Preferential flow in macroporous swelling soil with internal catchment: model development and applications

A new model of infiltration in swelling and shrinking clay soil was developed. The model consisted of two flow domains: soil matrix, with flow modelled by means of Darcy equation, and macropores; the latter was divided in turn into two sub-domains: shrinkage cracks, with aperture dynamically depending on matrix water content, and permanent macropores, independent of matrix saturation. In the shrinkage cracks sub-domain, a kinematic wave equation was derived by considering laminar motion of thin water films, along two parallel nearly vertical walls; in the permanent macropores sub-domain, a kinematic wave equation was assumed, with parameters physically related with macropores shape and dimension. Exchange of water between macropore domains and matrix was introduced in form of sink terms in the macropores mass balance equations, and as source terms in the matrix continuity equation. Infiltration through macropore walls was modelled using a diffusivity function derived from aggregates sorptivity measurements. The internal catchment was included by considering at each layer a fraction of dead end permanent macropores. Water ponding at the bottom of dead end macropores, infiltrated into the corresponding matrix layer. The model was tested against the results of infiltration transients through a large undisturbed swelling and shrinking clay soil column. Outflow rate from column bottom surface was constantly measured, while water content profile was registered at regular time intervals by means of five TDR horizontal probes. In order to quantify model parameters, characterisation of soil matrix was carried out, providing hydraulic conductivity curve, water retention curve, shrinkage characteristic and aggregates diffusivity. All of the other model parameters, thanks to their clear physical meaning, were estimated from direct observation of soil structure, except macropores morphologic parameters, some of which, although measurable, were obtained by calibration due to lack of relevant experimental data. The proposed model proved to be able to adequately simulate measured breakthrough curves as well as temporal evolution of water content profiles along the sample, especially with internal catchment process included.