Tuna Risk and Response (SA, 2004-2007)

Risk and Response Project

Download the Technical Report 42.5 MB

The physics and biogeochemistry of the Boston Bay area in south-west Spencer Gulf, South Australia were studied to assist the tuna farming industry in identifying and mitigating risks to the established tuna aquaculture industry in the region. This project was conducted within the Aquafin CRC (www.aquafincrc.com.au/home) in collaboration with SARDI; an overview of which is found in:

Tanner, J.E. and J. Volkman (Eds.) 2009. Aquafin CRC – Southern Bluefin
Tuna Aquaculture Subprogram: Risk and Response – Understanding the Tuna
Farming Environment. Technical report, Aquafin CRC Project 4.6, FRDC
Project 2005/059. Aquafin CRC, Fisheries Research & Development
Corporation and South Australian Research & Development Institute
(Aquatic Sciences), Adelaide. SARDI Publication No. F2008/000646-1, SARDI Research
Report Series No 344, 287 pp.

The hydrodynamic component is described in an appendix to the above study;

Herzfeld, M., Mideleton, J.F., Andrewartha, J.R., Luick, J., Leeying, W. (2008) Numerical hydrodynamic modelling of Boston Bay, Spencer Gulf. Technical report, Aquafin CRC Project 4.6, FRDC Project 2005/059. Aquafin CRC, Fisheries Research & Development Corporation and South Australian Research & Development Institute (Aquatic Sciences), Adelaide. SARDI Publication No. F2008/000745-1, SARDI Research Report Series No 342, 100 pp.

Key outcomes from the study are outlined below.

Boston Bay is situated in the lower western side of Spencer Gulf (Figure 1) on the southern coast of Australia, and is the base for a large tuna aquaculture industry.  This industry occupies the waters offshore of Boston Island, in an area hereafter called the tuna farming zone (TFZ).  This area has been subject to detailed studies of the interactions between aquaculture and the environment.  This study seeks to address environmental risks to the tuna industry, and to assess where nutrients released by the industry are dispersed to and what their potential environmental effects are.

Spencer Gulf/Boston Bay region.
Figure 1: Spencer Gulf/Boston Bay region.

A 3D primitive equation model was applied to Boston Bay to examine the hydrodynamics of the region. Using a nesting process the region was represented with high resolution while incorporating forcing due to wind stress, tides, low frequency sea level oscillations and pressure gradients due to temperature and salinity distributions. The open boundaries of this model were supplied from a larger scale regional model that encompassed the whole of Spencer Gulf. This model was in turn forced with data collected in the field. The model nesting is illustrated in Figure 2.

The regional grid is curvilinear with variable resolution over the domain. Seaward of Boston Island the resolution is ~1500 m with resolution increasing to > 6 km on the eastern side of the gulf. The model uses 23 layers in the vertical with 0.5 m resolution at the surface and ~8 m resolution near the maximum depth of 60 m. The grid size is 55 x 95 x 23; 45% of surface cells are wet cells and 30% of all cells in the grid are wet. A curvilinear grid was used to model the TFZ region. The grid spacing seaward from Boston Island is ~330 m and a maximum resolution of ~1 km exists on the offshore open boundary. The grid dimensions are 135 x 70 x18, with 0.5 m resolution in the vertical at the surface and ~4 m resolution at the bottom with a maximum depth of 30 m. This domain also consists of a high percentage of land cells, with 53% of the surface layer comprising wet cells and only 28% of the 3D domain being wet.

Two-tiered nesting strategy.
Figure 2: Two-tiered nesting strategy.

The model was simulated for 12 months for the period Sep 2005 – Aug 2006 and calibrated to sea level measurements at Port Lincoln, Whyalla and Wallaroo, and temperature, salinity and velocity at numerous locations in the Boston Bay region obtained from a dedicated field program (Figure 3). The model was able to successfully reproduce tidal and low frequency sea level oscillations in both regional and Boston Bay models (Figure 4). Temperature, salinity and currents (Figure 5) corresponding to tidal (diurnal) and weather band (periods 3-20 days) frequencies were also successfully reproduced by the model.

Details of moorings deployed in the TFZ. Mooring 4 was dragged during deployment #3; start and end locations are denoted 4c_start and 4c_end respectively. The thick black line corresponds to a CTD transect that was sampled monthly.
Figure 3: Details of moorings deployed in the TFZ. Mooring 4 was dragged during deployment #3; start and end locations are denoted 4c_start and 4c_end respectively. The thick black line corresponds to a CTD transect that was sampled monthly.
Modelled and measured sea level at Port Lincoln, Wallaroo and Whyalla.
Figure 4: Modelled and measured sea level at Port Lincoln, Wallaroo and Whyalla.
Modelled (red) and measured (black) north/south high and low frequency depth averaged currents at Mooring 4. Positive values are to the north.
Figure 5: Modelled (red) and measured (black) north/south high and low frequency depth averaged currents at Mooring 4. Positive values are to the north.

Both the data and model results showed there to be a strong (~20 cm/s) tidal currents (e.g. Figure 5a, Figure 6) that may be implicated in bottom stirring. Although the tide may be responsible for trajectories of over 8 km, the net displacements due to these currents are small (less than 1.4 km over a 3 hr period). Magnitudes of the weather-band currents are smaller than the tidal currents (< 5 cm/s), however, due to their longer periods these currents are the dominant contributor to residual flow and hence primarily responsible for transport and flushing of the region. Both data and model indicate the residual currents eastward of Boston Island (in the tuna farming zone) to be weak (~ 1 cm/s) and to the north/north east during both summer and winter. The transport due to these flows over a 3-month period is around 80km. The currents were also found to be strongly sheared in the vertical and so may be important to shear enhanced diffusion and dispersal. However, estimates of the flushing times based on tracers and Lagrangian tracking show flushing times scales of 10 days (Boston Bay) to 2 days for the outer bay region. The flushing time for the whole domain based on particle tracking is ~20 days. Flushing estimates are provided in Table 1.

Surface currents and sea level on flood and ebb spring tides.
Figure 6: Surface currents and sea level on flood and ebb spring tides.
Estimates of flushing times for January 2006.
Table 1: Estimates of flushing times for January 2006.

The tide in the region is classified as semi-diurnal mixed and is dominated by the semi-diurnal constituents M2 and S2, and the diurnal constituents K1 and O1. Coincidentally, all these constituents have approximately the same amplitude of ~0.18 m, and when they are out of phase they destructively combine to produce very little tidal variation for several days. During these periods, called the ‘dodge tide’, the tidal currents are small and transport is primarily wind driven (e.g. Figure 7). If wind-speed is low, then it is possible that the region is very poorly flushed. The model was able to reproduce the occurrence of the dodge tide, allowing predictive capability of the timing of these events.

Sea level and surface currents during the dodge tide, 8th January 2006.
Figure 7: Sea level and surface currents during the dodge tide, 8th January 2006.

There exists a degree of connectivity between the coastal zone and the outer bay region in summer that can be caused by local upwelling. Offshore (eastward) winds force surface waters offshore, resulting in compensatory onshore interior or bottom flow (e.g. Figure 8). In addition, the larger evaporation that occurs near the coast leads to dense water formation and bottom plumes that flow to the outer bay region. During winter, similar plumes result from coastal cooling rather than evaporation. The annual temperature cycle in the Boston Bay region is largely driven by atmospheric heating and cooling. Salinity is also controlled by atmospheric exchanges, but to a lesser extent, with advective processes playing a more dominant role. Large decreases in salinity in autumn occur, coincident with the flushing of Spencer Gulf when fresher compensatory oceanic flows enter western Spencer Gulf. Due to the atmospheric exchanges, the Boston Bay region was also found to stratified during summer and well mixed during winter.

Section of temperature and currents during strong offshore wind showing local upwelling.
Figure 8: Section of temperature and currents during strong offshore wind showing local upwelling.

Velocity profiles and density distributions reveal the TFZ to be stratified during summer and well mixed during winter. However, in summer periods of uniform properties exist through the water column and the variability in the depth of the mixed layer is large; instances of stratification typically occur during periods of light winds and large net heat input into the surface, and strong winds in conjunction with strong tidal mixing can destroy this stratification on a time-scale of days. Lighter water is found in the shallow coastal regions and Proper Bay as a result of differential heating. In the winter this trend is reversed. Convective cooling in conjunction with tidal mixing maintains the well mixed nature during winter.

The depth averaged seasonal flow can be characterised into three main sub-regions within the domain. Firstly offshore of Boston Island flow enters the domain in the south and exits in the north, with little penetration into the coastal margins. Secondly the Boston Bay /Proper Bay area can be treated as a separate system, with flow generally entering north ofBoston Island, flowing south to loop through Proper Bay and exiting to the east of BostonIsland, where a persistent anti-clockwise gyre exists off Cape Donnington. Thirdly Louth and Peake Bays exhibit northwards flow along the coast, fed by water seaward of Point Boston, and seasonally exhibiting gyres within Peake Bay and off Louth Island. A schematic of the residual flow is shown in Figure 9.

Schematic of depth averaged mean flow.
Figure 9: Schematic of depth averaged mean flow.

Analysis of residual flow and connectivity inferred from passive tracer distributions and particle tracking allows the region to be categorised into three main regions, consistent with passive tracer and particle tracking analyses;

  1. A region encompassing Proper Bay and the area landward of Boston Island, which has poor connectivity with the rest of the domain,
  2. Louth and Peake Bay regions, which also have poor connectivity with the rest of the domain,
  3. Regions outside these Bays and offshore of Boston Island with good connectivity with the remainder of the domain, but subject to greater flushing.

Particle trajectories under the influence of spring tides show the oscillatory nature of the tide (Figure 10a), especially near the offshore boundary. Gross displacements may be large, over 8 km, but small net displacements are observed. Neap tide trajectories exhibit little tidal motion, with particle displacement dominated by the wind (Figure 10b on 7 January 2006 for an easterly wind, showing net westward motion). Dodge tides reveal a similar situation to neap tides.

Particle trajectories during spring and neap tides.
Figure 10: Particle trajectories during spring and neap tides.

The Boston Bay numerical model was found to have satisfactory predictive skill, and as such is a suitable tool to couple sediment transport and biogeochemical models to for transport purposes.