Nitrogen species fate and transport modelling in a multi-layered sedimentary aquifer of the Swan Coastal Plain

The discharge of nitrogen contaminated groundwater to the marine environment of Cockburn Sound in Western Australia has resulted in significant water quality decline and marine habitat loss. Eolian and marine deposited, organic carbon depleted, sandy, surficial sediments that occur across the coastal plain provide little protection to the Superficial aquifer from contamination. Despite numerous efforts to reduce contaminant concentrations in groundwater discharging to Cockburn Sound, sizeable historical contaminant plumes persist. Stratified geological sequences and very low hydraulic gradients may account for the longevity of these plumes. However, the exploitation of these characteristics may also offer the potential for successful remediation.

A case study was undertaken to assess this potential in relation to a nutrient plume (ammonia and nitrate) at Summit Fertilisers. The study employed standard assessment methods such as groundwater monitoring and aquifer tests. More novel technologies in geophysics (electrical resistivity imaging) and mineralogy assessment (x-ray florescent analyser) have also been utilised in order to
characterise subsurface conditions and quantify contaminant distributions. Use of these techniques, particularly the use of geophysics, proved advantageous by reducing requirements for expansion of the monitoring network offsite as well as offering increased speed of data acquisition.

The bulk of contamination at the Site was found to reside in the Becher Sand of the Superficial aquifer. The higher proportion of fine grained sediments in the basal part of this unit, lowers its hydraulic conductivity so that effectively it acts as a local scale, discontinuous aquiclude. The presence of this aquiclude has reduced vertical migration of contaminants from the overlying Safety Bay Sand to the underlying Tamala Limestone at the Site.

Both ammonia and nitrate were observed to behave conservatively and consequentially should be highly mobile in the system. Although the mobility of the contaminants was very high, the plume was moving at only approximately 0.02 m/day to the north. This is inconsistent with high hydraulic conductivities in the Superficial aquifer (10-100 m/day) and regional head gradients (~0.0001 m/m). It is also inconsistent with the east-west direction of regional groundwater flow.

To understand these inconsistencies while making predictions of plume impacts and remediation, regional and local data were compiled into a conceptual model that formed the basis of a numerical groundwater model. This numerical model was constructed using MODFLOW 2000 coupled to MT3DMS in PMWIN. The aim of this modelling was to further develop our understanding of the system, in
particular to assess the stability of the plume and its sensitivity to changes in groundwater use on a both a local and regional scale.

While regional and local scale groundwater flow was well modelled, local site scale solute transport was not. This leads us to the conclusion that it must be strongly influenced by density driven flow, geological heterogeneities and the first flush effect in the contamination source, none of which were replicated due to a lack of data. Despite this, the model was able to confirm future persistence of an onsite contaminant source, the sensitivity of local groundwater flow to
anthropogenic factors (drainage basins and abstraction) and the likelihood of Site contamination negatively impacting neighbouring abstraction bores as well as Cockburn Sound.

Additional research is required to improve our understanding of contaminant transport mechanisms in the Superficial aquifer in the Kwinana Industrial Area (KIA). This is required for prioritising contamination management to ultimately minimise impacts to both human and ecological receptors.
Recommendations for further work following the outcomes of this study included:
• improvements to onsite stormwater collection/disposal to cease additional nutrient loading to the Superficial aquifer;
• expansion of the current onsite monitoring program to include sampling of water within onsite basins and assessment of density and viscosity, particularly during the first flush;
• confirmation of abstraction volumes from other groundwater users within the KIA;
• improved understanding of aquifer heterogeneities to refine our understanding of local and regional aquifer hydraulic properties; and
• updating the numerical model following the collection of this additional data, incorporating density coupled flow at the site and salt water interface.