Modelling the compound flood hydrodynamics under mesh convergence and 1 future storm surge events in Brisbane River Estuary, Australia

: Floods are the most common and destructive disasters around the globe, which becomes 13 more challenging in coastal areas due to higher population density and catchment area relative to 14 floods in an inland area. For effective coastal flood management to reduce flood adverse impacts it is 15 necessary to investigate the flooding processes and their joint interaction in a coastal area. This paper 16 selected the Brisbane River Estuary, Australia as an example and the MIKE 21 model is applied to 17 investigate the effects of mesh resolution on the flood discharge and to explore compound flooding 18 by computing variances in coastal flood assessments resulting from a separation of tidal and riverine 19 processes. The statistical results showed that the Nash-Sutcliffe coefficient, E of water level are varied 20 from 0.84 to 0.95 and the model simulated the 2011 flood extent results agreed with 90% accuracy 21 with the observed flood extent. Five mesh resolutions cases were analyzed and the result found that 22 the finer mesh resolution Case 5 was more appropriate for calculating the peak discharge with 2.7% 23 with estimated discharge. Compound flood event simulation results emphasized that not considering 24 the interaction of various flooding drivers caused 0.62 m and 0.12 m reduction in the flood levels at 25 Jindalee and Brisbane city gauges, and uncertainties in flood extent. Simulated results of flood at 26 Brisbane city gauge, showed that 2011 and 2013 floods with storm surge scenario 4 demonstrate, the 27 increase in flood level to be 12% and 34% respectively. The results recommend flooding assessment 28 by using mesh convergence with joint probability of compound flood under future storm surge for 29 planning and management of coastal projects. mesh make an 221 effect on water levels. When the results were consistent, then we used the coarser-resolution with 222 confidence, otherwise, we reduced the cell size until we got a consistent


Introduction 32
Flooding is a prevalent and most destructive catastrophe worldwide, which poses a severe threat to 33 lives and properties (Geravand, Hosseini, & Ataie-Ashtiani, 2020; Khalil & Khan, 2017; X. Liu & 34 Lim, 2017). Coastal flooding is likely to increase in the future (Sadler,Goodall,Behl,Bowes,& 35 surge is the rise of water level above the normal sea level along a coast due to reduced atmospheric 48 pressure and/or strong coastal winds (Karim & Mimura, 2008). The storm surge influences may 49 further increase when they coincide with riverine flooding (Zheng,Westra,Leonard,& Sisson,50 2014) and the resulting combination is known as compound flood events (Leonard et al., 2014;Wu, 51 Westra, & Leonard, 2020). Initially, the two involved flooding drivers involved were managed 52 individually in coastal flood management (Torres et al., 2015). However, studies show that storm 53 surge and extreme rainfall processes are statistically dependent, and thus their joint interface needs 54 to be considered (Hawkes & Svensson, 2006;Svensson & Jones, 2004;Zheng, Westra, & Sisson, 55 2013). For an effective coastal flood inundation assessment, multiple measures, such as flood 56 assessment models using good resolution digital elevation data (DEM) and considering a compound 57 flood event with future storm surges are required to be implemented to reduce flood adverse impacts. 58 In the past various researchers have given substantial efforts to simulate flood inundation in the 59 impacts assessments in IPCC recent report (IPCC, 2021) have urged us to consider and model future 84 impacts and take remedial measures to reduce their impacts. Flooding events specify the need for a 85 sustainable modelling approach to simulate the flood extent and propose measures to alleviate future 86 flooding. To access the flood hydrodynamics, the Brisbane River Estuary (BRE) Australia were 87 studied, which has been exposed to several flooding events over the past century, while, it has 88 This study examines and analyses BRE hydrodynamics by using MIKE 21 hydrodynamic model to 106 investigate the flood extent with various mesh resolutions. Further, to understand the interaction of 107 storm-tide and fluvial flooding mechanisms, we investigate a compound flood event in a BRE to 108 quantify how changing boundary set-ups at the entrance in Moreton Bay affect modelled water 109 levels and flood extent of the study area. In this paper, we address several of the issues outlined 110 above; the objectives of this study are threefold: (1) to develop a Brisbane River estuary and 111 Moreton Bay flood model for various flows events under converging mesh size; (2) to conduct a 112 simulation to assess the impact of compound riverine flooding and tides on water levels; and (3) to 113 analyse the future storm surge effect on flood extent. The outcome of this study will be significant 114 to comprehend the suitability of the hydrodynamic model to carry out flood modelling. Further, it 115 will help to identify the flood-exposed areas and to apply possible remedial strategies to overcome 116 the damage. The study results will help decision-makers to make a flood mitigation and management 117 plan of action. 118 The paper is structured as follows: the case study area, Brisbane River estuary (BRE) is described 119 in Section 2. Section 3 explains the Brisbane flooding, hydrodynamic model, the data requirements, 120 and methods, including mesh resolution effect and compound flooding in BRE. The results of the 121 hydrodynamic model calibration and validation along with mesh effects and compound flooding 122 influence are described and discussed in Section 4. Finally, conclusions are specified in Section 5. 123

Study area 124
The Brisbane River and Moreton Bay are located on the southeast coast of Queensland, Australia 125 ( Fig. 1). The lower part of Brisbane River is termed the Brisbane River estuary (BRE). Moreton 126 Bay is semi-closed coastal water situated at the mouth of Brisbane River. BRE and Moreton bay 127 experience semidiurnal tides with a tidal range of 2.5 m. The Brisbane River has the longest course 128 in sub-tropical SEQ, having a length of 344 km and has a catchment area of 13,600 kmP 2 P (Eyre, 129 Hossain, & McKee, 1998) to the Port Office Gauge which is located in the heart of Brisbane City. 130 The BRE is a micro-tidal estuary, with a mean spring and neap tidal range of 1.8 m and 1.0 m 131 respectively (Wolanski, 2014). It has a tidal influence up to 80 km from the river mouth. The Oxley 132 Creek and Bremer Rivers are major tributaries, which contribute to lower half catchment flows in 133 BRE and join the estuary at 34 and 73 km respectively, from the river mouth.

Data collection 159
To carry out this study we have collected, DEM bathymetry data, water level, flow data, tidal data, 160 flood measurement data, and flood extent satellite data. Data collected for the study with its 161 resolution and sources are shown in Table 1. 162 Where t is the time, and y are Cartesian coordinates, h is the water depth, s is the discharge, is 180 reference density of water, is water density, is atmospheric pressure (Pa), In CFLHD Eq. 1, as the local flow velocity is very less as compared to the local water depth so, it is 202 reasonable to disregard the velocity terms, and the CFLHD can be rewritten as Eq. 6 and further, the 203 CFLHD is rearranged to Eq. 7 with the reasonable assumption of ∆ ≈ ∆ . 204 Five meshes were generated and gradually attuned until all elements fulfilled the constraint in Eq. 207 7. Mesh quality was further enhanced by the smoothing tool to increase spatial regularity. The 208 details of the elements, nodes, element areas CFLmax and simulation running time in each case are 209 given in Table 2 Table 3. 244 Based on these four scenarios, storm tide inputs at Brisbane bar (Fig. 5) were used to simulate the 245 flooding behavior in BRE under low flow events. 246  and tidal data (Fig. 6 c&d) were used for the model performance. Finally, the model included BRE 260 and Moreton bay (Fig. 4c) with 2011 flood and tidal data (Fig. 6 e-h).  Fig. 7 (c&d). 297 The performance indices for calibrating gauging stations are shown in Table 4 The result of the comparison between observed and simulated flood extent is shown in Fig. 8. 308 Brisbane City experienced a major flood from 12 th January 10:00 am to 13 th January 6:00

Mesh resolution effects on discharge 323
The model performance results for the simulated discharge by using different mesh resolutions at 324 Brisbane City gauge is shown in Fig. 9. The results display a higher difference of coarser mesh with 325 the observed data and as mesh size become finer the observed and simulated discharges reduce, 326 indicating that the simulated discharges were correctly represented by the finer mesh resolution, as 327 also proposed by (Teng et al., 2017). The percentage difference of peak value of simulated 328 discharges with estimated discharge by Barton

Modelled water levels and flood extents under varying boundaries 340
The results of the interaction of storm-tide and fluvial flooding mechanisms by modelling a 341 compound flooding event in the BRE are shown in Fig. 10. The simulated and observed water levels 342 with and without river and tidal boundaries are presented at three gauge locations in Fig. 10. The 343 flood extents corresponding to these boundaries are presented in Fig. 11. Comparison of results at 344 the Jindalee gauge station (Fig. 10a) with and without tidal boundaries show that the peak water 345 level varied slightly, with a 0.62 m reduction at peak level without tidal input. Further, without tidal 346 boundary, the hydrograph has attained smooth rising and falling limbs without showing any tidal 347 variations. While, without discharge boundary i.e. Q=0, the tidal input moved up to the Jindalee 348 gauge, and caused a slight reduction in tidal levels. The comparison of peak water level with and 349 without tidal boundary at Brisbane City gauge (Fig. 10b) shows that the difference of peak flood 350 level could be as high as 0.12 m, while without riverine boundary the water level followed the tidal 351 wave pattern at Brisbane City gauge. At Brisbane bar, without tidal boundary, the water level 352 followed a straight line, with a slight increase in water level during the flood days, while the tidal 353 level at Brisbane Bar was slightly reduced without riverine boundary (Fig. 10c). 354 1, with a 25% increase in tidal level, the tidal input mainly remained inside the BRE, while causing 373 very minor flooding near the estuary mouth (Fig. 12 a). The tidal level inside the BRE was just below 374 the minor flood level of 1.7 m. In Scenario 2, with a 50% increase in tidal level, the tidal level crossed 375 the minor flood level and tidal inflow caused flooding in the tributaries adjoining the BRE (Fig. 12b). 376 However, with Scenarios 3 and 4, the flood extent increased near the BRE mouth and Brisbane City 377 gauge area (Fig. 12 c&d). Further, the flood water level surpassed the minor flood level and levelled 378 with a medium flood level of 2.6 m in Scenario 3 and 4 respectively. 379 Brisbane City gauge increased to 5 m due to the joint probability of riverine and tidal effects and 385 hence flooding extent and depth increase in the floodplain area, with more flooding at the BRE mouth 386 (Fig. 13 a). Similarly, the flood height for the 2013 flood increased from 2.24 m to 3.01 m due to the 387 joint probability of river and storm surge, crossing the medium flood level of 2.6 m at Brisbane city 388 gauge and leading to flooding in the Oxley creek area (Fig. 13 b). The modelling with the future storm 389 surge scenarios has shown that for flood inundation study and coastal planning in BRE, the 390 combination of riverine flow and the storm surge effect due to climate change were considered. As