Multiobjective parameter estimation of hydraulic properties for a sandy soil in Oman

Abstract

The estimation of soil hydraulic parameters for the description of the highly non-linear hydraulic conductivity and water retention curve is of crucial importance for numerical modeling of water and solute transport in the vadose zone. Soil cores have been taken from different depths of a sandy soil of an agricultural field site located in the Batinah region of the Sultanate of Oman and subjected to a comprehensive laboratory analysis, comprising simplified multistep outflow experiments without the use of microtensiometers, additional retention measurements with the pressure plate apparatus and particle size analysis. The measured retention curve and the dynamic cumulated outflow time series—characterizing the hydraulic conductivity curve—are used to estimate the soil hydraulic parameters of the Mualem-van Genuchten (MvG) soil model. Multiobjective parameter optimization runs were performed by coupling the HYDRUS-1D model with the AMALGAM algorithm to allow an enhanced diagnostics of the soil and model behavior by analyzing the trade-offs between two objectives. Furthermore, two different formulations of the MvG model were investigated regarding their ability to reproduce the measurements: (1) P1—with fixed shape parameter \(\tau\) and (2) P2—considering the shape parameter \(\tau\) as an additional fitting parameter. The estimated parameters for the investigated soil profile exhibit a considerable degree of heterogeneity with saturated hydraulic conductivities varying by three orders of magnitude within the soil profile. The multiobjective analysis showed that considerable trade-offs exist between the two objectives for some of the soil cores, indicating that the fit to the cumulated outfow data can be increased on the cost of the fit to the water retention. However, satisfying fits to both objectives simultaneously were achieved for both formulations of the MvG model.