Application of simple soil-water flow models to the transfer of surface-applied solutes to streams

In the context of leaching of surface-applied nitrate fertilizer to runoff, this paper evaluates the performance of five simple predictive models, based upon: (1) field capacity; (2) uniform displacement; (3) wetting front; (4) by-passing (preferential) flow; and (5) soil-water domains. Predicted behaviour and observed on an 85-cm depth soil are compared for leaching depth and outflow response. Predictive success of outflow by the models for 40 monitored rainfall events were: (1) 60%; (2) 52%; and (3) 76%. For the by-passing flow model (4), simple prediction of pedal excess flow (4a) gave a success of 80%; where calculations of the operational volume in preferential flow were involved (4b), variations from 20 to 100% success occurred for a range of operational volumes, the lowest volumes giving the greatest success (4c). The soil-water domain model (5) is a simple, predictive scheme involving the classification of outflow events according to antecedent soil moisture (θ_a); the method uses the mobile: retained water content (θ_r) at 2 bars and field capacity (θ_fc). Using soil-water domains alone was of limited success but when the domains were combined with rainfall intensity, the scheme specified outflow exactly. Thus, for θ_a < θ_r, no soil-water outflow occurred, even at high rainfall intensity; for θ_a >θ_fc outflow always occurred, even at low rainfall intensity and for θ_a >θ_r < θ_fc the simple by-passing model (4a) achieved 100% predictive success. The implications for fertilizer application are discussed.