Effect of antecedent soil moisture on infiltration and preferential flow in texture contrast soils

Preferential flow has been shown to be both common and widespread in agricultural soils, however the processes and mechanisms responsible for preferential flow in texture contrast soils have not previously been investigated. Through a combination of dye tracer experiments, soil morphology, in situ soil m oisture monitoring, infiltration studies, and soil water modelling, the effect of antecedent soil moisture on the occurrence, type and depth of preferential flow was investigated in a series of texture contrast soil profiles. Preferential flow was dominated by hydrophobicity induced finger flow in the A1 horizon and bypass flow through shrinkage cracks in the subsoil. Differences between sites in horizon thickness, chemical properties, presence / absence of an A2 horizon, and abundance of sand infills resulted in little variation in preferential flow. At low antecedent soil moisture, dye tracer infiltrated to 85 - 119 cm depth, infiltration bypassed up to 99 % of the soil matrix, and wetting front velocities were estimated up to 12 000 mm hr-1. Perched water tables and subsurface lateral flow did not develop due to the abundance of subsoil shrinkage cracks. At high antecedent soil moisture dye tracer infiltrated to 24 ‚Äö- 40 cm depth. While water repellence had been overcome or leached from the A1 horizon, infiltration of new water was impeded by difficulty displacing existing soil water further down the soil profile. This resulted in wetting front instability and lateral flow through the A1 horizon rather than within the A2 horizon or along the upper surface of the B horizon as reported in the literature. Occurrence of preferential flow was not related to rainfall intensity or rainfall magnitude. Rather, preferential flow was significantly more likely to occur when antecedent soil moisture was below approximately 50 - 60 % PAWC. Water repellence had a profound effect on the development of preferential flow, however the relationship between water repellence and antecedent soil moisture was not straightforward. Potential water repellence varied seasonally in relation to rainfall history in which water repellence was not re-established after rainfall unless input of new hydrophobic substances occurred. Ability to model and predict the frequency and magnitude of preferential flow was limited by poor model performance, attributed to parameter uncertainty and inability to simulate water repellence and lateral flow. Results indicate that in agricultural landscapes which contain texture contrast soils, shallow groundwater and waterways may be at risk of contamination by preferential transport of agrochemicals at low antecedent soil moisture.