Hillslope and stream connectivity: simulation of concentration-discharge patterns using the HYDRUS model

Nutrient concentrations and loads in streamflow are sensitive to rapidly changing stream chemistry and discharge during storms. Mechanistic models that can simulate water and solute movement at hillslope scales could be useful for predicting concentration-discharge (C-Q) patterns and thereby improve our quantitative understanding of terrestrial-aquatic linkages for targeted catchment management. Our objective was to use the HYDRUS model to represent hydro-biogeochemical processes in soils that drive seepage of water and solutes from soil profiles into streams. Specifically we compared measurements in the literature with HYDRUS outputs using two methods for simulating runoff. This model predicts runoff (R) as rainfall that is instantaneously in excess of infiltration, but it is not designed to route runoff as overland flow. Post-HYDRUS addition of seepage to runoff was used to simulate the delivery of dissolved or particulate constituents to a stream (method A). Alternatively, we demonstrated how simulations using HYDRUS could include a hypothetical layer at the top of the soil profile with extremely high porosity and hydraulic conductivity that enabled overland flow and down-slope infiltration, but in this case only dissolved constituents could be considered (method B). These methods were evaluated by comparing the simulated temporal patterns of discharge and concentration with observed patterns. The catchments considered were in Slovenia (4210 ha) and in Australia (11.9 ha). Methods A and B were shown to adequately simulate some aspects of published discharge-concentration patterns, e.g. runoff dilution or concentration effects, but the temporal patterns of discharge for both methods did not precisely match those measured at small time-steps (e.g. 15 minutes). This limitation was due mainly to inadequate simulation of the down-slope movement of runoff and down-slope infiltration of a portion of this runoff. Method A was generally more useful than method B. Despite this limitation, both methods, if used carefully, should be adequate for many purposes, especially when simulating longer time-steps. Additional hypothetical simulations illustrated the significance of soil hydraulic conductivity, soil water content, and vertical gradients in solute concentrations in soil. Two temporal types of dischargeconcentration patterns were observed; short-term hysteresis caused by runoff during and shortly after a rainfall event, and longer-term trends associated with infiltration and seepage. Clockwise and anti-clockwise hysteresis was demonstrated to be potentially due to the temporal asynchrony of peak discharge and peak concentration in runoff. Simulations also demonstrated advantages over using the more common approach of a 2or 3-component mixing model. Our results suggest that the HYDRUS model will be useful for the mechanistic simulation of within-soil processes that are needed to predict discharge-concentration patterns at hillslope scales.