Understanding and predicting deep percolation under surface irrigation

A lysimeter experiment was conducted in southeastern Australia to quantify the deep percolation response under irrigated pasture to different soil types, water table depths, and ponding times during surface irrigation. Deep percolation was governed by the final infiltration rate of the subsoil, the ponding time, the water table depth, and the amount of water stored in the rootzone between saturation and field capacity. These key variables were used to characterize both steady‐ and nonsteady‐state percolation in a conceptual model of deep percolation. The conceptual model was found to provide an effective representation of deep percolation for both the lysimeter and field‐scale water balance data. Steady‐state percolation during irrigation was the dominant process contributing to deep percolation on most of the studied soils. Nonsteady‐state percolation (redistribution) was very important for the sandiest soil type. The conceptual model provided better prediction of deep percolation than both data‐based model (artificial neural network) and process‐based modeling approach (1‐D Richards' equation model).