Technical Report 7

Hydrodynamic Modelling of the Port Curtis Region

Project CM2.11

M. Herzfeld, J. Parslow, J. Andrewartha, P. Sakov and I. T. Webster.

April 2004

1

Hydrodynamic Modelling of the Port Curtis Region CRC for Coastal Zone, Estuary and Waterway Management Technical Report 7 M. Herzfeld, J. Parslow, J. Andrewartha, P. Sakov and I. T. Webster. April 2004

National Library of Australia Cataloguing-in-Publication data: Hydrodynamic Modelling of the Port Curtis Region ISBN 0 9578678 8 3 (print)

This report may be copied and distributed for research and educational purposes with proper acknowledgement.

Further information should be addressed to:

CRC for Coastal Zone, Estuary and Waterway Management 80 Meiers Rd Indooroopilly Queensland 4068 Tel: 61 7 3362 9399 Fax: 61 7 3362 9372

Disclaimer: This report describes the pilot implementation of the MECO hydrodynamic model to Port Curtis. Although tidal height data have been used to calibrate water levels, the model has not been validated against measurements that could be used to assess its reliability for predicting currents and mixing within the harbour. Accordingly, the Coastal CRC and CSIRO cannot guarantee the accuracy of the model predictions, and does not recommend or endorse their use in applications where accuracy is critical

Table of Contents

Summary ______1 1. Background. ______2 2. Objectives. ______3 3. The Hydrodynamic Model. ______4 4. Model Domain.______6 5. Input Data. ______9 5.1 Wind Forcing. ______9 5.2 Surface Elevation.______11 5.3 Temperature and Salinity. ______14 5.4 River Flow.______15 6. Model Output. ______17 6.2 General Solutions ______17 6.2 Residual Currents ______24 6.3 Flushing Characteristics ______27 6.4 Transport Characteristics ______31 6.5 Connectivity ______38 7. Benefits and Outcomes ______44 8. Further Development ______44 9. Conclusions______46 10. References ______47

Summary

A computer ocean model was developed for the Port Curtis Estuary to examine the oceanographic characteristics of the region. The model highly resolved the region of interest and predicted sea level, currents and distributions of dissolved material in the water in response to the effects of wind, tides, slow sea level changes and density effects. Such a model could be used to predict the movements and mixing of contaminants throughout the harbour, knowledge that could be used to support management. The time period investigated was the first four weeks of 1999.

Results from the model show that the currents in the estuary are predominantly due to the effects of the tide, with the change in water level between high and low tide being as large as 4m. These large tides can generate very large currents, with some current speeds reaching up to 2ms-1 in the vicinity of North Channel. The tidal range is not constant but varies over an approximate 14-day period, with the smallest tidal range during this period being approximately half the largest range. The large tides also mix the water vertically so that any dissolved material (e.g. salt, temperature, contaminants) shows little variation from the surface to the bottom.

A particle in the estuary undergoes large displacement due to the tidal motion which may be as great as 15km. However, tidal displacements are of a back-and-forth nature, and after a large displacement on one phase of the tide the particle returns to basically its original position on the next. This results in little net displacement of particles over multiple tidal cycles. This displacement is directed up-estuary in the lower estuary and is evident as a series of small eddies in the upper estuary. The net displacement is mainly due to the interaction of the tides with the bottom. In the offshore regions outside the estuary the residual circulation is mainly due to the wind and is directed along-shore towards the north-west. Net flow enters the estuary through Gatcombe Channel and exits through North Channel. The net flow characteristics may vary if the climatic conditions under which the model was run change.

The flow regime within Port Curtis estuary allows dissolved material to be dispersed evenly throughout the estuary (i.e. the estuary is well connected), however material has difficulty leaving the estuary into the offshore environment. The e-folding flushing time (i.e. the time for total mass of material to decrease to ~1/3 of its original mass) for the estuary is of the order of 19 days in January 1999. This estimate applies to the estuary as a whole, and smaller sub- regions in the estuary have much shorter flushing times due to the well connected nature of the estuary. The flushing time for Rodds Bay is also much shorter as a result of good offshore exchange, having an e-folding time of 5 days in January 1999.

If material is input into Port Curtis estuary as an external source and subsequently dissolves, then appreciable concentrations of the material occur within the estuary and negligible concentrations are found seaward of Facing Island. Material input into Rodds Bay results in far smaller concentrations, with largest concentrations restricted to the areas south of the input location. Release at the dredging spoil site results in distributions of dissolved material in the form of a plume originating from the source and directed north-westwards along the seaward coast of Facing Island. The prevailing wind conditions are likely to influence offshore distributions, hence the distribution is expected to vary during the year.

The main output of the project is the development of the hydrodynamic model. With appropriate forcing data this model may be applied to any future, present or past time period. The simulations generated by the model and associated analyses provide a first order picture of the flow and distribution characteristics of the Port Curtis environment which may aid in management decision processes and generally provides enhanced understanding of the oceanography of the region. Once fully implemented, the model has the capacity to predict the outcomes of scenarios which may aid in management strategy evaluation. The model is currently at a pilot stage and requires further calibration and validation in order to achieve full confidence in solutions. The model should also be run under an expanded range of climatic conditions to better characterize variability in the region.

1 1. Background.

The Port Curtis region (Figure 1.1) is situated at the transition between the tropics and sub- tropics in Central Queensland on the eastern coast of Australia. Port Curtis is a naturally protected deep water harbour that is the largest port in Queensland and the second largest on the eastern Australian coast in terms of tonnage handled. The surrounding land has become industrialized in the past 30 years and is now home to major industrial activity. Gladstone is the major urban center in Port Curtis with a population of 27,000 and is associated with four major industries; aluminium, cement production, chemical production and electricity generation. All these industries discharge waste material into the harbour or atmosphere (under licence). The bathymetry in the harbour has also been modified by the development of shipping channels, land reclamation and coastline armoring. Dredging of the shipping channel occurs regularly, with the spoils deposited at a location approximately 9km south east from Facing Island. Heavy metal concentrations in the sediments of the estuary (e.g. Cr, As, Ni) are a concern. The residents of the region also use the waters in the Port Curtis area for recreational purposes, including fishing, sailing and access to the nearby southern reaches of the Great Barrier Reef. There exists a collective awareness in the community about managing the region’s aquatic environment in a sustainable manner.

Port Curtis is a macro tidal estuary with large barotropic tides having ranges up to 4m. The tide propagates into the estuary through the straits separating Facing Island from the mainland (Gatcombe Channel) and Curtis Island (North Channel) in the south-west, and through the Narrows in the north. Tides undergo a neap-spring cycle with a period of approximately 14 days, with ranges at the spring of ~4m and about 1m during the neap. Maximum currents during the spring phase may be as large as 2ms-1 in North Channel (Witt and Morgan, 1999). Fresh water flows may originate from the Narrows as a result of the nearby Fitzroy River flooding and the Calliope River which discharges into the estuary through Gladstone. The large tides ensure that the water column is vertically well mixed most of the time, and are also responsible for significant resuspension of fine sediment. Combined with very large deposits of silt from the hinterland in times of flood, the estuary maintains a highly turbid character. The region is characterized by extensive areas of tidal flats that become exposed at low tide and large areas of mangroves fringing the estuary which behave as a storage buffer for water at high tide. These mangroves and tidal flats have ecological significance, being home to numerous aquatic fauna and flora. The region is also home to stands of seagrass beds, notably Rodds Bay in the south, which in turn lure marine mammals (dugongs) to feed.

The combination of anthropogenic pressure, the presence of natural habitats and the community desire to appreciate the natural and recreational benefits of the region whilst sustaining industry make the management of Port Curtis a challenge. This project aims to provide support for the assessment of the impacts of various developments and management approaches.

Figure 1.1 : Port Curtis Region.

151o 10 / E 151o 20 / E 151o 30 / E 151o 40 / E The Narrows 23o 40 / S Curtis Island North Channel

Facing Island

Fishermans 23o 50 / S Landing Gladstone Port Gatcombe Channel

Queensland 24o S Rodds Bay

Depth (m)

4 14.5 25 2 151o 10 / E 151o 20 / E 151o 30 / E 151o 40 / E 2. Objectives.

In order to assess the physical characteristics of Port Curtis estuary this study aims to implement a prototype numerical hydrodynamic model that will provide predictive capacity for currents and mixing within the Port Curtis region. Data from Port Curtis will be obtained from tasks PC1 and PC3 and from various other sources. For this study, the hydrodynamic model will be developed to the stage of a working pilot model for Port Curtis. Model calibration will be undertaken later. Insight into exchange mechanisms of the estuary with the open ocean, flushing times, tracer dispersal distributions and residual flows can be gained from application of the model, which in turn is designed to ultimately aid management decisions relating to contaminant management. The model was forced with river flow, wind stress and surface elevation, temperature and salinity on the offshore limits of the domain. A pilot regional scale hydrodynamic model which extends into the Great Barrier Reef lagoon and several hundred kilometres north and south is developed as part of this activity and used to establish boundary conditions for the coastal model and subsequently to study connections between the Reef and the coastal study regions. The hydrodynamic model, its inputs, and model output is discussed in more detail below. Analyses are presented addressing the flushing characteristics of Port Curtis estuary, passive tracer distributions in response to the circulation, residual flow dynamics and connectivity. Limitations of the model, further effort required to improve confidence in model solutions and data requirements to facilitate these improvements are discussed.

3 3. The Hydrodynamic Model.

The hydrodynamic model used to simulate the physics of the Port Curtis region is MECO (Model for Estuaries and Coastal Ocean; Walker and Waring, 1998). This model has been developed by the Environmental Modelling group at CSIRO (Commonwealth Scientific and Industrial Research Organization) Division of Marine Research over the last decade. MECO is intended to be a general purpose model applicable to scales ranging from estuaries to regional ocean domains, and has been successfully applied to a variety of applications encompassing these scales to date. MECO is a three-dimensional finite difference hydrodynamic model based on the primitive equations. Outputs from the model include three- dimensional distributions of velocity, temperature, salinity, density, passive tracers, mixing coefficients and sea level. The equations forming the basis of the model are similar to those described by Blumberg and Herring (1987). Inputs required by the model include forcing due to wind, atmospheric pressure gradients, surface heat and water fluxes and open boundary conditions (e.g. tides). A schematic of the major forcing mechanisms captured by MECO is included as Figure 3.1. MECO is based on the three-dimensional equations of momentum, continuity and conservation of heat and salt, employing the hydrostatic and Boussinesq assumptions. The equations of motion are discretized on a finite-difference stencil corresponding to the Arakawa C grid.

Figure 3.1 : Schematic of forcing mechanisms in MECO

Inflow Surface fluxes Elevation Wind gradients

Mixing Density gradients

Bathymetry Open boundaries Bottom friction

The model uses a curvilinear orthogonal grid in the horizontal and a choice of fixed ‘z’ coordinates or terrain following σ coordinates in the vertical. The curvilinear horizontal grid was particularly useful in this application since it enabled high resolution to be specified in areas of the study region where small scale motions were present and larger resolution where they were not. The ‘z’ vertical system allows for wetting and drying of surface cells, useful for modelling regions such as tidal flats where large areas are periodically dry. This is likely to be important in a region such as Port Curtis which has a large tidal range and extensive drying areas. MECO has a free surface and uses mode splitting to separate the two-dimensional (2D) mode from the three-dimensional (3D) mode. This allows fast moving gravity waves to be solved independently from the slower moving internal waves allowing the 2D and 3D modes to operate on different time-steps, resulting in a considerable improvement in computational efficiency. The model uses explicit time-stepping throughout except for the vertical diffusion scheme which is implicit. The implicit scheme guarantees unconditional stability in regions of high vertical resolution. A Laplacian diffusion scheme is employed in the horizontal on geopotential surfaces. Smagorinsky mixing coefficients may be utilized in the horizontal.

4 MECO can invoke several turbulence closure schemes, including k-ε, Mellor-Yamada 2.0 and Csanady type parameterisations. A variety of advection schemes may be used on tracers and 1st or 2nd order can be used for momentum. This study used the QUICKEST advection scheme for tracers (Leonard, 1979) in conjunction with the ULTIMATE limiter (Leonard, 1991). This scheme is characterized by very low numerical diffusion and dispersion, and yielded excellent performance when resolving frontal features, which often occurred during tracer analyses. MECO also contains a suite of radiation, extrapolation, sponge and direct data forcing open boundary conditions. Input and output is handled through netCDF data formatted files, with the option of submitting ascii text files for simple time-series forcing. The netCDF format allows input of spatially and temporally varying forcing and initialization data in a grid and time-step independent manner. MECO is capable of performing particle tracking and may be directly coupled to ecological and sediment transport models.

5 4. Model Domain.

The simulation of the physics of the Port Curtis Estuary required the construction of three model grids. A large scale ‘super’ grid was developed to generate tidal harmonics suitable for forcing a regional scale grid. The regional grid supplied the initial and open boundary conditions for a smaller grid of the study region nested within the larger grid. In the absence of field-derived temperature, salinity and surface elevation measurements to apply to the open boundaries, this strategy is the only way of adequately driving the model through the open boundaries. The domain nesting is illustrated in Figure 4.1, the regional domain in Figure 4.2 and Port Curtis (local) domain in Figure 4.3.

The super grid was required to be executed in two-dimensional depth averaged mode only, as the goal was to reproduce the sea level on the boundary of the regional grid. The super grid was used with a rectilinear grid of resolution 4.4km. A rectilinear grid covers the regional grid with a resolution of 2.2km and 22 layers in the vertical with 3m resolution at the surface and 80m resolution near the maximum depth of 600m. River flows representing the Fitzroy and Calliope Rivers are included, and south Keppel Bay is connected to Port Curtis estuary via The Narrows, which has approximately the same cross sectional area as the real geography but is wider and shallower in the model owing to the discretization used.

A curvilinear grid was used to model the Port Curtis region where high resolution is achieved in the estuary and Rodds Bay, with coarser resolution near the offshore boundary. The grid spacing varied from ~200m inshore to 600m at the seaward boundary. There are 19 layers in the vertical with 0.5m resolution at the surface and 5m resolution near the maximum depth of 30m. This domain consists of mostly land cells, with 38% of the surface layer comprising wet cells and 24% of the 3D domain being wet.

Due to the relatively large spatial extent of the domain using a fine resolution grid, the run- time ratio of the model is approximately 40:1 (i.e. the model simulates 40 days of results in 1 day of real time). This is determined by the stability constrains on the model which limit the maximum time-step to be used for 2D and 3D modes. These constraints are in turn dependent on the grid resolution, the water depth, stratification and the size of the grid. Considerable effort was invested into optimizing these time-steps so as to achieve the largest run-time ratio possible. However, the model speed remains relatively slow, and this means that the model is only suitable for simulations of several neap-spring tidal cycles duration (e.g. up to 60 days) and thus is appropriate for investigating flushing and circulation dynamics on these shorter time-scales at present.

6 Figure 4.1 : Nested domain consisting of super grid (black), regional grid (red) and local grid (blue).

Figure 4.2 : Regional Domain

7 Figure 4.3 : Port Curtis Domain with the discretized grid superimposed

8 5. Input Data.

The model was forced with wind, river flow from the Calliope and Fitzroy (via the Narrows) Rivers and elevation, temperature and salinity on the oceanic open boundary. The model simulation period was chosen as January 1999, and all data are collected to span this period. The sources of the forcing data are detailed below.

5.1 Wind Forcing. Wind speed and direction data were obtained from the Bureau of Meteorology at the locations depicted in Figure 5.1 and interpolated onto the regional and Port Curtis domains to provide a temporally and spatially varying wind-field.

Figure 5.1 : Wind Measurement sites

A sample of the wind-field at selected sites is shown in Figure 5.2 (a) and (b) for the first three months of 1999. The mean wind speed and direction during this period is shown in Table 5.1.

Table 5.1 : Mean wind speed and direction for wind measurement sites

Site Mean Wind Speed (ms-1) Mean Wind Direction (oT) Gladstone 4.2 106 Miriam Vale 3.9 123 Heron Island 4.6 130 Lady Elliot Island 4.9 150 Sandy Cape 4.8 147 Yeppoon 3.6 103 Rockhampton 2.4 118 Bundaberg 3.1 158 Rundle Island 5.6 126 Mean 4.1 129

9 Figure 5.2 (a) : Wind Speed at Measurement Sites

Figure 5.2 (b) : Wind Direction at Measurement Sites

10 Figure 5.2 and Table 5.1 indicate that for the first three months of 1999 the mean wind in the Port Curtis region was a relatively light (~4 ms-1) south-easterly.

5.2 Surface Elevation.

The time series of surface elevation prescribed on the open boundaries of the Port Curtis domain were supplied from output of the regional model. The elevations used in the regional model consist of a high frequency component (tidal component with periods < 1 day) and a low frequency component with periods of days to weeks. The tidal component applicable to the regional domain was derived from output of the super grid, which in turn was forced only on the open boundaries with tides constructed from a global tidal model (Eanes and Bettadpur, 1995).

Tidal records were obtained at hourly intervals over a variety of locations (courtesy of National Tidal Facility and Queensland Dept. Transport) shown in Figure 5.2.1. Since the phase and amplitude of the tide are variable over the study region, these data do not possess adequate spatial resolution to construct an accurate tidal forcing of the regional open boundaries; hence the use of the global tide model which is capable of supplying tidal data at arbitrary resolution. However, the measured data are useful for calibrating the tide, and the Burnett Heads data were used for extracting the low frequency variations (e.g. coastally trapped waves, storm surges) with which the southern boundary of the regional model was forced.

Figure 5.2.1 : Tidal Observation Locations

Using the global tide model directly to force the open boundaries of the regional model resulted in poor performance in the vicinity of the northern and southern cross-shelf boundaries. This is due to the global tide model’s tendency to supply inaccurate constituents near the coast, and reflection and over-specification problems with the cross-shelf open boundary conditions themselves. Further away from the cross-shelf boundaries, the model performed adequately, suggesting that it is the offshore open boundary that is predominantly

11 responsible for driving the tide in the domain interior. This scenario prompted the use of the super grid, whose cross-shelf open boundaries were distant enough from the zones encompassing the regional model’s open boundaries so as to provide accurate elevations in these areas. The super grid was run in 2-dimensional mode only to provide time series of surface elevation on the regional grid’s open boundaries. These time series were then decomposed into the tidal constituents, which were subsequently used to force the tidal component in the regional model. This nesting approach provided better results than directly imposing the global tidal model constituents on the regional model’s boundaries. Comparison of sea level output at Burnett Heads from the super grid model and global tide model with observation is presented in Figure 5.2.2. It should be noted that open boundary conditions generally all suffer from reflection problems which contaminate solutions in the vicinity of the boundary. Using this nesting procedure ensures that this error is minimized and smooth velocity distributions exist (i.e. lacking in eddies, re-circulation and excessive speed or shear) across the boundary.

Figure 5.2.2 : Elevation from ‘super’ grid and global tide model : comparison to observation.

The major tidal constituents and their amplitudes at the boundary locations illustrated in Figure 5.2.3 that were derived from the super grid are presented in Table 5.2.1. Note that these constituent’s amplitude and phase vary spatially around the open boundary perimeter.

Table 5.2.1 : Tidal Harmonics for the Regional Model

Tidal Constituent Northern Boundary Offshore Boundary Southern Boundary Name Amplitude (m) Amplitude (m) Amplitude (m) M2 1.230 0.684 0.891 S2 0.449 0.241 0.306 K1 0.322 0.242 0.233 O1 0.153 0.119 0.110 S1 0.002 0.002 0.002 Q1 0.034 0.026 0.023 P1 0.103 0.078 0.076 N2 0.258 0.151 0.188 NU2 0.055 0.029 0.037 K2 0.126 0.065 0.081 L2 0.044 0.019 0.023 2N2 0.046 0.024 0.027 MU2 0.013 0.023 0.027 T2 0.026 0.014 0.018

It is observed from Table 5.2.1 that the dominant constituents in the region are those due to M2 and S2, hence the tide possesses a semi-diurnal character. It is also noted that the

12 amplitude of these constituents is quite large. The semi-diurnal constituent N2 is also quite large. Of the diurnal constituents K1 and O1 dominate.

Figure 5.2.3 : Locations of tidal constituents in Table 5.2.2.

The long period component was extracted from the low passed Burnett Heads record observations (Figure 5.2.4), which is located on the southern boundary of the regional domain. The tidal component from the super grid were then superimposed and the resulting sea level was applied on the southern boundary of the regional domain. The long period components at Burnett Heads is applicable to the coast only, and an offshore profile was imposed on the amplitude to correctly specify the long period wave over the shelf. The offshore extent of this wave was treated as a calibratable parameter which was tuned to provide the best match of modelled sea level to observation at South Trees. Modelled surface elevations are compared to those measured at South Trees in Figure 5.2.5 from which it is observed that agreement is good, i.e. the model reproduces sea level well in the Port Curtis region.

Figure 5.2.4 : Low Frequency sea level at Burnett Heads

13 Figure 5.2.5 : Segment of Surface Elevation at South Trees : measured (black) and modelled (red) (a) Tide and low frequency

(b) Low frequency

Elevations on the offshore open boundary of the Port Curtis domain were subsequently forced with output of the regional domain. Obviously these elevation signals contained both the diurnal and long period fluctuations.

5.3 Temperature and Salinity.

The temperature and salinity (TS) distributions used as initial conditions in the regional model were obtained from output of the global circulation model ACOM3 (Australian Community Ocean Model, (Schiller (2003)). The resolution of this model is ½ x 1/3o which is relatively coarse but constitutes the best three-dimensional TS distributions available. The temperature and salinity on open boundaries of the regional domain were also forced with ACOM output. The temperature and salinity distributions interpolated onto the regional domain from ACOM output for 1 January 1999 (initial condition) are presented in Figures 5.3.1 and 5.3.2. The regional model is relaxed to ACOM on a time scale of 10 days so as to implicitly include surface fluxes (e.g. changes in sea surface temperature due to atmospheric heating and cooling) and any motion resulting from temperature and salinity distributions generated on scales much larger than the regional model can capture (i.e. baroclinic basin scale phenomena).

14 Figure 5.3.1 : Temperature Distribution at 1 Jan 1999

Figure 5.3.2 : Salinity Distribution at 1 Jan 1999

The Port Curtis temperature & salinity initial conditions and boundary forcing are provided from output of the regional model. No surface heat or moisture fluxes are included in the local model at this stage, as calibration to temperature and salinity was not attempted. It is unclear what effect these fluxes may have on temperature and salinity.

5.4 River Flow.

River flow records were obtained for the Calliope River at Castlehope and Fitzroy River at the Gap. The former flow was input at Gladstone in the Port Curtis model and regional models and the latter input into the regional model only. The Fitzroy River maintains influence on the Port Curtis region via flow through the Narrows, and output from the regional grid at the Narrows was used as an open boundary condition in the Port Curtis model. Due to resolution restrictions, the representation of the Narrows in the regional grid was wider than in reality, however, the depth was adjusted to maintain correct cross-sectional area. The modelled salinity of water entering the Narrows should be subject to validation if future calibration to temperature / salinity is undertaken. The Boyne River was omitted due to the presence of the Awoonga Dam reducing flows to negligible levels. Daily flows were supplied and are displayed for the first three months of 1999 in Figure 5.4.1. Due to the close proximity of the Calliope River open boundary to marine water in the model, in conjunction with the large tidal excursion and well mixed nature of the estuary, the salinity of this inflow was not assumed to be fresh and was set at 20 psu. This value was based on experience obtained from similar

15 modelling exercises, extrapolated to a well mixed environment, and is not expected to be a highly accurate representation. However, no attempt is made in this study to calibrate to temperature and salinity, and the baroclinic circulation resulting from this freshwater inflow is expected to be small in comparison to tidal circulation, hence the salinity of this input is of secondary importance to the volume (barotropic contribution). The salinity of this inflow should be explicitly measured prior to any future calibration to temperature and salinity. The temperature of the Calliope River was assumed to be equal to the low pass filtered air temperature at Gladstone (Figure 5.4.2). Flow for the Fitzroy is displayed in Figure 5.4.3.

Figure 5.4.1 : Calliope River Flow

Figure 5.4.2 : Air Temperature at Gladstone

Figure 5.4.3 : Fitzroy River Flow

16 6. Model Output.

6.2 General Solutions The circulation in the Port Curtis region is dominated by the tide, with surface currents flowing into the estuary on the flood tide, primarily through North and Gatcombe Channels (Figure 6.1.1). Flow is also directed into Rodds Bay from offshore. Currents are significantly weaker (up to a factor of 2) during the neap flood tide in comparison to the spring flood tide. Large elevation differences exist between locations within North Channel and seaward of Facing Island during times of peak flow (greater than 25cm) and a consistent sea level gradient exists along the length of the estuary. The circulation during the peak ebb flow is the reverse of the flood tide (Figure 6.1.2). Flow is directed along the estuary and exits through the Facing Island Channels. The waters in Rodds Bay flow offshore. Sea level is highest in the Narrows and decreases constantly along the estuary. The neap ebb tide currents are again weaker than the spring tide currents.

Figure 6.1.1 : Surface currents at flood tide (ms-1) (a) Spring tide (b) Neap Tide

Depth averaged and bottom currents distributions are very similar to the surface currents (Figures 6.1.3 to 6.1.6), indicating that momentum is quite well mixed vertically and there exists little vertical structure to the three-dimensional flow. The large barotropic tide generates large currents, and combined with the shallow bathymetry large bottom friction results which generates a well mixed water column.

17 Figure 6.1.2 : Surface currents at ebb tide (ms-1) (a) Spring tide (b) Neap Tide

Figure 6.1.3 : Depth averaged currents at flood tide (ms-1) (a) Spring tide (b) Neap Tide

18 Figure 6.1.4 : Depth averaged currents at ebb tide (ms-1) (a) Spring tide (b) Neap Tide

151o 10 / E 151o 20 / E 151o 30 / E 151o 40 / E

23o 40 / S 23o 40 / S

Current 1 ms−1

23o 50 / S 23o 50 / S

24o S 24o S 1300 19 Jan 1999 +10

Sea−Level (m)

0.2 0.55 0.9

151o 10 / E 151o 20 / E 151o 30 / E 151o 40 / E

Figure 6.1.5 : Bottom currents at flood tide (ms-1) (a) Spring tide (b) Neap Tide

19 Figure 6.1.6 : Bottom currents at ebb tide (ms-1) (a) Spring tide (b) Neap Tide

Temperature and salinity have not been subjected to calibration and therefore are not expected to accurately reflect the thermohaline structure within the region. Although the low- pass filtered air temperature is a reasonable approximation for river temperature, more confidence could be applied to solutions using measured data. Similarly, the river salinity was fixed at 20 psu whereas measured river salinity should preferably be applied. Surface heat and water fluxes were not imposed and in the absence of good temporal and spatial measurements of temperature and salinity it is unclear how important these processes are in establishing the water properties in the estuary. However, the solutions do provide qualitative insight into the advection and diffusion characteristics of the estuary.

Temperature and salinity solutions for the flood and ebb tide are displayed in Figures 6.1.7 – 6.1.10. The temperature and salinity distributions are generally well mixed in the estuary and Rodds Bay with less than 0.5oC and 0.05psu horizontal variability throughout the domain. The exception is in Port Curtis estuary where the Calliope River, and to a lesser extent the Narrows, act to cool and freshen the estuary. The salinity of the Calliope input is significantly lower than that of the estuary and this acts to significantly lower salinity in the estuary. The large tidally driven advection and diffusion in this region distribute the low salinity river water over much of the estuary. Offshore the salinity distribution is uniform and temperature structure reflects the mixing regime, where cooler SST is found in regions where a deep mixed layer mixes cool subsurface water to the surface. An interesting phenomenon is evident in North Channel, where plumes of warmer/fresher estuarine water are advected through the strait on the ebb tide to create a distinct plume. As the tide turns and water converges into the strait the plume is pinched off to create a fresh pool which is advected with the ambient flow. This occurs on every tidal cycle to create an apparent ‘puffer’ effect of any tracer through the strait. The effect is more evident during spring tides.

Vertical diffusivity for the flood and ebb tide is displayed in Figures 6.1.11 and 6.1.12 respectively. These mixing coefficients were generated using the Mellor-Yamada level 2.0 mixing scheme (Mellor and Yamada, 1982). Very large mixing is evident in Port Curtis estuary, with the flood tide generating stronger mixing than the ebb and the spring tide greater than the neap. The south-eastern and north-western tips of Facing Island are regions of particularly strong vertical mixing. The offshore regions, particularly towards the south-east, also are exposed to strong mixing.

20 Figure 6.1.7 : Surface temperature at flood tide (oC) (a) Spring tide (b) Neap Tide

Figure 6.1.8 : Surface temperature at ebb tide (oC) (a) Spring tide (b) Neap Tide

21 Figure 6.1.9 : Surface salinity at flood tide (psu) (a) Spring tide (b) Neap Tide

151o 10 / E 151o 20 / E 151o 30 / E 151o 40 / E

23o 40 / S 23o 40 / S

23o 50 / S 23o 50 / S

24o S 24o S 0700 19 Jan 1999 +10

Salinity (psu)

34.7 34.88 35.05

151o 10 / E 151o 20 / E 151o 30 / E 151o 40 / E

Figure 6.1.10 : Surface salinity at ebb tide (psu) (a) Spring tide (b) Neap Tide

22 Figure 6.1.11 : Surface vertical diffusivity at flood tide (m2s-1) (a) Spring tide (b) Neap Tide

Figure 6.1.12 : Surface vertical diffusivity at ebb tide (m2s-1) (a) Spring tide (b) Neap Tide

23 6.2 Residual (Net) Currents

The transport of material throughout the Port Curtis Estuary is achieved through advective and diffusive processes. Although the currents generated by tidal forcing are large, they are periodic by nature and only result in net flow through non-linear interactions with topography. It is the net flow that is important from a flushing perspective, since transport by this flow provides a mechanism that may potentially remove material permanently from the estuary. Mechanisms that may contribute to this residual circulation are:

1. Rectified tidal currents 2. Low frequency barotropic currents 3. Wind 4. Thermohaline circulation

Four model scenarios were conducted to assess the impact of various forcing mechanisms on the residual circulation in Port Curtis Estuary. These scenarios are:

1. Forced with tide only 2. Forced with tide and low frequency sea level 3. Forced with tide, low frequency sea level and wind 4. Forced with tide, low frequency sea level, wind and TS gradients

Mean currents were obtained by obtaining a running average of the model velocity components at every time-step. If the velocities are oscillatory, this will bias the mean depending on the phase when the running mean is terminated. Since the tidal velocities are large in Port Curtis, this does result in some oscillation of the mean currents over a tidal cycle, but these oscillations are less than 1mms-1 (Figure 6.2.1) and therefore introduce negligible error to the mean. The results presented below are for the first month of 1999, and residual currents are expected to alter if any of the mechanisms outlined above change. The residual currents for scenario 1-4 are displayed in Figure 6.2.2 – 6.2.5 respectively.

Figure 6.2.1 : Time variation of averaged currents in Mid-Estuary (U1 & U2 are the velocity components)

24 Figure 6.2.2 : Mean Surface Currents for Tide Forced Currents only (Scenario 1). The right panel depicts Port Curtis region at higher resolution.

Figure 6.2.3 : Mean Surface Currents for Tide and Low Frequency Forcing (Scenario 2).

25 Figure 6.2.4 : Mean Surface Currents for Tide, Low Frequency Forcing and Wind (Scenario 3)

Figure 6.2.5 : Mean Surface Currents for Tide, Low Frequency Forcing, Wind and Temperature & Salinity (Scenario 4).

Generally, current speeds are less than 10cms-1, except in North Channel. The flow enters Port Curtis Estuary through Gatcombe Channel, and exits through North Channel. The currents are generally directed up the estuary; in the region between the Calliope River input and the Narrows, the flow degenerates into a series of gyres. A transition zone between the

26 up-estuary flow and the gyres exists where the flow is somewhat chaotic. Mean flow is quite small near the narrows. An anticyclonic (counter-clockwise) gyre also exists in the mouth of Rodds Bay. There exists little variation between the scenarios in the estuary, indicating that the residual flow in this region is predominantly the result of the tide, which is generated by non-linear interactions of the tide with topography (topographic rectification; Robinson, 1983). Offshore the tide appears to induce a large, relatively weak, cyclonic eddy (Figure 6.2.2) which strengthens slightly upon the inclusion of low frequency sea level and atmospheric pressure (Figure 6.2.3). Offshore flow increases again with the inclusion of wind (Figure 6.2.4) and the eddy is no longer visible with currents directed towards the north-west, parallel to the coast. The inclusion of thermohaline forcing does not alter this pattern (Figure 6.2.5). Note that the net flow pattern may change if any of the forcing mechanisms become radically different to those used in this analysis. Large flood events may also impact on net flow.

From a flushing perspective, the residual currents act to advect material into the estuary rather than out. Since the scenarios indicate that that the residual currents in the estuary are relatively insensitive to wind direction and strength, flushing is not expected to improve substantially in the presence of persistently favourable wind. Topographically rectified tidal currents are the result of the generation of topographically induced vorticity, thus any change to the bathymetry may impact upon the mean flow. Freshwater inputs do not appear to contribute to residual thermohaline circulation, owing to the strong tidal mixing preventing the establishment of large baroclinic pressure gradients.

6.3 Flushing Characteristics

Passive tracers were used to obtain an estimate of the flushing characteristics of the estuary. A passive tracer was initialised in a sub-region of the estuary (Figure 6.3.1) with a concentration of 1 and zero elsewhere, and the total mass in this sub-region was calculated throughout the simulation. Full forcing was applied to the domain (i.e. wind, tide, low frequency sea level and temperatre / salinity effects). The e-folding time for flushing this sub- region is encountered when the total mass was reduced to 1/e (~38%) of the initial mass. A time series of the normalized total mass in the sub-region is displayed in Figure 6.3.2, from which it can be seen that total mass oscillates on the tidal frequency as tracer is brought into the domain on the flood and removed on the ebb. The general trend of tracer decrease is obtained by fitting a curve to the total mass, from which it can be seen that the e-folding time for this sub-region is approximately 19 days. The passive tracer distribution at the end of the simulation is shown in Figure 6.3.3 (note that this is a snapshot and varies considerably with the phase of the tide).

Figure 6.3.1 : Distribution of passive tracer in a sub-region of the Port Curtis Estuary (yellow region) for flushing estimation.

27

Figure 6.3.2 : Normalized total mass for the Port Curtis sub-region

Figure 6.3.3 : Passive tracer at day 31

It can be seen that at certain locations in the estuary (particularly between the Calliope River and Fisherman’s Landing) the concentration of passive tracer remains at 33% of the original concentration after 1 month. This period encompasses approximately two complete neap- spring tidal cycles, but note that under different forcing conditions (e.g. tide, wind or river flow) the flushing characteristics are expected to alter.

The flushing characteristics of Rodds Bay were investigated by defining a sub-region as depicted in Figure 6.3.4.

28 Figure 6.3.4 : Distribution of passive tracer in a sub-region of Rodds Bay (yellow region) for flushing estimation.

The time series of the normalized total mass in the Rodds Bay sub-region is displayed in Figure 6.3.5, from which it can be seen that the e-folding time for this bay is approximately 5 days. The tracer distribution after one month is displayed as Figure 6.3.6. Relatively low concentrations (<5%) are found throughout much of the domain. Tracer exhibits the tendency to be transported north-westwards along the coast, to enter Port Curtis estuary through Gatcombe Channel and exit through North Channel. Higher concentration tracer persists in the most inshore reaches of Rodds Bay. Overall the flushing characteristics of Rodds Bay appear quite good.

Figure 6.3.5 : Normalized total mass for the Rodds Bay sub-region

29 Figure 6.3.6 : Passive tracer at day 31

30 6.4 Transport Characteristics

Point sources of tracers were continuously input to the surface at Gladstone and Fisherman’s Landing (Figure 6.4.1) with unit loads (assumed to be 1 gs-1 ~ 31,500 kg/year, giving output concentrations in units of gm-3, or mgL-1) for a one-month period. This period encompasses two neap-spring cycles; ideally a longer time period should be employed to allow the tracers to reach a quasi-steady state, preferrably seasonal or annual simulations. Surface tracer concentrations were output at 1 hour intervals and post-processed to compute the 5th, 50th (median) and 95th percentile distributions for the whole simulation, providing a statistical description of the distributions resulting from tracer transport over this period. Note the response of the tracers to the interaction of the source input with the system dynamics is linear, so that if the load were increased by some arbitrary factor then the corresponding concentrations can be scaled accordingly.

Results are displayed as Figure 6.4.2 for Gladstone and 6.4.3 for Fisherman’s Landing. Results are interpreted thus: given that a continuous unit load is input at Gladstone and its distribution throughout the domain allowed to reach quasi-steady state, at any given location in the domain one would expect to find the concentrations less than those shown in Figure 6.4.2 (a) for 5% of the time, less than those in Figure 6.4.2 (b) for 50% of the time and less than those in Figure 6.4.2 (c) for 95% of the time. Note that the concentration scales in the figures for the three percentiles differ from one another. These percentiles were calculated for the whole simulation period, including the spin up period before quasi-steady state was achieved, hence the results represent distributions expected from the time of input until quasi- steady state is achieved, (i.e. if material is input at a certain time and we wait until quasi steady state, then we can expect the following percentile distributions). Resulting distributions for both releases are confined to the estuary, with negligible concentrations beyond Facing Island. The Gladstone releases generally result in higher concentrations throughout the lower estuary and inshore of Facing Island, with maximum concentrations centered around Gladstone. The Fisherman’s Landing release remains largely centered around that source, with the higher percentile distributions resulting in a plume of high concentration exiting the estuary through Auckland Channel. The Gladstone release has concentrations between 0.001 and 0.002 mgL-1 throughout much of the estuary – note that if the input were in terms of nitrogen then resulting concentrations in the water column from this unit input alone are significant. Median concentrations from the Gladstone release in Rodds Bay remain less than 1x10-10 mgL-1 (Figure 6.4.4), thus input loads at Gladstone would need to be of the order of 3.15 x1011 kg/year for concentrations in this region to exceed 0.001 mgL- 1.

Figure 6.4.1 : Estuary point source release locations

151o 10 / E 151o 15 / E 151o 20 / E

23o 45 / S 23o 45 / S

Fishermans

Gladstone 23o 50 / S 23o 50 / S Auckland Channel

151o 10 / E 151o 15 / E 151o 20 / E

31 Figure 6.4.2 (a) : Gladstone release 5 percentile distribution

Figure 6.4.2 (b) : Gladstone release median distribution

Figure 6.4.2 (c) : Gladstone release 95 percentile distribution

32 Figure 6.4.3 (a) : Fisherman’s Landing release 5 percentile distribution

Figure 6.4.3 (b) : Fisherman’s Landing release median distribution

Figure 6.4.3 (c) : Fisherman’s Landing 95 percentile distribution

33 Figure 6.4.4 Gladstone release median distribution at Rodds Bay

A further two point sources were released with unit loads at locations outside the estuary depicted in Figure 6.4.5. The spoil site is the location where the dredging spoils from Port Curtis estuary are deposited. The tracer release was at 10m depth in this instance (bottom depth is ~15m). The Rodds Bay location, where tracer is input is at the surface, is intended to be indicative of tracer transport within Rodds Bay. Percentile distributions for the spoil site and Rodds Bay are displayed in Figures 6.4.6 and 6.4.7 respectively. It should be noted that these are surface distributions, even though the source for the spoil site was located at 10m depth.

Figure 6.4.5 : Point source release locations

34 Figure 6.4.6 (a) : Spoil site release 5 percentile distribution

Figure 6.4.6 (b) : Spoil site release median distribution

Figure 6.4.6 (c) : Spoil site release 95 percentile distribution

35 Figure 6.4.7 (a) : Rodds Bay release 5 percentile distribution

Figure 6.4.7 (b) : Rodds Bay release median distribution

Figure 6.4.7 (c) : Rodds Bay release 95 percentile distribution

36

Tracer distributions for the spoil site show a tendency for tracer to be advected north- westwards along the seaward side of Facing Island (with the mean flow). There exists negligible tracer south-east of the spoil site. The 5 percentile distribution exhibits a large pool of maximum concentrations at a location north-westwards of the source in addition to around the source itself. This is due to the rotational nature of the tidal currents causing tracer to pinch off at the source and advect north-westwards with the mean flow in the form of a distinct pool, which subsequently diffuses to negligible concentrations within the next tidal cycle. This pool is not evident in the median concentrations, where the tracer forms a plume extending north-westwards with maximum concentration near the source. Little tracer reaches Port Curtis estuary or south-eastwards of the source. The 95 percentile distribution is similar in spatial extent to the median, except maximum concentrations are located in closer proximity to the source, and extend further towards the south-east. Generally, the greatest far-field concentrations from the spoil site are found on the seaward side of Facing Island. As a location for depositing dredging spoils removed from the estuary, the spoil site appears well located since very little material is transported back into the estuary. Note that this supposition is based on the assumption that the fine sediment associated with the spoil sinks slowly enough that it is treated as a neutrally buoyant tracer.

The Rodds Bay percentile distributions reveal the tendency for tracer to be transported directly south towards the region between and Norton Point on Hummock Hill Island and Becher Point. The 5 percentile distribution is confined to the area between Innes Head and Becher Point with concentrations less than 6x10-4 mgL-1. The median distribution enters Seven Mile Creek and Boyne Creek, and also propagates north-westwards along the seaward side of Hummock Hill Island with concentrations less than 1.5x10-3 mgL-1. The 95 percentile distribution is similar to the median, with higher concentrations found opposite Rodds Peninsula and small amounts of tracer entering Port Curtis estuary. The 95 percentile concentrations are generally less than 2x10-3 mgL-1. Little tracer is transported around Rodds Peninsula to offshore waters. The region most impacted by tracer originating from Rodds Bay appears to be the Seven Mile Creek region and the coastal margins around Middle Head, Hummock Hill Island, Wild Cattle Island and to a lesser extent Boyne Island, with small amounts of tracer entering Port Curtis. Tracer transport is generally north-westwards along the coast. Given this transport tendency it is unlikely that tracer released north-westwards of Rodds Bay would be transported into the Bay. .

37 6.5 Connectivity

The connectivity of the domain can be examined by observing the behaviour of neutrally buoyant particles released at particular locations. Particles were released along a vertical plane located near Fisherman’s Landing, where the plane extended horizontally from start to end locations shown in Figure 6.5.1 and vertically from the surface to the bottom (4m). The particles were released from random locations in this plane at a rate of 2 particles / hour from an initial pool of 10,000 particles. These particles were subsequently advected with the circulation to provide insight into how various regions of the domain are connected. The particles are also subjected to random motion representing the effects of diffusion (i.e. sub- grid scale effects). Therefore, any two particles released from the same place at the same time are expected to undergo different trajectories due to this random motion. When a particle crosses the offshore open boundary of the model domain it is placed in the initial pool for re- release. The particle distribution after approximately one neap-spring cycle is shown in Figure 6.5.2, and at the end of the simulation in Figure 6.5.3. This distribution is the projection of particles at all depths onto the surface.

Figure 6.5.1 : Particle release locations

The even distribution of particles present throughout the estuary indicates that the estuary itself is well connected. More particles appear to exit the estuary through North Channel than Gatcombe Channel, consistent with the residual current distribution. Few particles are observed beyond Facing Island, indicative of the relatively poor connectivity between the upper estuary and offshore waters.

Due to the large number of particles in the domain, an animation of the particle trajectories best conveys the connectivity of the region, although observation of isolated particle trajectories does supply insight into the dynamics of the system. The predominant tidal constituents are M2 and S2, meaning that the tide is predominantly of semi-diurnal character. Particle trajectories are therefore expected to undergo two along-estuary excursions per day (i.e. one every 6 hours), with the net displacement of start and end locations being indicative of the residual flow. Trajectories for particles 1 & 2 (released in the 1st hour corresponding to the ebb of a spring tide) and 12 & 13 (released in the 6th hour; flood of spring tide) are displayed in Figures 6.5.4 and 6.5.5 respectively. Particle trajectories are plotted for 48-hour blocks over a neap-spring cycle (see Figure 5.2.5 (a) for a typical neap-spring cycle). These trajectories clearly show the large tidal excursion associated with the strong tidal currents. Although the excursion is large, net movement is usually small, with the particle returning to a location near its original position. Tidal excursion can be as large as 15km (e.g. Figure 6.5.5 (a), red trajectory, hours 35 – 41) with only ~3km net displacement over a 12-hour period. At times, net displacement can be less than 2km over a 48-hour period (e.g. Figure 6.5.4 (b), red

38 particle). The tendency appears for particles to slowly progress down estuary along the mainland side of the estuary and exit offshore through North Channel.

Figure 6.5.2 : Particle distribution after 1 neap-spring cycle

Figure 6.5.3 : Particle distribution after 2 neap-spring cycle

39 Figure 6.5.3 : Trajectories for particles 1 and 2 (released during the ebb of a spring tide) (a) Day 0 – 2 (b) Day 2 – 4

151o 10 / E 151o 15 / E 151o 20 / E

Start End 23o 45 / S

23o 50 / S

0km4

151o 10 / E 151o 15 / E 151o 20 / E

(c) Day 4 – 6 (d) Day 6 – 8

40 (e) Day 8 – 10 (f) Day 10 – 12

o / o / o / 151 10 E 151 15 E 151 20 E 151o 10 / E 151o 15 / E 151o 20 / E

Start Start o / End End 23 45 S S 23o 45 / S

o / 23 50 S S 23o 50 / S

DepthDepth Range −7000 to 10m10m DepthDepth Range −7000 to 10m10m o / o / o / 151 10 E 151 15 E 151 20 E 151o 10 / E 151o 15 / E 151o 20 / E

(g) Day 12 – 14 (h) Day 14 – 16 Note : when particles leave the model domain through the open boundary they are returned to the source, hence the blue trajectory over land.

41 Figure 6.5.4 : Trajectories for particles 10 and 11 (released during the flood of a spring tide) (a) Day 0 – 2 (b) Day 2 – 4

(c) Day 4 – 6 (d) Day 6 – 8

42 (e) Day 8 – 10 (f) Day 10 – 12

151o 10 / E 151o 15 / E 151o 20 / E

Start End S 23o 45 / S

S 23o 50 / S

DepthDepth Range −7000 to 10m10m

151o 10 / E 151o 15 / E 151o 20 / E

(g) Day 12 – 14 (h) Day 14 – 16

43 7. Benefits and Outcomes

This study provides an overview of the circulation characteristics within the Port Curtis estuary and the interaction of the estuarine circulation with the offshore environment. Various regions of stakeholder interest (e.g. spoil site, Rodds Bay) have been investigated to provide insight into transport characteristics of potential contaminants. The primary objective of developing a pilot hydrodynamic model of the region has been realized. In addition, calibration to sea level and analyses of tracer distribution, flushing rates, residual circulation, connectivity and the general oceanography of the region have been provided. These outcomes provide a first order overview of the system that may aid in management decisions. In particular, knowledge of the flushing rates of the system combined with estimates of contaminant input will provide an estimate of how fast introduced contaminant would be removed from the harbour. The predictions of tracer trajectories from hydrodynamic modelling allow the assessment of where contaminants introduced at a point source are likely to go. Although not part of the hydrodynamic model development, currents and mixing are the main determinants of resuspension rates and transport of deposited sediments. Models of fine sediment transport from dredge spoil sites for instance would need to be based on a quantitative understanding of the hydrodynamics of the harbour. Likewise the transport and fate of nutrients and consequent primary production within the harbour can be simulated using biogeochemical models. Transport and mixing of nutrients and phytoplankton require knowledge of the hydrodynamics. Full application of the hydrodynamic model should be undertaken after calibration and validation are completed.

8. Limitations and Further Development

The hydrodynamic model is currently at a pilot stage and requires further calibration and validation in order to achieve full confidence in solutions. Currently, the model has only been calibrated to sea level, where good agreement was observed between modelled and measured elevation. Since the estuary is tidally dominated this implies that the currents are reasonably accurate. Also, as the residual flow appears to be dominated by topographically rectified tidal currents, the satisfactory elevation calibration implies that conclusions based on tracer distributions and flushing can be considered at least first-order accurate. However, the non-linear nature of the residual generation has the potential to amplify error and the residual circulation should be verified using field measurements, e.g. conservative tracers such as salinity after a flood event or conservative contaminants.

There has been no attempt to calibrate the model to temperature and salinity, and this should be undertaken so as to gain confidence in the predicted distribution of these variables. Temperature and salinity distributions are generally established in response to the larger scale circulation and modulated by local effects, therefore if good agreement is achieved between mean modelled and measured TS then confidence in broad circulation patterns is achieved. Local forcing conditions (e.g. wind) are, however, capable of altering velocity distributions on short time and space scales, thus the local circulation near point sources can only be expected to be well represented if the local forcing conditions are accurate. This is hardly ever the case since measured atmospheric variables (i.e. wind) are usually only available at a few locations in the domain and interpolated such that any local variations are smoothed over. Therefore, if velocity data (e.g. ADCP) are collected for calibration purposes, then the local forcing conditions should be attempted to be measured and represented in the model.

Temperature and salinity are cheaper and easier to measure than velocity, hence further calibration is recommended to be undertaken using Temperature/Salinity chains gathering high temporal resolution data at a few key locations, supplemented with coarser temporal transects with high spatial resolution. TS measurements should also be collected at the location on the Calliope River corresponding to the modelled Calliope flow for forcing this inflow. Also, TS data in the Narrows should be collected to verify the nesting procedure in this region. Calibration to TS may require the addition of surface heat and salt fluxes into the model, hence measurements of relevant atmospheric variables in the study region (wet and dry bulb air temperature, pressure, wind, cloud cover, global shortwave irradiance,

44 precipitation) may be necessary. Higher spatial resolution wind measurements would improve confidence in results. Since a good calibration to sea level has been achieved with the model, further elevation measurements are not considered necessary.

It should be noted that the analyses presented in this study are based on simulations of one- month duration only, and circulation patterns are expected to change at other times of the year in response to altered climatic conditions (e.g. floods, storm surges, wind events). The extent to which seasonal influences impact the model solutions is unknown, hence the conclusions drawn in this study cannot be applied universally without further model application during different timeframes. The current state of the model is capable of performing annual simulations to address this issue.

The inclusion of the intertidal areas is also recognized as possibly needed with the model, and it is anticipated that any further application of the model would attempt to include wetting and drying of these regions. Given these limitations the model calibrates well against sea level and provides some useful insight into the hydrodynamics of the region.

45 9. Conclusions

A 3D primitive equation model was applied to Port Curtis estuary 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 high resolution restricted the lengths of simulations to several neap-spring tidal cycles, and therefore attention was focussed on dynamics having these time-scales. The first month of 1999 were simulated.

The model results confirm that this region is tidally dominated. Large semi-diurnal tidal amplitudes of up to 2m, predominantly due to the M2 and S2 tidal constituents, result in large currents that may reach maximum velocities of 2ms-1 through North Channel. Flow is directed up the estuary towards the Narrows on the flood tide, and down the estuary on the ebb. Surface elevation undergoes a neap-spring tidal cycle with a period of approximately 14 days. The tidal wave propagates through the Narrows both from Port Curtis and Keppel Bay to meet midway along the Narrows resulting in small currents there. The water column is well mixed in Port Curtis estuary, with negligible vertical gradients of momentum or density. This is attributed to large bottom stress resulting from the strong tidal flow.

Although the currents are large in the estuary, the residual (net) flow appears small. Residual flow inshore can be attributed to topographic rectification resulting from the non-linear interaction of tides with the bathymetry, while offshore the residual flow is predominantly wind driven with a small contribution from the barotropic gradients. Residual currents are directed up-estuary in the lower estuary, and formed a series of small gyres in the upper estuary. Offshore, flow was directed along-shore towards the north-west, and enters the estuary through Gatcombe Channel to exit through North Channel. Magnitudes of the residual currents were less than 0.1 ms-1. These residual currents were based on forcing for January 1999, and under different forcing conditions the character of the mean flow may alter.

The Port Curtis estuary region appears to be well connected throughout, however the estuary is poorly connected with the offshore region seaward of Facing Island. This is evident in flushing, passive tracer and particle analyses. Tracers are transported efficiently throughout the estuary but inefficiently transported out of the estuary to offshore regions. The e-folding flushing time for the estuary is of the order of 19 days in January 1999. This estimate is not expected to dramatically alter seasonally if tidally driven flow dominates the circulation. This flushing estimate is applicable to the estuary as a whole, and due to the well connected nature of the estuary any smaller sub-region within the estuary is expected to have significantly shorter flushing times. In contrast to Port Curtis estuary, Rodds Bay is well flushed and well connected to offshore regions, with an e-folding flushing time of 5 days in January 1999.

The input of unit loads into Port Curtis estuary results in appreciable concentrations within the estuary and negligible concentrations seaward of Facing Island. Unit inputs of 1 gs-1 into the estuary would result in significant median concentrations if the input were considered to be a limiting nutrient for primary production. Input into Rodds Bay results in far smaller concentrations around the point source, with tracer restricted to the areas south of the source. Release at the dredging spoil site results in tracer distributions forming a plume originating from the source and directed north-westwards along the seaward coast of Facing Island. The prevailing wind conditions are likely to influence distributions offshore, thus seasonal variability is expected.

Although the model has been calibrated to sea level, additional calibration to temperature and salinity is required to provide greater confidence in the prediction of broad scale circulation within the regional domain. The model has only been simulated for a relatively short period in the summer of 1999 and a broader range of climatic conditions should be simulated to gain greater confidence in the statistical descriptions of tracer distribution and residual circulation. An attempt to better parameterize the inter-tidal areas should also be undertaken.

46 10. References

Blumberg, A.F. and Herring, J., (1987) Circulation modelling using orthogonal curvilinear coordinates, in Three-Dimensional Models of marine and Estuarine Dynamics, eds. Nihoul, J.C.J and Jamart, B.M., Elsevier.

Eanes, R. and S. Bettadpur (1995) The CSR 3.0 global ocean tide model. Center for Space Research, Technical Memorandum, CST-TM-95-06.

Leonard, B.P. (1979) A stable and accurate convective modelling procedure based on quadratic upstream interpolation. Computer methods in applied Mech. and Eng., 19, 59 – 98.

Leonard, B.P. (1991) The ULTIMATE conservative difference scheme applied to unsteady one-dimensional advection. Comp. Methods in Appl. Mech. and Eng., 19, 17 – 74.

Mellor, G.L. and T. Yamada (1982) Development of a turbulence closure model for geophysical fluid problems. Rev. Geophysics and Space Phys., 20(4), 851-875.

Robinson, I.S. (1983) Physical oceanography of coastal and shelf seas. Elsevier Oceanograpgy Series 35 (B. Johns Ed.), pp 321-356.

Schiller, A. (2003) Effects of explicit tidal forcing in an OGCM on the water-mass structure and circulation in the Indonesian throughflow region. Ocean Modelling, 6, 31-49.

Walker, S.J. and J.R. Waring (1998) A multiple grid, 3-dimensional, non-linear, variable- density hydrodynamic model with curvilinear horizontal coordinates and level (z) vertical coordinates, CSIRO Marine Research, Report OMR-118/120.

Witt, C. and C. Morgan (1999) Stuart oil shale project, stage 2. EIS marine water quality and flow modeling. WBM Oceanics Australia Report 11774.R1.2

47