In arid and semi-arid environments, streamflow could be seasonal with highly nonlinear responses to rainfall and antecedent soil moisture. Physically-based modelling of such catchments is challenging due to the sensitivity to initial conditions, and the need to model all the components of the water balance appropriately. This paper investigates complexities around modelling the entire water cycle applying an integrated modelling framework, MIKE SHE, to a southwestern Victorian ephemeral catchment used as a pasture for grazing.
The three-dimensional Darcy equation and the one-dimensional Richards equation are used to couple the saturated and unsaturated zones. Surface flow is represented by the two-dimensional diffusion wave approximation. Root water uptake is described by the Kristensen & Jensen (K&J) dynamical reduction model, whose parameters were ascertained through model calibration and sensitivity analysis. Catchment discharge and climatic data were obtained from onsite gauges and a nearby weather station.
Good calibration and validation fit was achieved for discharge hydrograph and groundwater heads at the catchment outlet. Due to unmeasured heterogeneities in soil properties, upper slope bores showed some discrepancies from the observed levels; however, differentials conform to observations in one of these bores, indicating that the model provides realistic head fluctuations. Recursive simulation results showed that transition from an arbitrary initial condition to a steady-state solution depends heavily on the natural rainfall regime variations. A sensitivity analysis of probable combinations of K&J parameters revealed that the entire parameter space produced negligible variations in water table dynamics but corresponded to a more pronounced diversification in distributed evapotranspiration.
Despite simplifying assumptions, MIKE SHE proved capable in capturing ephemeral catchment flow dynamics. Recursive model run could be helpful in investigating attenuation rate in sensitivity to initial conditions as for this study it depended largely on the amount of precipitation suggesting the irrelevance of the steady state approach for ephemeral catchments.