Potential for forecasting inundation from flooding and storm surge: a case study of a small urban catchment during ex Tropical Cyclone Oswald.

Takehiko Nose 1, Joanna M. Burston 1, Daniel Gee 1, and Rodger Tomlinson 1 1 Griffith Centre for Coastal Management, Griffith University, Southport, Qld, ; [email protected]

Abstract Forecasting inundation resulting from the dynamics of storm tide combining with freshwater flooding in estuarine environments and the behaviour of wave set-up in tidal inlets is complex yet important for effective disaster management. As part of a Griffith University project sponsored by the (Qld) State Government, a case study of a small urban catchment with a wave-dominated tidal inlet is used to investigate the potential for numerical modelling of such processes. Ex Tropical Cyclone (TC) Oswald, January 2013, generated freshwater flooding, storm surge and extreme wave conditions along eastern Qld coastline, causing extreme water levels in the case study site of Currumbin Creek on the Gold Coast. A field survey following this event combined with measured discharge, wave, and hydrographic data allows analysis of the hydrodynamic response of the system using numerical modelling. A numerical model built using DHI’s MIKE21 is used to simulate interactions between the riverine flood, storm surge and waves within the tidal inlet during the ex-TC Oswald event. In particular, the effect of possible wave set-up at the entrance on the tailwater level is investigated. We conclude that dynamical forecasting of inundation due to a combination of storm tide, extreme waves and freshwater flooding in such environments requires detailed inputs, including accurate discharge, tailwater levels and up-to-date bathymetry.

Keywords: numerical modelling, Currumbin Creek, wave setup, storm surge, tropical cyclones, coastal estuary

1. Introduction to investigate potential improvements to Qld’s Water levels in tidal inlets during extreme weather capabilities in forecasting storm surge inundation events exhibit complex behaviour due to the risk for emergency planning purposes. Here, a case number of and dynamic interaction between the study has been developed to investigate the physical processes operating. These processes complexities involved in forecasting storm tide include storm surge, elevated riverine flow in inundation in tidal inlet environments. The aim of response to extreme rainfall, astronomical tide and this study is to investigate the contribution of various possible wave set-up, as well as evolution of the processes to the extreme water levels measured morphology due to channel scouring and longshore from the field survey of Currumbin Creek carried out sediment. Understanding the response and after ex-TC Oswald through numerical modelling. sensitivity of extreme water levels for urbanised tidal inlets is important from an emergency management The Gold Coast was chosen as the case study perspective. Much of the risk posed by storm tide location given historical impacts of TCs and along the Gold Coast, Qld, is to properties with population exposure to flooding. The Gold Coast exposure to flooding from the combination of storm has been exposed to infrequent TC events including tide and riverine flooding in small urban catchments. a series in 1974 including TC Pam, TC Dinah However, the dynamics of storm tide combining with (1967), the 1954 cyclone and several events from freshwater flooding in such environments and the the 1880s, to 1930s [5]. These cyclonic events have behaviour of wave set-up in tidal inlets are not well caused major coastal impacts including flooding, understood. The current practise of flood engineers beach erosion, and barrier breakthrough. Since and disaster managers when assessing the risk of these major events, the Gold Coast has inundation in real-time is to apply a tailwater level experienced intensive development including consisting of the astronomical tide combined with a proliferation of canal estates along its urban static estimation of the peak storm surge from BoM waterways. warnings to a flood model (pers. comm., Gold Coast City Council (GCCC)). The present practise is to In January 2013, the Gold Coast was impacted subtract any allowance for wave set-up from the upon by ex-TC Oswald, with accompanying heavy tailwater level for Gold Coast catchments. rainfall, persistent east-north-east winds and associated storm surge and high wave conditions This study, part of the Griffith Centre for Coastal coinciding with spring tides (Section 3) causing Management (GCCM)’s “Coastal Emergency severe beach erosion and some localised Management Decision Support System” project, an inundation. While this event was not particularly initiative of the Qld State Government in partnership severe, it underlined the vulnerability of the Gold with DHI Australia, Qld Cyber Infrastructure Coast to TC events and provided an opportunity to Foundation and Emergency Management Qld, aims investigate the physical processes contributing to oceanic inundation. GCCM staff conducted a field recorded around 09:00 28 th January (all times given survey of the open beaches and of the inundation in AEST (+1000 UTC)). around the small urban catchment of Currumbin Creek (Figure 1).

2. Study Area Currumbin Creek (28.135 °S, 153.465 °E) is located between Currumbin Beach to the south and Palm Beach to the north on the Gold Coast of eastern Australia and has a catchment area of 48 km2 (Figure 1). The creek entrance can be classified as a tidal inlet rather than an estuary given the generally low freshwater discharge, and is trained with rock training walls on both the northern and southern sides. The southern training wall connects to Currumbin Rock. Along with annual maintenance Figure 2. Synoptic chart showing the position of ex-TC dredging of the entrance, these walls serve to keep Oswald, 11:00 AEST 28 th January 2013 (source: BoM). the entrance open to the sea, improving water quality within the estuary and mitigating flood impacts [7]. The entrance morphology is dynamic in response to longshore transport from the south and episodic ocean swell events [6]. The Currumbin Creek area has high ecological significance and is a popular recreational destination for surfers, boating and fishing.

Figure 3. Measured wind (a) speed and (b) direction at the Gold Coast Seaway for ex-TC Oswald, Jan 2013 (data source: BoM).

The peak significant wave height recorded by the Gold Coast Waverider buoy during the event was 6.27 m at 10:38 28 th January 2013 with an associated peak wave period of 11.71 s (data source: DSITIA) (Figure 4). The peak wave direction was persistently east to east-north-east throughout Figure 1. Location of Currumbin Creek, SE Qld, Australia. the event. This would be a 5-10 year event in terms of wave height for the Gold Coast [1]. A maximum 3. Event Description: TC Oswald, Jan 2013 water level of 1.395 m AHD was measured at the TC Oswald developed in the Gulf of Carpentaria on Southport tide gauge located within the Gold Coast 21 st January 2013, making landfall as a Category 1 Broadwater (Figure 5) at 09:50 28 th January, being system on the west coast of Cape York Peninsula 0.565 m above the predicted high tide of 0.83 m [2]. The following morning, Oswald dissipated to an (data source: Maritime Safety Qld). The tidal ex-TC and tracked southward just inland of the east residual fluctuated between two peaks: 0.708 m at coast of Qld and NSW before moving offshore near 15:30 27 th January and 0.707 m at 01:50 28 th Sydney on 29 th January (Figure 2). This event January, and a minimum of 0.446 m at 21:10 27 th impacted the Gold Coast from 26 th –28 th January, January. Numerical modelling indicated that this with persistent east to east-north-east sub-100 km/h fluctuation was attributable to a variation in offshore wind gusts measured at both the Gold Coast wind speed at this time. The tidal residual at this site Seaway and Coolangatta (Figure 3). The maximum is not representative of the open ocean residual 1-min wind speed of 22.5 m/s (81 km/h) was given the complexity of this system with connections to Moreton Bay, multiple freshwater inputs and the Flooding was evident for , constriction of the Gold Coast Seaway. Currumbin Creek and the system although no properties were inundated. The flood event was estimated to be approximately a one in 20 year event by GCCC flood engineers.

3.1. Field Survey Description GCCM staff undertook a field survey of the Currumbin Creek catchment and surrounding open beaches on 1 st February 2013 to measure the spatial position and elevation of the highest debris line position, used to infer the peak water level, using a Leica GS15 GPS with a positional accuracy of 0.02 m for open locations.

The field survey results generally show an increase in the inferred peak water level from within the creek to the open ocean (Figure 7). A debris line position Figure 4. Measured wave (a) heights and (b) peak period of 1.7 m AHD was measured beneath the Gold for the Gold Coast Waverider buoy for ex-TC Oswald, Jan Coast Hwy Bridge. At Thrower Drive Bridge, debris 2013 (data: DSITIA). line positions of 1.46 m and 1.42 m AHD were measured on the western and eastern sides of the creek, respectively. Further upstream, the debris line position on the bank opposite the Currumbin Waters canal estate was 1.44 m AHD.

East of the Gold Coast Hwy Bridge, the debris line position was generally higher than at the bridge. The debris line position in the north-western corner Figure 5. Predicted tide, measured and residual water of the estuary mouth varied between 1.75-2.16 m levels within the Gold Coast Broadwater for ex-TC Oswald, Jan 2013 (data: MSQ). AHD. Along the northern bank of the estuary mouth, the debris line position varied between 2.03-3.23 m Ex-TC Oswald was associated with a heavy rainfall AHD. There was no evidence of wave over-topping over the Gold Coast. While there is no official rain of the dune on the northern side although the gauge within the Currumbin catchment, 261.8 mm breakwater was overtopped by waves as evidenced fell at Tallebudgera Road and 602.8 mm fell at by debris. On the southern bank of the estuary Tallebudgera Dam from 26 th – 28 th January. At the mouth, the debris line position was 1.88 m AHD. river gauging station monitoring Currumbin Creek, the water level and discharge increased to a peak of A sand delta extends into the creek mouth from its 2.89 m and 155.172 m 3/s at 22:00 27 th January southern bank and the peak debris line positions (Figure 6) (data source: Qld DNRM). here were 2.28-2.66 m AHD on the seaward- exposed northerly side and between 1.62-1.91 m AHD on the non-exposed southerly bank. At the pocket beach just within the training walls on the southern side of the inlet, the debris line position varied from 2.25-3.24 m AHD. The higher levels at this location indicate that this beach was subject to some wave energy during the event. On the easterly-exposed northern end of Currumbin Beach,

Figure 6. Discharge at Nicholls Bridge on Currumbin the maximum run-up level was 5.08 m AHD. On Creek during ex-TC Oswald, January 2013 (data: DNRM). Palm Beach, the maximum run-up level was 3.95 m AHD, which did not overtop the dune. The southern The coastal impacts experienced along the Gold end of Palm Beach has north-north-east exposure Coast as a result of ex-TC Oswald included and would have been subject to slightly lower wave extensive erosion of open beaches, especially those energy than Currumbin Beach. with an easterly to north-easterly exposure [3]. Overtopping of dunes and resultant oceanic inundation was observed in several locations. Figure 8. Model mesh and bathymetry for Currumbin Creek showing up- and downstream measurement points.

Figure 7. The maximum extent of inundation around the The model forced with discharge measured at Currumbin Creek entrance (m AHD) due to ex-TC Nicolls Bridge as the upstream boundary condition Oswald, January 2013 (imagery from OpenSeaMap.com). and astronomical tidal forcing on the open ocean

boundary for the calibration period shows good 3.2. Field Survey Interpretation agreement to the measured water levels at both The general increasing trend in peak debris line locations (Figure 9). To consider the model’s position from Thrower Dr Bridge to the Gold Coast performance in a period of high river discharge, the Hwy Bridge then towards the creek entrance could results of the TC Oswald simulation to follow were be due to several different dynamic processes. compared with output from GCCC’s in-house flood These include the penetration of wave set-up for a model and found to be comparable in timing and limited distance into the estuary, the dynamics of water level (pers. comm., Dr H. Mirfenderesk, 2013) the incoming tide and storm surge against the (not shown). This calibrated model was determined riverine flow and a change in bed level during the to be suitable for further use. event. In particular, the difference in inferred peak water level on either side of the delta feature may indicate that elevated water levels due to wave run- up were experienced on its seaward side, being a minimum of (2.28–1.91 m=) 0.37 m. The contributing processes to the peak water levels within the Currumbin Creek tidal inlet during this event are subsequently explored.

4. Numerical Modelling of Currumbin Creek 4.1. Model Set-Up and Calibration In order to simulate the inundation associated with the ex-TC Oswald event, a hydrodynamic model of Currumbin Creek was built in the finite volume 2D depth-integrated hydrodynamic model MIKE21 FM by DHI (Figure 8). Bathymetry for the creek was obtained from a hydrographic survey conducted for Figure 9. Sample comparison of modelled and measured GCCC in September 2012. The offshore bathymetry water levels for (a) upstream and (b) downstream was interpolated from a combination of beach locations in Currumbin Creek, April – May 2011. survey data (GCCC) and RAN chart data. The model was calibrated using water level 4.2. Simulation of ex-TC Oswald measurements collected by DHI for the period 14 th The calibrated model was used to simulate the ex- th April - 30 th May 2011 for two sites: upstream TC Oswald event for the period 09:00 25 January - th (28.16151°S, 153.46324°E) and downstream 09:00 30 January 2013. The upstream boundary (28.13584°S, 153.47737°E) (Figure 8). discharge was that measured at Nicolls Bridge (Figure 6). The offshore water level boundary condition was a combination of the predicted astronomical tide and storm surge generated using levels seaward of that location (Table 1). The peak GCCM’s calibrated Coral Sea tidal model forced water level at this bridge was not replicated by with global tidal model boundary conditions and forcing the model with the river flow, the tide and the BoM’s modelled ACCESS-R wind and pressure tidal residual (either that modeled or measured in fields [4]. The modelled winds had good agreement the Gold Coast Broadwater), although the ~1.4 m with the measured winds at Gold Coast Seaway in water level at the Thrower Drive Bridge further both speed and direction (Figure 10) as well as inland was replicated in run (4). No model other sites along the Qld coastline (not shown). simulations resulted in a slope in the peak water They also produced waves with good agreement in levels towards the entrance as observed in the field height, period and direction to those measured at survey, even when a change to the entrance DSITIA’s Gold Coast waverider buoy when used to morphology was made to simulate sand build-up. force GCCM’s Coral Sea spectral wave model (not shown). These results show that the modelled wind Table 1. Hydrodynamic model results for Currumbin data have sufficient accuracy and resolution to Creek for ex-TC Oswald, Jan 2013. replicate the peak wind in the event. Model Run Peak Water Level at Gold Coast Hwy Bridge (m AHD) Measured 1.70 m (1) Upstream river flow 0.56 m (2) River flow + tide 1.07 m (3) River flow + tide + 1.17 m modelled storm surge (peak 0.19 m) (4) River flow + tide + 1.42 m measured tidal residual at GC Seaway (peak 0.704 m) River flow + tide + measured 1.54 m (change in timing of tidal residual at GC Seaway peak to previous high tide – + (Entrance constricted with not in agreement with 0.5 m AHD sand bar) observations of residents)

A preliminary spectral wave model was set up to Figure 10. Comparison of ACCESS-R modelled with look at the penetration of waves into Currumbin measured wind speeds (both converted to 3s gusts) at Gold Coast Seaway (data source: BoM). Creek. The wave boundary conditions were supplied from the regional Coral Sea wave model The modelled storm surge offshore of Currumbin forced with BoM’s ACCESS-R wind fields. Waves peaked at 0.19 m, much lower than the 0.708 m up to 0.25 m were found to propagate to the sand residual measured within the Broadwater (Figure 5). delta feature but not beyond for a simulation This difference was investigated, and found to be including the water level variations and also local attributable to the fact that the Broadwater residual wind forcing. This would account for the difference was associated with the storm surge generated in the maximum measured water level on either side within the complex Moreton Bay system due to the of the sand delta feature, but not those on the persistent east-north-east winds as ex-TC Oswald northern bank. moved southwards, setting up a gradient in water level from north to south. In the regional Coral Sea 5. Discussion model results, the direction of flow through the Gold The field survey of the extreme water line resulting Coast Seaway throughout the Oswald event was from ex-TC Oswald for Currumbin Creek showed a seaward regardless of the phase of the tide (not distinct gradient in water level towards the entrance. shown). Therefore, the measured water levels We were unable to replicate this gradient by within the Broadwater were not representative of the numerical modelling of the river flow and storm tide, open ocean conditions at this time. even considering different entrance bathymetries. We postulate that wave set-up and run-up caused 4.3. Model Results elevation of the water level in estuary mouth during The model simulations were run by incrementally this extreme event, grading over a distance of adding the processes considered in order to approximately 500 m inland. From preliminary wave understand the contribution and interaction of each modelling exercises, the role of wave set-up in this to the resultant water level. Model output was event was not resolved. We conclude that accurate compared to the inferred peak water line data estimation of run-up levels would require more collected in the field survey, mainly the 1.7 m level detailed modelling of wave propagation. close to Gold Coast Hwy Bridge and higher water The maximum measured run-up level of 5.08 m • The importance of monitoring of open AHD on Currumbin Beach implies a wave run-up of ocean water level conditions for a minimum of 3.9 m given the maximum (tide + understanding storm tide; surge) ocean water level of ~1.2 m in the event. • The knowledge gap in the behaviour of Using empirical equations for the maximum offshore wave set-up across river entrances in Hrms and Tp and an estimated beach slope of 0.11, extreme events affects inundation the R wave run-up is calculated to be 4.02 m [10], 2% forecasting; which is inclusive of the wave set-up component of • The complex behaviour of water levels 1.68 m [8], which fit well with the measurements on within individual estuaries precludes the use the open beach. However, it is obvious that wave of design levels derived from set-up of this magnitude was not acting across the measurements at such sites to be applied creek entrance, although some intermediate level of as open ocean tailwater levels. wave set-up may have been contributing to the tailwater level (in addition to the modeled storm 7. Acknowledgements surge) given the improved model water level in the We acknowledge the kind contributions of data and estuary when using the measured tidal residual advice from Saeed Shaeri (PhD Candidate, GCCM), from the Gold Coast Broadwater. GCCC, DHI, MSQ, BoM, DSITIA and DNRM.

A case study of the trained Brunswick River 8. References entrance by Hanslow et al. (2001) [9] investigated [1] Allen, M. and Callaghan, J. (1999) Extreme wave the effects of wave height and riverine flooding on conditions for the south Queensland coastal region. In: wave set-up, and identified a water level anomaly 14th Australasian Coastal and Ocean Engineering over 0.5 m associated with TC Roger (1993) not Conference, 14-16 April 1999, Perth WA. attributed to storm tide or riverine flow. They found [2] BoM (2013) Special Climate State 44 – extreme that when compared with the wave set-up on a rainfall and flooding in coastal Queensland and New natural beach, wave set-up in a similar river South Wales. 5th February 2013. entrance is typically smaller under similar wave [3] Burston, J.M., Gee, D., De Boer, J, (2013) Coastal heights at the same depth, possibly due to the river impacts of ex-Tropical Cyclone Oswald 2013 on the Gold current causing local set-down as it spreads out of Coast, Queensland. Griffith Centre for Coastal the entrance, counteracting the effect of the local Management Technical Report. May 2013. wave-driven set-up. The behaviour of this process under extreme wave conditions is not clear, and [4] Burston, J.M. and Symonds, A.M. (2013) Coral Sea Hydrodynamanic Model Set-up and Calibration. Griffith some proportion of the wave set-up could still be Centre for Coastal Management Report. May 2013. acting as an increased tailwater level. They also raise the possibility that the anomaly could have [5] Callaghan, J. (2012) Known Impacts of Tropical been due to coastally trapped waves. In summary, Cyclones, East Coast, 1958-2008. Archive for HardenUp Queensland. http://hardenup.org/ we have not been able to identify the process leading to the peak water levels measured in the [6] Castelle, B., Bourget, J., Molnar, N., Strauss, D., Currumbin Creek from numerical modelling of Deschamps, S., and Tomlinson, R. (2007) Dynamics of a riverine flow and storm tide alone. wave-dominated tidal inlet and influence on adjacent beaches, Currumbin Creek, Gold Coast, Australia, Coastal Engineering. 54(1): 77-90. 6. Implications of Findings The findings of this case study have implications for [7] GCCC (2012) Currumbin Creek Dredging. Gold Coast both emergency managers and the overarching City Council Fact Sheet. http://www.goldcoast.qld.gov.au/ storm tide forecasting project. It underlines the currumbin-creek-dredging-4520.html. complexity of extreme water level behavior in tidal [8] Hanslow, D. & Nielsen, P. (1993) Shoreline set-up on inlets, in particular its dynamic and interrelated natural beaches, Journal of Coastal Research SI 1-10. response to the riverine flow, oceanic water levels, [9] Hanslow, D., Nielson, P., and Hibbert, K. (2001) Wave wave penetration and changing bed morphology. Setup at River Entrances. Coastal Engineering Forecasting inundation at high spatial resolution in Proceedings , 1 (25). doi:10.9753/icce.v25. such environments would require complex coupled [10] Nielsen, P. and Hanslow, D.J. (1991) Wave runup models with detailed bathymetric, flow and oceanic distributions on natural beaches. Journal of Coastal water level inputs. Research 7: 1139-1152.

Some other findings of this study include: • The difficulty of validating model output against debris line data and hence;