Uncertainty in Estimation of Dual Permeability Model Parameters for Flow and Tracer Transport in Laboratory Soil Columns.

A large number of parameters are required when dual permeability model (DPM) is used for simulating flow and transport in the fractured/macroporous unsaturated zone. Uncertainty in estimation of these additional hydraulic and solute transport parameters (for macropore and matrix/macropore interface) can be associated with measurement error, use of invoked conceptual model and/or the parameter optimization technique. Specific objectives of the study are: i) to quantify uncertainty in parameter estimation from several soil column experimental data, and ii) to study the association among various input and output parameters. To realize the objectives, various controlled flow and transport experiments were carried out with bromide solution (KBr) in: 1) a soil column with 1 central macropore (0.1 cm dia) and 2) two soil columns with multiple macropores (3 macropores of 0.1 cm dia; 19 macropores of 0.1 cm dia) in one-half of the column cross-section. For the second objective, Artificial Neural Network (ANN) was used on data generated from experimental observations and numerical model (Hydrus-1D) runs of the various columns. Results from sensitivity analysis highlight the variation in key parameters affecting physical non-equilibrium in different macroporous soil column designs. Saturated water content was found important for variably saturated flow experiments in homogeneous and central macropore columns, while matrix saturated hydraulic conductivity was a key parameter in bromide transport experiments in multiple macropore columns. Initial results from Hydrus-1D show positive correlation between outflow rates, bromide breakthroughs, pressure head profiles, and parameters, such as, the size and no. of macropores, density/area fraction of macropores, etc. ANN will be used to quantify these relationships.