Storm Surge Atlas

The storm surge model used in this study is a depth-averaged ocean current model developed to simulate currents and sea surface elevations on continental shelves (Hubbert et al. 1990, 1991; Hubbert and McInnes, 1999a,b).  The model, known as GCOM2D, solves a set of fluid dynamical equations over a grid comprised of equally spaced points in an east-west and north-south direction over the region of interest.  Finer resolution grids have a greater concentration of grid points per unit area and therefore can resolve in greater detail the horizontal variation in parameters such as water depth, currents, topography and bathymetry.  GCOM2D is driven by wind stresses and atmospheric pressure gradients acting on the ocean surface, sea-level heights at its lateral boundaries due to tides and atmospheric conditions, and friction of the ocean floor (Hubbert et al. 1990, 1991; Hubbert and McInnes, 1999a, b). 

GCOM2D features a moveable coastal boundary that allows the flooding and draining (i.e. inundation) of the coastal terrain due to the storm tide to be simulated (Hubbert et al. 1990, 1991; Hubbert and McInnes, 1999a, b).  The GCOM2D runs over successively finer regions utilising the results of the lower resolution, outer simulations as boundary conditions.  This ‘nesting’ technique is an economical way of maximising grid resolution over the region (McInnes et al. 2003). In this study, model simulations are carried out over two regions.  The lower resolution simulations are conducted with a grid spacing of 1km over the entire region shown in Figure 1.  Simulations are then conducted over the smaller region centred on Cairns at 100m resolution shown in Figure 2.

Topography of the region was obtained from the AUSLIG GEODATA 9-second Digital Elevation Model (DEM) with grid spacing of nine seconds in longitude and latitude (approximately 250m resolution) (Auslig 1994).  It was enhanced in the Cairns region using data from the 20m DEM of Zerger (2002). 

Wind speed and pressure fields required to force the storm surge model are derived using the analytical wind profile model of Holland (1980).  Cyclone profiles are determined by specifying the central pressure, radius of maximum winds, cyclone track including the direction of approach, translational speed and location of crossing; and the wind profile shape.  The surface wind is then derived following the procedure described in Hubbert et al. (1991).   

Output stations were determined for the low resolution grid (1 km) and high resolution grid (100 m) where the output storm surge levels were required. Twelve stations along the Cairns coast were determined and are detailed in table 1.

Table 1: High resolution (100m) station output locations.

The proximity of the cyclone crossing locations were chosen to make landfall north of Cairns so as to place the most intense onshore winds in the vicinity of Cairns.  The three cyclone crossing locations are Yule Point (16.57S 145.51E), Redcliff Point (16.68S 145.58N) and Taylor Point (16.78S 145.69E) as show in Figure 3.   The tide chosen for the simulations was 1.64m Lowest Astronomical Tide (LAT) equating to 0m Australian Height Datum (AHD), which is the closest tide to Cairns Mean Sea Level (MSL) at 1.70m LAT (Maritime Safety Queensland 2006).

Figure 3. Crossing Locations of the tropical cyclones in this study.

Tropical cyclone parameters

From the results of a sensitivity analysis, the tropical cyclone parameters that were selected to be simulated in the GCOM2D inundation model are listed in Table 2

Table 2:  Cyclone profile parameters chosen for storm surge simulations

Spatial Analysis Methodology

For each tropical cyclone scenario simulation, the model GCOM2D calculates the maximum surge penetration which generates an output indicating the highest surge height for each grid cell (100m) relative to Australian Height Datum (AHD).  To map the inundation Arc-Info GIS was used to convert the model output data to a raster grid file of inundation.  The raster image was passed through a neighbourhood filter which calculates the mean with a 3 x 3 window.  This filter had the effect of removing extremes from the data, producing a smoother surface without compromising the original output data.  To define the inland extent of the storm surge inundation the filtered inundation raster was multiplied with a Cairns coast mask created from a 1:100 000 vector dataset which excluded all sea point height values.  The datasets used in this project are listed in Table 3

Table 3  Datasets used in spatial analysis