Macquarie Harbour (TAS, 2013-Current)


Macquarie Harbour is located on Tasmania’s west coast and is one of the largest salt-wedge estuaries in Australia. It is home to a growing salmon farming industry, and also adjoins the South-West World Heritage Area containing the Gordon & Franklin rivers.

The harbour is appoximately 35 by 9 km in size, and is fed by fresh water from a catchment of 6900 sq. km, the majority of which belongs to the Gordon and King Rivers. Both rivers are regulated to some extent by hydro-electric power schemes and have average flows of approximately 265 & 55 cumecs respectively. But these numbers can vary markedly with rainfall and power generation The power schemes control about 25% of the flow in the Gordon (capable or outputting 280 cumec), and 60% of the flow in the King. The water from the Gordon River is of high quality, while that of the King is very poor owing to the effects of the mining industry.

The harbour is connected to the Indian Ocean through a relatively narrow entrance, 400m wide but containing a much narrower navigable channel. Currents of up to 4 knots prevail during strong tides.

Depths in the harbour range up to 50m, but much of it is very shallow, particularly the perimeter and the sill separating the harbour from the ocean.

Field studies by Cresswell et al (1989) have suggested that the harbour consists of a 3-layer system ie a top layer controlled by river run-off, seasonal heating and cooling, a mid layer at about 20m which has a long residence time and is low in oxygen, and a lower layer within the deep basins that contains modified marine water brought in by tides over the sill.



A numerical model of Macquarie Harbour was originally developed as part of the Aquafin CRC (Volkman et al) to demonstrate the transferability of modelling approaches from the then existing aquaculture projects (D’Entrecasteaux Channel and Port Lincoln) to other sites. The model has since been updated to take advantage of new forcing data sets that enable its running in near real-time. The model is currently uncalibrated, and results are indicative only. We invite comparison of these results with observations, particularly of salinity and temperature, and we would readily accept feedback on the comparisons, along with a copy of any data that would assist in the calibration of the model.

The development of the current model follows on from the water quality work of Koehnken (2005) and it should ultimately include the implementation of both full 3D hydrodynamic and biogeochemical components. At this stage however, only the hydrodyamics have been implemented, but this could still allow a detailed analysis of the temperature/salinity regime of the harbour to be determined and hence confirm the structure of the harbour proposed by Cresswell et al (1989). It would also enable the determination of the flushing time of the harbour using either tracer and/or particle methods for a range of forcing scenarios.

Meromictic Lakes

To properly model the entirety of Macquarie Harbour, the model domain needs to extend up the Gordon River to a point beyond the reach of the tides and the salt-wedge. That point lies approximately 39 km upstream of the river mouth, between Butler Island and the Franklin River confluence. As such, the model domain includes that part of the river which flows adjacent to 3 unique lakes – Lake Morrison, Lake Fidler and Sulphide Pool. These lakes are meromictic i.e. they remain permanently stratified with a layer of anoxic saline water lying below a layer of oxygenated fresher water. There are only a handful of such lakes in Australia and only ~150 throughout the world. What makes these lakes more significant is the fact that they are shallow and would normally mix at regular intervals, in the presence of sufficient wind, to become holomictic. However it appears the Gordon lakes receive regular salt-intrusion from the river to maintain their meromixicity. The mechanisms for this intrusion have been the subject of much study (Tyler et al, 2001) ever since the lakes were first discovered in 1977 when the Gordon Power Scheme was being constructed.

It is thought that up to 3 different but related meteorological events need to coincide to provide the salt-water replenishment of the lakes. First, there needs to be a period of low river flow to allow the salt-wedge to push upstream past the lakes, second there needs to be a period of increased sea-level to allow the river level to reach the narrow interconnecting channels which feed the lakes, and (possibly) third, there needs to be an increased flow event to further lift river level and to entrain water of the appropriate salinity into the surface layer.

To date, a limited amount of modelling (Tyler et al, 2001) has been carried out to assess how and when the above circumstances might arise, both in the years prior to and after the flows of the Gordon were modified by the power scheme. This modelling used the POM hydrodynamic model to simulate steady flow scenarios (with a simple 12 hour tide) and provide estimates of times for the salt-wedge to penetrate beyond the lakes, as a function of river flow. The flow data records for the Gordon-below-Franklin (Palmer et al, 2001) were then scanned to estimate how many times salt-intrusion might have occurred, Four occasions between 1970 and 1999 were identified.

The above 4 low-flow periods all occurred in Summer and all were at times when the Gordon flow was interrupted by Hydro Tasmania to carry out dam/turbine maintenance. It is thought that ever since the Gordon flow has become regulated, with the consequence that low-flow periods no longer occur naturally, that natural salt replenishment of the lakes is unlikely to take place. To this end it is understood that Hydro Tasmania recently investigated using a barge to transport saline water from near the harbour entrance, and infusing this water into the deeper waters of Lake Fidler in an effort to artificially maintain its meromixicity.

A full 3D hydrodynamic model of both harbour and Gordon River, such as the one described here with realistic meteorology, tides and river flows, should enable detailed investigation of the circumstances under which salt-replenishment of the merimictic lakes might occur.

Model Implementation

Using an early version of the model, a preliminary run was conducted to investigate the ability of the model to predict the likelihood of salt intrusion into the meromictic lakes. That work is descibed in a following section.

The updated version of the model, as described in the next section, was then implemented and the sea-level results, both tides and low-frequencies, compared with the measured Strahan data. The agreement was not as good as expected and it was decided to improve this by nesting the model not directly inside OceanMAPS, but inside another already-operating near real-time model of Tasmania/Bass Strait. The latter has a 4km resolution and is itself forced by global tide harmonics, ACCESS and OceanMAPS. It has the advantage of having its open boundaries far removed from the Tasmanian coast where OceanMAPS is likely less accurate. The advantage provided by the nesting was evidenced by the significantly improved tidal and low-frequency signal predicted for Strahan.

But while the tides now appeared very good, there seemed room for improvement in the low-frequency signal which is responsible for the mean levels both in the harbour and along the rivers including up to the meromictic lakes.

To this end, advantage was taken of the extremely high correlation seen betwen the modelled low-frequency sea-level offshore, and the modelled signal inside the harbour. This is no doubt due to the very narrow entrance to the harbour. By calculating the lag and change in amplitude between the offshore signal and the harbour signal, this process could be applied in reverse to the low-pass filtered measured Strahan data, to infer an estimate of the offshore low-frequency signal. Accordingly, the model surface elevation on the offshore open boundary was now relaxed to the inferred signal with the result that sea-level comaparisons for Strahan are vastly improved.

The model has now been fully implemented in near real-time mode using the techniques described above. It runs contunuously drawing on the forcing data provided in near real-time from Hydro’s Strahan tide gauge and river stations, and BoM’s ACCESS and OceanMAPS models. In the following sections, near real-time output is provided in the form of time-series of sea-level at Strahan, in the Gordon near Lake Fidler and offshore, and salinity and temperature in the surface waters near Liberty Point. Snapshots and animations are provided for horizontal and vertical sections of salinity and temperature across the entire harbour. Please note, however, that at this stage the model is uncalibrated.

Model Improvement
Further enhancements to the model should include :-

1) Calibration using measured salinities and temperatures from within the harbour.

2) Improved bathymetry, particularly near the harbour entrance and in the Gordon River.

3) Addition of input from the Franklin River to the total Gordon flow.

4) Addition of biogeochemical modelling to complement the hydrodynamics.


The global ocean operational model OceanMAPS is used to provide offshore on the open boundaries. OceanMAPS is operated by BOM (Bureau of Meteorology) and is a data assimilating, eddy resolving model with 10 km resolution in the region surrounding Australia.

Surface fluxes are derived from the atmospheric model ACCESS-A, operated by the Bureau of Meteorology.

River flow and temperature data are provided by Hydro Tasmania.

AVHRR SST images processed by the CSIRO Remote Sensing Project are included for comparison with model output.


Cresswell G.R., Edwards R.J. and Barker B.A. (1989) Macquarie Harbour Tasmania, Seasonal Oceanographic Surveys in 1985, Papers and Proceedings of the Royal Society of Tasmania, Vol. 123.

Koehnken L. (2005) Overview of Water Quality in Macquarie Harbour and Assessment of Risks due to Copper Levels. Prepared for DPIWE March 2005.

Palmer L., McConachy F. and Peterson J. (2001) Basslink Integrated Impact Assessment Statement, Potential Effects of Changes to Hydro Power Generation, Appendix 2, Gordon River Hydrology Assessment. Prepared for Hydro Tasmania June 2001.

Tyler P.A., Terry C. and Howland M.B. (2001) Basslink Integrated Impact Assessment Statement, Potential Effects of Changes to Hydro Power Generation, Appendix 11, Gordon River Meromictic Lakes Assessment. Prepared for Hydro Tasmania June 2001.

Volkman J.K., Thompson P., Herzfeld M., Wild-Allen K., et al (2009) A Whole-of-Ecosystem Assessment of Environmental Issues for Salmonid Aqualculture. Aquafin CRC Project 4.2(2).