaches B1 and B2 which both have average RMSEs of 0.35, because Function A performs better in estimating WRCs than does Function B. The proposed approaches could be incorporated into simulation models for improved prediction of water, solute, and gas transport in soils.</p>