Dynamic River-Aquifer interaction modelling and optimal interception of saline groundwater discharge

Abstract

The Murray River reach at Mildura, Australia, has been identified since the early 1970s as a place where saline groundwater discharge to the river has been exacerbated by the presence of a weir and a lock. It is believed that salt-laden groundwater has been entering the river in response to a normal operating head difference of 3.6 metres between upstream and downstream water levels. Deep groundwater, which has salinities several times those of seawater moves upwards to the river under a strong vertical gradient. In an effort to intercept this flow, the New South Wales Government install ed a dewatering network in 1979 known as the Buronga Salt Interception Scheme. Effluent is piped to a large gypsum swamp a few kilometres to the north. The area has been the subject of several modelling studies since the mid 1980s, initially with an analytical model, then a finite element model and more recently a finite difference now model and solute transport model. A coupled optimisation model has been used to derive optimal pumping rates for an expanded network of interceprion bores by insisting that groundwater heads remain at or below river levels. The river and the aquifer system are highly interactive. As a result, there is a highly dynamic variation in fluxes 10 and from the riv cr. The latest numerical model replicates the dynamics of the system faithfully. At times of high river flow, relatively fresh river water recharges the aquifers. During low river flow, saline groundwater enters the river below the weir. The river salt load estimates from the solute transport modelling are in good agreement with estimates based on multiplying water fluxes with local salinities. Given the good agreement, the latter approach is regarded as a sufficient indicator of sail loads. achieved with much less effort. The solute model predicts a "warm spot" immediately downstream of the weir, where elevated groundwater salinities can be expected. The solute modelling was useful in showing the importance of including the dynamics of hydraulic stresses, rather than the common assumption of steadystate flow. It was also instructive in showing the dilution that occurs upstream of tile weir and that flood recharge had to be added to the flow model to account for some observed fresh water zones. Salt load estimates with transient modelling are about five limes higher than steady-stare estimates. due to river dynamics and the slower response time of groundwater. After a high river flow event, groundwater levels near the river will remain elevated for some tirrc and salt-laden groundwater will discharge to the river until a new equilibrium is achieved. It is ironic that a fresh high flow event in the river causes a subsequent salt pulse in the river due to discharge of salty groundwater from a temporary mound. It is clear from the simulation modelling, and measured vertical head differences that the interception scheme's operation could be improved. The optimisation analysis recommends an increase of about 45% in pumping rates.