Open access publications in the SLU publication database

Abstract

Preferential flow through macropores is one of the major pathways through which pesticides can leave the root zone and enter ground and surface waters. Dual-permeability models account for preferential flow by including a separate flow domain coupled to the less permeable soil matrix. The exchange of water and solutes between flow domains is usually described by first-order expressions. Key model parameters regulating the degree of preferential flow are difficult or impossible to derive from direct measurements. Therefore, the objective was to evaluate the possibilities of parameter identification through automatic calibration in the dual-permeability model MACRO. An improved version of the model, with implicit numerical solutions introduced to shorten model run times, was first developed. This enabled the use of computationally intensive calibration methods. Generalised likelihood uncertainty estimation (GLUE) and sequential uncertainty domain parameter fitting (SUFI) were applied to a comprehensive field data set containing measurements of soil water contents, drainflow and flux and resident concentrations of bentazone and bromide. All groups of data were needed to get highly conditioned and unbiased parameter estimates. However, the parameter describing mass exchange between pore domains was poorly conditioned even with a very comprehensive data set. SUFI decreased the initial uncertainty domains significantly for all parameters except for the parameter describing mass exchange between pore domains in the subsoil. However, random sampling from SUFI posterior uncertainty domains resulted in larger prediction uncertainty compared to GLUE. This is because these domains contain parameter combinations that are poor simulators, mainly due to parameter correlations. GLUE and a parameter identification method based on the localisation of information (PIMLI) were applied to real and numerically generated data from laboratory microlysimeter experiments. The data contained enough information to reduce the uncertainty of the parameter describing mass exchange between pore domains, the saturated matrix hydraulic conductivity and the macroporosity. Simultaneous identification of the macroporosity, the macropore hydraulic conductivity and the parameter describing macropore tortuosity was not possible, probably because of parameter correlations. Measurements with large information content for parameter identification were generally found during the first irrigations after solute application.