Sensitivity of simulated stream salinity to soil chloride profile

Removal of native vegetation and intensification of agriculture has led to salinisation of soils and streams in the south-west of Western Australia. Four significant changes have typically been observed: an increase in streamflow volume; a relocation of deep groundwater levels; expansion of riparian wet zones; and an increase in stream salt loads. Can we predict stream salt loads following deforestation without data on the chloride profile, the vertical distribution of chloride in the unsaturated zone? This question is addressed in the context of the 1 km2 Wights experimental catchment which was clearfelled and converted to pasture. Time series data extending over 16 years for rainfall, saltfall, streamflow, stream chloride and deep groundwater levels were available. Using the CATPRO lumped hydrosalinity model a generalised chloride profile was parameterised and jointly calibrated to the streamflow, stream salinity and groundwater time series. It was found that the chloride profile parameters were virtually unidentifiable. Although qualitatively different assumptions about the initial chloride profile produced different groundwater chloride concentration time series, they produced virtually identical stream salinity time series. It is suggested that time series of groundwater chloride concentration observations will be necessary to better identify chloride dynamics in lumped models such as CATPRO.