Stream salinity regime shifts related to dry and wet periods

It is well assumed stream salinity increases during droughts due to more evapoconcentration and less dilution of saline groundwater; however, changes in stream salinity during prolonged droughts, such as the Millennium drought, are not well studied. There are long-term records, 20+ years, of high-frequency electrical conductivity data in several streams in Victoria which serves as a surrogate for salinity; however, the usefulness of these data are limited due to a lack of temporal explanatory power in existing water quality models that predict salinity concentration as a function of streamflow. Therefore, eight (8) newly derived models that explain concentration with quick and slow flow are evaluated along with seven (7) existing concentration-discharge (C-Q) models. The best model is one of the newly derived ones. It significantly improves temporal performance, and provides an objective estimate of baseflow. The residuals from this model are then analysed using a hidden Markov model approach to identify when shifts in stream salinity have occurred on a monthly time-step. Of the 23 catchments evaluated, six (6) show a decrease in salinity during and persist following the drought, and the correlation of the shifts with water table levels suggests the decline is at least partly attributable to less saline groundwater flowing into the stream. This finding contradicts the well adopted assumptions that salinity increases during droughts and suggests less recharge is occurring. Shifts in salinity at many of the sites (8) are more challenging to diagnose, but four (4) of these sites also show changes in the water table aligning with shifts in stream salinity. Two (2) sites show no distinct shift in salinity, and the groundwater at these sites declines and recovers. Some catchments are more sensitive to prolonged dry periods with less recharge which may have further implications for salt accumulation in soils due to less infiltration.