The Delft3D-Flow hydrodynamic model was used to predict the impact of the Hong Kong-Zhuhai-Macau Bridge on the tidal flow in the Pearl River Estuary. In addition, the impacts of the flow currents on nearby bridge piers and artificial islands were assessed. To obtain additional spatial and temporal observations relative to the tidal gauge station observations, the sea surface height, which was derived from satellite altimetry ENVISAT RA-2, was closely correlated with the simulated tidal levels on individual days. These variables were modeled in the presence and absence of the Bridge, and the sea surface height validated numerical model was used to study the impacts of the Bridge on local water levels and circulation patterns. The construction of this Bridge resulted in small changes in tidal levels in that area. The average highest and lowest tidal levels changed by less than 0.01 m, which is about 0.4% of the highest tidal level and 2% of the lowest tidal level. The total tidal range changed by nearly 0.015 m. These results indicated that coastal engineering hardly affects the tidal flow in the Pearl River Estuary. In addition, complex eddy and circulation patterns were discovered near artificial islands. It is important to understand the complex flow patterns to ensure traffic safety during bridge construction and to determine the associated impacts of the bridge on tidal currents in the Estuary.
Estuary areas are complex water regimes that are influenced by fresh and saline water. These areas are often dramatically affected by human activities. Thus, infrastructure construction in these areas affects coastal tidal flow and water quality because the water flow is obstructed by the constructed bridge piers and artificial islands. These changing hydrological regimes may threaten safety during engineering operations. To assess the significance of these effects, well-developed, -calibrated, and -validated numerical models are often used. Remotely sensed data are often integrated to validate numerical model. These data complement gauge station measurements, especially when only a few stations exist. The environmental impacts of bridge piers and artificial islands can be investigated with shallow water hydrodynamic models (Liu, 2006).
Previously, hydrodynamic models were often used in coastal engineering projects. For example, the Delft3D and UNIBEST numerical models were used to predict future shoreline evolution and to determine the actual impacts of constructing a deep-water harbor jetty near Nouakchott (Elmoustapha et al., 2007). Furthermore, the effects of reservoir construction on tidal hydrodynamics and suspended sediment distributions in the Danshuei River Estuary were modeled by Liu (2007). A detailed hydrodynamic model was developed to simulate 3D stratified tidal flow and larger-scale eddy-dynamics with a horizontal large eddy simulation (HLES) to assess the safety of constructing the Stonecutters Bridge in Hong Kong (Morelissen et al., 2010). A two-dimensional finite volume numerical model based on an unstructured triangular mesh was used to study the hydrodynamic impacts and the extent of flood inundation during different tidal renewable energy projects in the Seven Estuary, UK (Xia et al., 2010). An unstructured-grid Finite Volume Coastal Ocean Model (FVCOM) was used to study the hydrodynamic impacts of large-scale structures in the Changjiang Estuary, China (Ma et al., 2011). In addition, researchers have studied flow scour around bridge piers to evaluate the effects of bridge construction on the coastal environment (Richardson et al., 1998; Yang, 2005; Abdeldayem et al., 2011). 2-D mathematical models were used to simulate flow effects around the manufactured island of the Hong Kong-Zhuhai-Macao Bridge in the Pearl River Estuary (PRE), which confirmed its potential use (Wang et al., 2010; Li et al., 2011 a). Numerical models should be well-rounded, calibrated, and validated to investigate the impacts of estuarine engineering projects on tidal flow. Thus, remotely sensed data can be used as complementary data for model validation. However, remotely sensed data and hydrodynamic models have rarely been integrated to study the effects of coastal engineering on tidal flow quantitatively.
Remote sensing is widely used as a technology for periodically acquiring synoptic measurements of the earth’s surface to monitor environmental changes. In addition, hydrological parameter monitoring can benefit from remote sensing. For example, for tidal level monitoring, the sea surface height (SSH) can be derived from radar altimetry data. After satellites with radar altimetry were initially launched in the 1970 s, hydrologists have investigated the potential of using these data for monitoring water levels (Peter et al., 2001; Leon et al., 2006; Cai and Ji, 2009; Joecila et al., 2012). Because the coverage and data acquisition time are different relative to traditional hydrological gauge station measurements, satellite radar altimetry measurements can supplement traditional monitoring methods. In this study, a hydrodynamic model was validated by satellite altimetry data to investigate the influence of bridge construction on the tidal flow dynamics in an estuary.
When using hydrodynamic models to study tidal flows under the influence of bridge piers, the stability and accuracy of the hydrodynamic model must be considered. Generally, field measurements have been used to validate the accuracy of model results. In this study, sea surface heights were determined from satellite altimetry at various virtual stations to validate the hydrodynamic model. The ENVISAT RA-2 data were selected in this study by comparing the ground track coverage of the satellite altimetry within the study area. The hydrodynamic model was used in 3D mode to resolve horizontal eddy and water level elevation obstructions that were caused by the bridge piers and artificial islands in the cross-sea bridge construction area. The main deck construction will involve steel deck segments from dynamically positioned barges. Accurate flow models and predictions are necessary to plan the construction phases and to allow ships to navigate the area safely during the 7-year construction period.
To assess the impact of the Hong Kong-Zhuhai-Macau Bridge (HKZMB) construction on tidal flow in the PRE, a three-dimensional hydrodynamic model was developed based on the hydrological characteristics of the study area. The model was validated with regularly observed tidal levels from both field observations and satellite altimetry. Next, the validated model was used to predict the water level and flow. Complex flow patterns were revealed by studying the changes in the average highest and lowest average tidal levels and by studying the horizontal eddies near the bridge piers and artificial islands. This study aims to predict the tidal flow changes before and after the construction of the HKZMB. These results can provide scientific support for safe operations during bridge construction and be used to analyze tidal flow predictions in the PRE.
Study area and data
The PRE, also known as the Lingdingyang Estuary, is a deltaic estuary that forms part of the larger Pearl River Delta. The Pearl River includes the West (Xijang), North (Beijiang) and East (Dongjiang) Rivers and drains into the South China Sea. The drainage system of the mouth of the Pearl River is extremely complex (Figure 1). The channel network discharges into open water through eight main outlets, including the Humen, Jiaomen, Hongqimen and Hengmen outlets, which discharge into the PRE. The Modaomen and Jitimen outlets discharge directly into the open sea. In contrast, the Hutiaomen and Yamen outlets discharge into the Huangmaohai Estuary, which is the second largest estuary in the Pearl River Delta. We focused on five of the PRE outlets, including the Humen, Jiaomen, Hongqimen, Hengmen and Modaomen.
The HKZMB is a 50-km-long sea-crossing bridge that is essential for linking the three following economic zones: Hong Kong, Zhuhai and Macau. The construction of the HKZMB is an important strategic move for promoting regional economic development throughout the Pearl River Delta. The HKZMB is a large bridge system numerous segments, artificial islands and undersea tunnels that connect Hong Kong, Macau and Zhuhai, which are three major cities located in the Pearl River Delta. Construction of the HKZMB formally began on 15 December 2009 and is scheduled to be finished in 2015 or 2016.
Complex hydrological and geomorphic conditions occur in the PRE. The bathymetry of the PRE is characterized by three shoals and one underwater channel (Figure 2). The hydrological regime of the Pearl River is governed by the East-Asian monsoon (Dong et al., 2004). The summer monsoon period, which occurs between April and September, accounts for more than 80% of the annual discharge volume of the Pearl River. The combined annual average flow of the main branches was 286 × 109 m3 between 1955 and 2005 (Dai et al., 2008). The annual runoff to the Pearl River is 34 × 109 m3, which includes 26.7 × 109 m3 from the Xijiang River (80% of the Pearl River discharge), 4.6 × 109 m3 from the Beijiang River and 2.7 × 109 m3 from the Dongjiang River. Regarding river discharge, the Pearl River is second only to the Yangtze River in China.
In this study, the hydrodynamic PRE model considers the discharge from five main outlets. The daily water discharge data for the five outlets were obtained from the Pearl River Water Resources Bureau website. To validate the simulation, the hourly-observed tidal level data at the Macau and Hong Kong gauge stations were downloaded from China’s Marine Service Network website. The daily wind synoptic data were provided by the Hong Kong Observatory (HKO).
Radar altimetry ENVISAT RA-2 data were used to validate the PRE hydrodynamic model. The ENVISAT satellite, which is designed to help the scientific community understand the Earth’s environment and the processes that are responsible for climatic changes, was launched by the European Space Agency (ESA) in February 2002 (Frappart et al., 2006). The ENVISAT orbits on a 35-day repeated cycle, which provides global scale data to measure ocean topography, water level variations in large river basins and land surface elevation and to monitor sea ice and the polar ice caps (Wehr and Attema, 2001). The RA-2, which is carried by the ENVISAT, is an instrument that precisely detects the two-way delay of the radar echo from the Earth’s surface and measures the power and the shape of the reflected radar pulses. RA-2 is a nadir-looking pulse-limited radar altimeter that is based on the heritage of the ERS-1 RA and functions at the main nominal frequency of 13.575 GHz (Ku-Band). In addition, another channel (3.2 GHz S-Band) is operated to estimate the ionospheric propagation delay and for geophysical applications. Thus, SSH can be extracted from RA-2 data through a series of corrections and used as a virtual station to validate the hydrodynamic model. The ENVISAT RA-2 dataset and the GDR and Cyclic reports used in this study are available in the ESA FTP archives (ftp://diss-nas-fp.eo.esa.int) after registration. As an observation sequence, the RA-2 GDR data were selected for the PRE for the study period. The RA-2 measurement points are shown in Figure 1.
Hydrodynamic model set-up
The Delft3D-FLOW model was used because the program was developed specifically to model unsteady water flow in shallow seas, coastal areas, estuaries and rivers (Delft Hydraulics, 2011). The Delft3D-FLOW model solves the shallow water equations with specified boundary conditions in two or three dimensions. The continuity and horizontal momentum equations were solved with the implicit finite difference method (ADI) on a staggered grid. For the vertical grid, a sigma-coordinate approach was used. The mass balance for each grid cell was considered in the continuity equation. The Delft3D-FLOW model was used to predict the influence of the bridge construction on tidal flow in this study.
In the PRE hydrodynamic model, the computation domain was divided to generate computation grids in the horizontal and vertical directions. The PRE stretches approximately 105 km longitudinally and 55 km latitudinally. An orthogonal horizontal curvilinear grid layout was used to map the computation domain at the bridge piers and artificial islands. Vertically, there are 7 sigma levels with finer resolutions near the surface and bottom. Depth was triangularly interpolated from the bathymetry of the computing grid points. The time step used in this study was 1 s, which corresponds to the basic CFL condition.
Given the external forces in the numerical experiment, three factors were considered, including tidal constituents, river discharges and meteorology. At the open boundary of the PRE, five major tidal constituents were included as follows: S2, M2, N2, K1 and O1. The exacted amplitudes and phases for each tidal constituent that was interpolated from the cotidal chart are specific to the grid points at the open boundary. Discharges were obtained from the river hydrographical stations. In addition, meteorological data from observatories near the PRE were used. Wind speed and direction are provided on a daily basis and are collected at the Waglan Island observation station.
To identify the impacts of the HKZMB construction on tidal flow variations in the PRE, two periods of 10 days each were selected based on the revising time of the ENVISAT satellite over the PRE. The periods were between 10 October and 20 October 2009 and between 15 November and 25 November 2009. During each period, two scenarios were designed to simulate the variations with and without the bridge piers and artificial islands (assuming that all other factors were equal). The first scenario was without the bridge effects and the second scenario included the bridge piers and artificial islands. To a certain degree, the related parameters must be recalibrated based on the calibration period parameters. After repeated adjustment, the optimal parameters were obtained to model the Pearl River Estuary. The bottom roughness (the Manning Roughness coefficient) was set to 0.018 and the horizontal eddy viscosity was set to 0.2 m2·s−1. The 3D turbulence was based on the k-ϵ model. Finally, the observed tidal levels at the Macau and Hong Kong gauge stations were used to validate the PRE hydrodynamic model.
Bridge pier generalization and grid refinement
The HKZMB consists of two main parts, including the HKZMB over the Pearl River and the HKZMB to Hong Kong. The longest bridge section will be 22.8 km long and span between the artificial islands for the Macau tunnel exit and the Macau border facilities. This section will include three cable-stayed spans of between 280 and 460 m in length. The HKZMB is connected to the inland Hong Kong highway system through the North Lantau Highway Connection. The HKZMB will begin at San Shek Wan on Lantau Island in Hong Kong and extend from Tai O across the Pearl River Estuary to a cross-harbor tunnel before arriving at Macau and Zhuhai ports and connecting with Gongbei in Zhuhai and the Oriental Pearl in Macau. The HKZMB will stretch across the Estuary and consist of more than 400 piers and 4 artificial islands, including the Zhuhai-Macau Port, the East and West Artificial Island and the Hong Kong Port.
The bridge piers and artificial islands in the discharge area of the waterway will affect flood discharge and the local river hydrological regime. However, the sizes, water depths, and locations of the bridge piers and islands may not coincide with the computation grid scale. Therefore, we generalized the bridge piers and artificial islands with numerical simulations. Because this computation was notably large and included many piers and islands (especially in the Estuary), methods to refine the grid locally and modify the water resistance coefficients were used to maintain computation efficiency. To simulate the impact of bridge piers on tidal flows, computation grids were locally refined near bridge piers for both rows and columns. The grid cells that contained bridge piers or islands were set to dry with no water discharge. Locally refined grids can potentially reveal the tidal flows and horizontal eddies around bridge piers and islands while maintaining computational efficiency.
Virtual station and its sea surface height extraction from satellite altimetry
Due to the large tidal level variations and the limited observations from traditional hydrologic stations in the PRE, satellite altimetry was used to provide additional water level information. A virtual station can be setup based on altimetry to define the intersection of a satellite ground track with a water body and to supplement traditional measurement data sources that are used in numerical models. The virtual SSH from the station and the height (or topography) of the ocean’s surface were used as hydrological parameters for monitoring the Estuary.
where Id is the ionosphere Ku-band correction, is the model dry tropospheric correction, is the sea state of the Ku-band, is the wet tropospheric correction, is the pole tide correction, and is the solid earth tide correction. In this study, the SSHs at the different virtual stations were calculated with Equation (2), which was derived from the ENVISAT RA-2 and covered the entire Pearl River Estuary.
The hydrodynamic model was carefully set up with refined local grids. The model was calibrated and validated for the specified PRE hydrologic regime. The ENVISAT RA-2 data were processed to validate the hydrodynamic model and to link them to gauge station measurements. Therefore, the RA-2 coverage points were set as the virtual stations to measure the hydrological parameters periodically. The virtual stations and traditional gauge station measurements were taken together, which enabled for comprehensive monitoring of the PRE hydrological conditions.
Results and discussion
Model validation with the data observed with the gauge station and the sea surface height observed with the virtual station
Numerical simulation results for both periods were validated against the observed tidal levels at the Macau and Hong Kong (Victoria Harbor) gauge stations. Figures 3a,b show that the simulated water level fluctuations corresponded to the 4 hourly-observed tidal levels at the Macau station. The errors for the largest and smallest tidal heights were less than 0.15 m. The simulated tidal levels were delayed by approximately 10 min relative to the observed data. Figures 3b,d show that a good correlation occurred between the 4 hourly results from the numerical simulation and the gauge observations at the Hong Kong station. The deviations of the largest and smallest tidal height were less than 0.10 m. In addition, the phase was synchronic between the simulated water levels and the observed data. When the model was carefully set up, it was highly accurate and required only limited calibration. The model validation showed that the simulated results adequately represented the complex currents in the PRE. Thus, the model is a good tool for predicting tidal flow during the construction of the HKZMB.
The traditional hydrological gauge stations that are fixed along the coastal line potentially play a limited role in validating whole estuary simulations. The high accuracy of the satellite imagery (with a centimeter resolution) and its repeating cycle of two visits to the PRE every 35 days make the ENVISAT RA-2 data suitable for observing water coverage. The SSH data for the PRE was obtained from the ENVISAT RA-2 data as described in the previous section. The ENVISAT RA-2 GDR data acquired on 13 October, 17 October, 17 November, and 21 November in 2009, were processed with the Basic Radar Altimetry Toolbox (BRAT) that was developed by the ESA (ESA & CNES, 2011). The processed SSH data were compared with the water levels that were obtained from the numerical simulations for the corresponding sites and times. Figure 4 provides the linear fit between the ENVISAT RA-2 derived SSH data and the simulated water levels. The ENVISAT RA-2 SSH data are well-correlated with the simulated water levels (based on a significance test and the 99% confidence interval).
The reliability of the valid measurements from the virtual station depends on their effective spatiotemporal coverage. The 35-day repeating cycle of the ENVISAT provides more frequent observations than field measurements alone. Thus, the valid measurements and the altimeter data determine the temporal coverage of the measurements. Furthermore, the integrated observation network, which includes the virtual stations and gauge stations, provides adequate spatial measurement coverage and enables intensive spatial coverage. After validation with the virtual and gauge station observations, the accuracy of the model satisfies the accuracy requirements for predicting tidal levels in the PRE.
Predicted tidal level change before and after bridge construction
The construction of the HKZMB will affect local tidal levels and flow patterns. In this section, the changes in tidal levels and flow patterns are investigated. To examine the influence of the bridge piers on tidal flow, the areas around the bridge piers were assessed to estimate the HKZMB construction. Figures 5a,b depict the different water levels near the bridge piers and artificial islands. The darker colors indicate larger water level differences before and after bridge pier and artificial island construction. A distinct boundary occurs and curves toward the incoming water direction along the direction of the bridge piers. The bridge piers tend to increase the level of incoming water flows in front of the piers and decrease the level of water behind the piers. The high-low water level boundary is concurrent result from the both conditions. Along the axis of the bridge piers, the water level is significantly more affected near the bridge piers, which results in an arc-shaped boundary. Occasionally, this boundary is not a straight line (Figure 5b) because the flow direction is not perpendicular to the bridge pier line, which has a more complex flow pattern.
Nine points of interest (P0 to P8) along the Lingdingyang waterway (from north to south, Figure 2) were selected to examine tidal level changes before and after the HKZMB construction. The changes in the average high tide levels (CAHTL) and the average low tidal levels (CALTL) were calculated and tabulated at these points and are presented in Table 1. Most tidal level changes were less than 0.01 m. However, the change in the average high tidal level at P3 and the change in the average low tidal level at P4 were greater. The average high tide levels decreased downstream of bridge construction, and the average low water levels increased at all points except P6, P7 and P8, which were located in the external Estuary. Thus, the HKZMB construction will decrease the high tidal levels and increase the low tidal levels on the north side of the HKZMB construction area. In contrast, the opposite pattern was observed on the south side of the bridge in the open water areas. Overall, the tidal range changed by less than 0.015 m. In addition, the tidal ranges shrink on the upstream side of the HKZMB construction areas and increased downstream of the open water areas. This result occurred because the piers and artificial islands obstruct water following the HKZMB construction.
Tidal flow changes and horizontal eddies near the bridge piers and artificial islands
In the PRE prediction model for HKZMB construction, horizontal eddies were observed near the bridge piers and artificial islands, which could threaten bridge deck lifting operations during HKZMB construction. Horizontal eddies will cause different water level and complex current patterns. To ensure safe operations, complex flow patterns and eddies should be understood before barge positioning and navigation. Therefore, it is important to identify the horizontal eddy characteristics near the bridge construction areas. For example, the horizontal eddy near the East Artificial Island of the HKZMB (Figure 6) demonstrates a depth averaged velocity distribution pattern. Due to tidal flow effects and island obstructions, the simulated eddy moved from the East Artificial Island to the southwest before moving in the opposite direction. The horizontal eddy disappears with time, which reflects the periodic variations of the horizontal eddy.
After the HKZMB construction, based on the evolving eddy formation affected by the HKZMB construction may extend up to 2–3 km from the bridge piers and artificial islands. These results corresponded with previous results (Li et al., 2011b). Therefore, only the local flow patterns are affected by the HKZMB construction.
In this study, a tidal flow model was used to predict tide levels in the PRE in China before and after HKZMB construction. This model was validated with observed tidal levels that were obtained from hydrological gauge stations. Furthermore, a virtual station concept that extends the spatial and temporal coverage of traditional gauge station observations was used to improve the model’s prediction accuracy. The SSHs that were derived from remotely sensed satellite data were used to validate the model. This validation indicated that the simulated results adequately represent the complex currents in the PRE. Thus, the proposed method provides accurate tidal flow predictions for coastal engineering. Regarding the gauge station data, the SSH from the virtual station provided additional temporal and spatial data. With this data, the hydrodynamic model for the coast area was validated. Remotely sensed data integrating numerical simulation can be useful for investigating the tidal flow impacts of coastal engineering.
Based on the scenario simulations and results comparison, the water levels increase due to the resistance of the bridge piers and artificial islands. A distinct water level difference boundary appears along the direction of the bridge piers and curves towards the direction of the incoming water. In addition, the eddy processes varied and the eddy migrated horizontally near the bridge. Thus, the HKZMB construction hardly affects the tidal flow of the PRE. In addition, this affect is limited to the bridge area. During the HKZMB construction, the complex flow patterns and eddies must be understood to ensure safe barge positioning and navigation. The accurate tidal flow model and prediction that are provided in this study will allow for better planning for the different HKZMB construction phases and allow for safe navigation during the next three- or four-year construction period.
In this study, changes in the PRE tidal flows following HKZMB construction were predicted at a large scale with a hydrodynamic model. Due to the small size of the bridge piers in the study area, it was difficult to consider the shape of the bridge piers within the large Estuary. Thus, local analogue simulation and physical experiments in the laboratory will be implemented in the future on a small scale to determine the effects of bridge pier shape on water flow.
This work is funded by the National Natural Science Foundation of China (NSFC) (Grant Nos. 41101415 and 41331174), the Hong Kong Research Grants Council (RGC) General Research Fund (Grant No. B-Q23G), the National High Technology Research and Development Program (863 Program) (Grant Nos. 2012AA12A304 and 2012AA12A306), the Major Science and Technology Program for Water Pollution Control and Treatment (2013ZX07105-005); LIESMARS Special Research Funding and Special funds of State Key Laboratory for equipment.