Development of a 3-D hydro-geochemical model to assess water quality and acidification risk in the Murray Lower Lakes, South Australia

Widespread decreases in rainfall throughout the Murray-Darling Basin, superimposed on the significant allocation of water for irrigation, resulted in an order of magnitude reduction in flow over the period from 2004-2009. This resulted in a dramatic shift in flow regimes and water quality in the lower River Murray, and in particular substantial decline in the water level of the Lower Lakes at the base of the Murray, which are disconnected from the ocean by barrages to prevent salt-water intrusion up the river. A consequence of the declining water level in the Lower Lakes was the exposure of pyrite-bearing lake sediments, leading to an increased risk of acid sulfate soil impacts. The potential for large-scale acid generation in sediments created the potential for acid and dissolved metal release to degrade water quality, with associated ramifications for biodiversity. Here we document a coupled model system for simulating lake water quality that adopted an existing three– dimensional (3-D) lake hydrodynamic-biogeochemical model (ELCOM-CAEDYM) and was coupled with a novel soil hydro-geochemical model suited to simulation of acid sulfate soils. The soil hydro-geochemical model required a simplified vertical representation to allow computationally efficient application across a spatially heterogeneous domain and for dynamic linking with the 3-D lake hydrodynamic-biogeochemical model, however, the vertical structure of acid sulfate soils can not be overly simplified due to the importance of the unsaturated zone soil moisture profiles and the variability of pyrite with depth for acidity dynamics. A separate 2-D sensitivity analysis using the full Richard’s equation solution was used to identify dominant system dynamics and parameterise the simplified vertical processes in the soil model, as well as horizontal connection to the lake. The coupled soil-lake model was validated against a range of surface water quality data collected during 2009. Several acidification events that occurred around the perimeter of the lake system during this period were well captured by the model, allowing us to identify the key processes controlling the generation and mobilisation of acidity. The model showed that the dominant flux of the acidity occurs following pulses of rain that collect acidity from surficial regions of the exposed sediment and drive lateral flow to the lake edge. Acidification of groundwater and subsequent baseflow to the lake was a much less significant delivery mechanism, and in this case the diffusive flux of acidity following reflooding was also low on average. The model demonstrated the large spatial variability in the manifestations of acid sulfate soils, and the potential for ‘hotspot’ locations that require priority management. The validated model has subsequently been used to estimate the range of timeframes for large-scale lake acidification through forecast scenarios, in order to allow for planning mitigation strategies and to highlight targeted areas for further field and laboratory research to help reduce model uncertainty.