Parameter sensitivity to climate and landscape variability of a simple, lumped salt and water balance model

Abstract. A salt and water balance model is developed to represent salinity generation following land use changes in Western Australia. The model consists of five interconnecting stores: (i) Dry, Wet and Subsurface unsaturated Stores, (ii) a transient Stream zone Store and (iii) a saturated Groundwater Store. The salinity generation process in Western Australia is highly dependent on annual rainfall, potential energy for evaporation, salt fall and land use history of a catchment. We selected six experimental catchments with different land use histories across a climatic gradient to test the model and assess parameter sensitivity. The model was successful in representing the streamflow and salinity generation processes of all catchments. In the process of application, we classified the model parameters into three sets: (i) "known", (ii) "fixed" and (iii) "variable". The "known" parameter set is calculated a priori from catchment attributes. The "fixed" set comprises regionalised parameters that remain unchanged across all catchments once calibrated in one catchment. The "variable" set of seven physically meaningful parameters were calibrated at one catchment, estimated a priori for other catchments and then subsequently adjusted for best fit. The "variable" set represents: (i) the depth (d), spatial distribution (b, c), relationship of the lateral hydraulic conductivity with moisture content (ia) and vertical conductivity (K_uv) of the top soil, (ii) lateral conductivity (K_ll) of the groundwater system, and (iii) salt release (C_u) from top soil. Sensitivity analyses of key model parameters show that the relationship of the top soil lateral hydraulic conductivity with soil moisture content (ia) is the most sensitive parameter. Other sensitive parameters include the depth of the top soil and its spatial distribution (d, b, c).