An intensive observation and a three-dimensional finite-difference model of Environmental Fluid Dynamic Code were used to study saltwater intrusion in the Pearl River Estuary. The model simulation was carried out for December 2007 to February 2008, which covered the in situ observation. The model was forced by observed winds, tidal elevation at open boundaries, and river discharges from eight inlets in the Pearl River Estuary. The agreements with observation were verified in terms of current, salinity and tide elevation. Using the computed tidal current, the residual current was also calculated to estimate material transport. In addition, the Lagrangian method was used to track the trajectory of particles from the mouth of Humen and Modaomen inlets. By computing the flushing time of these two channels, the authors find that there exists close relationship between flushing time and saltwater intrusion. Potential impact of sea-level rise on saltwater intrusion was also studied. Sensitivity experiments indicate that sea-level rise can impact on the saltwater intrusion in the Pearl River Estuary, such as increasing salinity concentration and intrusion distance, especially during neap tide in winter (dry season).

Introduction

The Pearl River is the third longest river in China, after the Yangtze River and the Yellow River. It has three principal tributaries, namely the Xijiang River, Beijiang River and Dongjiang River. The Pearl River Estuary (PRE) is a micro-tide, partially mixed estuary located on the north shelf of the South China Sea. There are eight inlets in the PRE: Humen, Jiaomen, Hongqimen, Hengmen, Modaomen, Jitimen, Hutiaomen and Yamen, as shown in Figure 1. The largest source of freshwater discharge into the PRE is through Modaomen, with monthly averaged discharges ranging from about 5522 m3s−1 in wet season to 777 m3s−1 in dry season (Wong et al., 2003). Tides and freshwater inflows are two major external forcing mechanisms that control the hydrodynamic processes in the PRE. Another important factor that governs the circulation in the PRE is the surface wind stress. In the PRE, wind alternates between the northeasterly monsoon in winter and the southwesterly monsoon in summer. Extensive efforts have been made to understand the basic hydrographic characteristics of the PRE, such as tide elevation (Ye et al., 1986; Yang et al., 1999; Li et al., 2000; Mao et al., 2004), tide current and circulation (He, 1986; Wong et al., 2004). Moreover, many researchers in China have done much work on salinity and temperature distribution in the PRE (Wong et al., 2003; Dong et al., 2004).

Figure 1.

Bathymetry of the Pearl River Estuary and its adjacent coastal area. Major inlets are marked. Numbers 1–6 denote Gaoyao, Shijiao, Guangzhou, Sanzao, Dahengqing, and Denglongshan, respectively. (Color figure available online.)

Figure 1.

Bathymetry of the Pearl River Estuary and its adjacent coastal area. Major inlets are marked. Numbers 1–6 denote Gaoyao, Shijiao, Guangzhou, Sanzao, Dahengqing, and Denglongshan, respectively. (Color figure available online.)

Saltwater intrusion occurs and has become a serious environmental problem in the PRE, as the river discharge decreases and tidal mixing relatively strengthens in winter. It provides a potential threat to residual water supply and urban industrial production. The saltwater intrusion in the PRE has received more and more attention due to both economic and scientific concerns. However, there are few numerical simulations to investigate the influence of river discharge and sea level rising on saltwater intrusion in the PRE. In this paper, we present results from a three-dimensional numerical simulation of the PRE for the period covering December 2007 to February 2008, a period of low river discharge. This numerical model is calibrated and verified using in situ data during December 2007 to January 2008, including a spring and neap tide cycle. We focus on the following processes in the estuary: (1) estuarine salinity distribution, (2) impacts of lateral inflow, and (3) variation of saltwater intrusion caused by continually rising sea level.

Field observations

The investigation period is divided into two phases, spring tide period (December 26 to 28, 2007) and neap tide period (December 31, 2007 to January 2, 2008) in the PRE. Observed hydrodynamic parameters include water temperature, salinity and currents. Moreover, biological and chemical elements were observed, such as nitrate, phosphate, dissolved oxygen, and so on. Three ships were used during each period. There were three cross sections (L1, L2 and L3) and four time series stations (M1, M2, H1, and H2) as shown in Figure 2. Both hydrodynamic and bio-chemical data were collected for the cross sections, but only hydrodynamic observations were obtained for the time series stations. Current and salinity measurements for the four stations (M1, M2, H1, and H2) were taken on four small boats using RCM9 ocean current meter and sbe 37-SM CTD. The vertical measurements were taken for three layers, namely near-surface layer, middle layer and near-bottom layer. Measurements were repeated at about 1-hour interval at each station for 50 hours in each period.

Figure 2.

Location of tidal gauge stations, time series stations and vertical cross sections of observation.

Figure 2.

Location of tidal gauge stations, time series stations and vertical cross sections of observation.

Model setup

Model

Environmental Fluid Dynamic Code (EFDC) developed by Hamrick (1996) is used in this study. EFDC is a public-domain modeling package for simulating three-dimensional flow, transport and biogeochemical processes in surface water system. The model solves the three-dimensional, vertically hydrostatic, free surface, turbulent averaged equations of motion for a variable density fluid. The two turbulence transport equations implement the Mellor-Yamada level 2.5 turbulence closure scheme (Mellor and Yamada, 1982; Galperin et al., 1988). The EFDC uses orthogonal curvilinear coordinates in the horizontal and a sigma coordinate transform in the vertical. The detailed descriptions about this model can be found in various publications, such as Ji et al. (2007) and Xu et al. (2008). The EFDC has been widely applied to bays and estuaries (Kuo et al., 1996; Lin et al., 2007; Xu et al., 2008).

Model configuration and initialization

The EFDC was configured as a domain that includes eight inlets, Lingdingyang Bay and the adjacent shelf area to about 60–100 m depth, which is approximately from 112.6°E to 115.5°E and from 21.1°N to 23.1°N (Figure 3). The bottom topography in the Pearl River was interpolated from a data set provided by Sun Yat-Sen University in Guangzhou, China. There were 1532 observed sections for depth data perpendicular to river banks in the Pearl River. In the Lingdingyang Bay and the adjacent sea, depth data are interpolated from etopo2 (2-minute gridded global relief data), having a high resolution of 0.5 km × 0.5 km to represent the actual bottom topography.

Figure 3.

Model grid.

Figure 3.

Model grid.

Based on features in the study area, a curvilinear orthogonal grid is used to represent the complex geometry of the PRE. The model grid consists of 19,655 grid cells in the horizontal direction. The horizontal resolution is about 0.1 km inside the inlets and is approximately 3 km at the coast of Guangdong. There are nine sigma layers in the vertical direction, with finer resolution near the surface and bottom. Time steps used in integrating the model is 30 seconds.

In order to obtain the initial value of salinity, the model was firstly spun up for one year using climatologic data interpolated from WOA01 salinity data and included M2 tide only at the open boundary. The mean river discharges during dry and wet seasons are the same as those used by Wong et al. (2003) during this spin-up period. This spin-up time is sufficient for the salt content in the estuary to reach a near-equilibrium state, which is independent of the initial salinity distribution.

Surface and open lateral boundary conditions

Hourly wind records at Station Sanzao for the simulation period are obtained from the Pearl River Water Resource Conservancy, which is assumed to be spatially uniform for the entire estuary. The wind is generally from the northeast with speed rarely exceeding 6 ms−1. Owing to the lack of observed river discharge data at each inlet, we use the ratio of river discharge (Table 1), which is estimated from the Pearl River Water Resource Conservancy, to derive each inlet's discharge according to stream gauge stations at Shijiao and Gaoyao. The freshwater discharges in Xijiang and Beijiang rivers are measured daily at these two stations. Therefore, we could approximately estimate the total river discharges to the PRE using data from these two stations and derive each inlet's daily discharge.

Table 1.

Flow ratios(%) at eight inlets during 1999 and 2006.

HumenJiaomenHongqimenHengmenModaomenJidimenHutiaomenYamen
1999–2006 14.8 13.8 12.3 16.1 28.6 3.6 4.6 6.3 
HumenJiaomenHongqimenHengmenModaomenJidimenHutiaomenYamen
1999–2006 14.8 13.8 12.3 16.1 28.6 3.6 4.6 6.3 

There are three open boundaries respectively at east, south and west of the model domain. The tidal solution was forced by sea-level elevation at these open boundaries, derived from tidal records at Station Sanzao. Temperature and salinity at the open boundaries are fixed to be the climatological values interpolated from WOA01. In addition, radiation boundary condition is used, which allows gravity waves inside the model domain to leave the open boundaries freely.

Model Validations

Model-observation comparison method

Owing to irregular time intervals for sampling along the cross sections, it is difficult to compare the model result with measurement at exact observation time. Therefore, we only compare the simulated currents and salinity with measurements at the four time series stations. Model horizontal velocities are linearly interpolated to instrument depth at each station. A few overall statistics are computed for differences between the simulated and the observed data, in regard to seawater elevation, and magnitudes of vector velocity and salinity. Two primary indices are used to denote goodness of fit. One is the velocity difference ratio (VDR) defined as the ratio of the sum of squared magnitudes of the vector velocity differences to the sum of squared magnitudes of the observed velocities (Han et al., 2008). That is:

formula
where Vm is the simulated horizontal velocity and Vo is the observed horizontal velocity. If the VDR is lower, the result is better, with VDR = 0 being the exact agreement. Another is the salinity different ratio (SDR) defined as the ratio of the sum of squared salinity differences to the sum of squared magnitude of the observed salinity, that is:
formula
where Sm is simulated salinity and So is observed salinity.

Validation against tide gauge data

Hourly tidal elevation is gauged at Stations Guishan Island and Jiuzhugang near the river mouths, and at Station Neilingding for the Lingding Bay. Figure 4 shows computed and observed tides at three tide gauge stations from December 25 to 30, 2007. There are little difference between phases of simulation and observation, and the computed amplitude generally agrees well with the observed. The correlation coefficients are generally greater than 0.93 at the three stations.

Figure 4.

Time series of simulated (solid line) and observed (dotted line) surface tidal elevation at the three stations of Jiuzhugang, Neilingding and Guishandao in December 2007. (Color figure available online.)

Figure 4.

Time series of simulated (solid line) and observed (dotted line) surface tidal elevation at the three stations of Jiuzhugang, Neilingding and Guishandao in December 2007. (Color figure available online.)

Evaluation against salinity and moored current

The simulated salinity and current are evaluated against observed data at the four time series stations using the statistics method described in section 4.1. The results show that the model simulates salinity and current satisfactorily. In general, the model gives better simulation of salinity in the coastal waters than inside the estuary, and maximum errors occur at the stations near the inlets. We find that the model simulates salinity at Stations H2 and M2 well and SDR and VDR of those two stations are less than 0.07, but has relatively larger errors at H1 and M1. The largest discrepancy occurs at H1 (the SDR is approximate 0.5), which is located in the narrow channel of Hengmen with salinity strongly influenced by freshwater discharge. Errors in the freshwater source estimation from Hengmen can significantly affect the accuracy of the localized salinity simulation. Its salinity is underestimated in the model, which indicates that the discharge from Hengmen may be overestimated. In addition, model grid sizes are also too coarse to resolve this narrow channel.

Material transport and saltwater intrusion

Residual current and Lagrangian tracer

Residual current, which is generally obtained by averaging tidal velocities over tidal periods, could play a significant role in material transport. The residual current in the PRE mainly results from freshwater inflows, wind, density-driven flows and so on. In general, larger river discharge may result in greater residual current. Figure 5 shows that maximum residual currents occur near river mouths, such as in Modaomen and Hengmen channels, where residual current could reach 15–20 cm s−1. The freshwater inflows from Hengmen, Hongqimen and Jiaomen make the residual current on the west of Neilingding Island flow from northwest to southeast. Owing to the influence of topography and Coriolis force, the residual current in Humen Channel flows southward. With increasing traveling distance, the residual current turns from eastward to westward near the Neilingding Island.

Figure 5.

Simulated surface residual current in winter 2007. (Color figure available online.)

Figure 5.

Simulated surface residual current in winter 2007. (Color figure available online.)

In order to investigate the influence of residual current on transporting the pollutants, we released two Lagrangian particles in the model. The initial positions of the particles are shown on Figure 6. The computation result shows that the particles moved forward in a spiral way. Particle 1 is the indicator near Humen; it first moves from north to southeast and arrives at Neilingding Island in five days, then turns to southwest because of the topography. Particle 2 released near Hengmen has similar trajectory and pass west of Lantau Island, and finally flow out of the PRE. Both of these 2 particles are released on the surface layer, so their movements are mostly driven by the freshwater and surface wind stress. Horizontally, high salinity water occupied the eastern part of the PRE, while low salinity water occupied in the western side (Figure 7). The bottom water with a salinity > 5 psu mostly intruded into the eastern side of the PRE and reach the Neilingding Island. The salinity difference between the surface and bottom water was less than 3 psu, especially in the upper of the PRE. This is also in agreement with observations of Dong et al. (2004).

Figure 6.

Tracks for drifting Lagrangian particles in the PRE.

Figure 6.

Tracks for drifting Lagrangian particles in the PRE.

Figure 7.

The distribution of simulated tidally averaged surface (a) and bottom (b) salinity in the winter 2007.

Figure 7.

The distribution of simulated tidally averaged surface (a) and bottom (b) salinity in the winter 2007.

Flushing time

Flushing time is an important indicator of material transport in an estuary, such as the process of saltwater intrusion (Ji et al., 2007). It is usually defined as the time needed to replace the freshwater already in the estuary (freshwater volume) at the rate of freshwater inflow. There are some different ways to compute flushing time in the literature, but some commonly used formulas, such as the tidal prism and the Knudsen formula, may significantly underestimate the time needed (Ji et al., 2007). The formula used in this study is briefly illustrated below, while detailed introduction to this method can be found in Ji et al. (2007).

By its definition, the flushing time of an estuary, Tf, can be calculated as:

formula

where R is the rate of total freshwater inflow, and Vf is the freshwater volume of an estuary. Volume Vf can be calculated as:

formula

where So equals to seawater's salinity, S is the salinity in the estuary, denotes mean salinity in the estuary, and V equals to estuarine volume.

We compute the flushing time of Modaomen and Humen channels in the PRE from December 1, 2007 to January 30, 2008, and the results are shown in Figure 8. We can see that the flushing time is quite sensitive to freshwater inflow rate, which is the dominant force in determining the variation of flushing time in both Modaomen and Humen channels. During the spring tide period between Day 26 and Day 28, the flushing time of these two channels are very short due to high freshwater inflow rate. As the freshwater in both channels decreases, the flushing time increases during the neap tide between Day 31 and Day 33. To illustrate the relation of flushing time with saltwater intrusion in the estuary, the distances from a 15-psu isohaline to the mouths of Modaomen and Humen are presented in Figure 9. Figures 8 and 9 clearly indicate that the flushing time and the saltwater intrusion are closely related. The rapid downstream flushing is accompanied with the large inflow event, which results in short flushing time, between Day 20 and Day 27 in both inlets. This event flushes the 15-psu isohaline downstream about 8 and 12 km in Modaomen and Humen channels, respectively. On the other hand, tidal flows play an important role in moving salinity upstream. This can be seen from the relatively quick migration of salinity in both inlets upstream 15–20 km between Day 27 and Day 32, after the flood has flushed the salinity downstream.

Figure 8.

Flushing time of Modaomen and Humen channels from December 1, 2007 to January 30, 2008. (Color figure available online.)

Figure 8.

Flushing time of Modaomen and Humen channels from December 1, 2007 to January 30, 2008. (Color figure available online.)

Figure 9.

The distances from 15-psu isohaline to the mouths of Modaomen and Humen from December 1, 2007 to January 30, 2008. (Color figure available online.)

Figure 9.

The distances from 15-psu isohaline to the mouths of Modaomen and Humen from December 1, 2007 to January 30, 2008. (Color figure available online.)

A better understanding of vertical mixing and vertical distribution of salinity is obtained by analyzing model results in two vertical cross sections in the PRE, which approximate to the locations of the observed vertical sections (L1 and L3) and are shown in Figure 2. The analysis of saltwater intrusion in the cross sections is divided into two periods for spring and neap tides. Figures 10 and 11 show that tidal mixing plays an important role in affecting the stratification during the spring tide in winter. The maximum vertical differences of the salinity are about 5 psu (20 psu) during spring (neap) tide in the Humen Channel. It indicates that the salinity stratification, which is eroded during spring tide, is formed during neap tide due to weaker tidal-induced mixing. The comparison of saltwater intrusions in Modaomen and Humen inlets between neap tide and spring tide indicates that the salinity intrusion is stronger in spring tide in Humen, while it is the opposite in Modaomen, both are confirmed by observations of these two cross sections. These phenomena are caused by several factors, such as different topography conditions and river discharges. Modaomen Inlet, which has the largest river discharge among the eight inlets in the PRE, has fewer creeks than Humen Inlet. By computing the flushing time of these two inlets, we find that the flushing time of Humen is much longer than Modaomen and the average ratio of the flushing time of Humen to that of Modaomen is 1.2 in spring tide and 1.6 in neap tide.

Figure 10.

Simulated vertical distribution of salinity along cross section L1 (Humen) during spring (top) and neap (bottom) tides. (Color figure available online.)

Figure 10.

Simulated vertical distribution of salinity along cross section L1 (Humen) during spring (top) and neap (bottom) tides. (Color figure available online.)

Figure 11.

Simulated vertical distribution of salinity along cross section L2 (Modaomen) during spring (top) and neap (bottom) tides. (Color figure available online.)

Figure 11.

Simulated vertical distribution of salinity along cross section L2 (Modaomen) during spring (top) and neap (bottom) tides. (Color figure available online.)

Influence of sea-level rise on saltwater intrusion

The gradual rise of sea level is one of the most important aspects of climate change, especially because it is likely to accelerate in the future as global warming progresses (Mark et al., 2002). It has caused many environmental problems including saltwater intrusion in the PRE. Chinese scientists have studied extensively sea-level-rise rate in this area (Shi et al., 2008). Recent study indicates that the sea level of the PRE has a rising trend of 0.3 ± 0.05 cm per year over the period from 1993 to 2006, and the relative sea level in some parts of Guangdong coast may rise up to 30 and 50 cm in 2030 and 2050, respectively, if the factors of ground subsidence and sea-level fluctuation are taken into account (Shi et al., 2008).

The impact of sea-level rise on saltwater intrusion is a matter of public concern. Some scholars concluded that the distance of salinity intrusion would increase with maximum distance of about 4 km in Humen Channel and of 3 km in Modaomen Channel during spring tide in dry season, based on Ippen and Harloman diffusion theory and method (Li et al., 2000). These previous studies mostly focused on conceptual model and did not include some important factors such as river discharge and wind, when computing the distance of salinity intrusion. We conduct three experiments to examine the influence of sea-level rise on saltwater intrusion in the PRE, in which the level rises are assumed to be 20, 50 and 80 cm, respectively. Model configurations of these three runs are the same as the benchmark run, except that the mean sea levels at the three open boundaries are increased by 20, 50 and 80 cm, respectively. All other controlling factors, such as wind, heat fluxes and freshwater discharges are unchanged. Therefore, the sea-level rise is approximately equivalent to a water-depth increase in these experiments.

Figure 12 presents the isohaline of 0.25 psu in the benchmark run and the run of a sea-level rise of 50 cm. Table 2 lists the averaged salinity difference between the benchmark experiment and the three sea-level-rise experiments in the two channels during spring and neap tides. Graphic comparisons and statistical analyses indicate that sea-level rise can have significant impacts on the saltwater intrusion in the PRE, such as increased salinity concentration and distance of salinity intrusion. Table 2 shows that the differences between the benchmark run and the 20-cm sea-level-rise experiment are relatively small, but as the sea level increases, the differences become more obvious. The 80-cm sea-level-rise experiment shows that the sea level rise of 80 cm has a great influence on the saltwater intrusion in the PRE by in increasing the salinity of water and the length of intrusion. In addition, the impact of sea-level rise on salinity intrusion is relatively bigger during neap tide in the dry season, it may be due to the stronger stratification during neap tide.

Figure 12.

Variation of 0.25-psu isohaline caused by sea-level rise of 50 cm during spring tide. (Color figure available online.)

Figure 12.

Variation of 0.25-psu isohaline caused by sea-level rise of 50 cm during spring tide. (Color figure available online.)

Table 2.

Difference of average salinity and distance of salinity intrusion between the benchmark experiment and the three sea-level-rise experiments in the two channels during spring and neap tides.

20 cm50 cm80 cm
SectionSalinity (psu)Length (km)Salinity (psu)Length (km)Salinity (psu)Length (km)
L1       
 Spring 0.6 1.9 1.4 6.3 2.3 8.9 
 Neap 0.7 4.0 1.6 8.1 2.6 12.3 
L2       
 Spring 0.3 1.1 0.8 4.8 1.5 5.9 
 Neap 0.3 1.7 1.0 5.4 1.8 7.1 
20 cm50 cm80 cm
SectionSalinity (psu)Length (km)Salinity (psu)Length (km)Salinity (psu)Length (km)
L1       
 Spring 0.6 1.9 1.4 6.3 2.3 8.9 
 Neap 0.7 4.0 1.6 8.1 2.6 12.3 
L2       
 Spring 0.3 1.1 0.8 4.8 1.5 5.9 
 Neap 0.3 1.7 1.0 5.4 1.8 7.1 

Conclusions

To study saltwater intrusion and material transport in the PRE, a hydrodynamic model for the area was developed using observed sea level, wind and river discharge for the period December 2007 to February 2008. The model results were verified with the observations from the winter cruise in December 2007 and January 2008. Furthermore, the characteristics of residual current and the relation between flushing time and saltwater intrusion were examined, so was the influence of sea-level rise on salinity intrusion. Results of the simulations give new insights into the processes of saltwater intrusion and major factors affecting the intrusion, such as river discharge and tidal force. Freshwater inflow rate is the dominant force in determining the variation of flushing time in both Modaomen and Human channels. Flushing time is useful to analyze saltwater intrusion, and flushing time and saltwater intrusion have close relationship with each other. Horizontally, low salinity water occupied the western side of the PRE, while high salinity water intruded into the eastern side. Modeling results indicate that sea-level rise can have adverse impact on the saltwater intrusion in the PRE, and the adverse effect becomes greater as the sea level increases.

Acknowledgements

This study was supported by the CAS/SAFEA International Partnership Program for Creative Research Teams (No. KZCX2-YW-T001) and the National Natural Science Foundation (40625017, U0733002 and 40830851). We would like to thank the editor and two anonymous reviewers for comments that helped to improve the manuscript.

References

Dong, L., Su, J., Wong, L., Cao, Z. and Chen, J.
2004
.
Seasonal variation and dynamics of the Pearl River plume
.
Continental Shelf Research
,
24
:
1761
1777
.
Galperin, B., Kantan, L. H., Hassid, S. and Rosati, A.
1988
.
A quasi-equilibrium turbulence energy model for geophysical flows
.
Journal of Atmospheric Science
,
45
:
55
62
.
Hamrick, J. M.
1996
.
User's manual of the environmental fluid dynamics computer code
,
Virginia
:
Virginia Institute of Marine Science/School of Marine Science, the College of William and Mary
.
Special Report in Applied Marine Science and Ocean Engineering No. 331
Han, G., Lu, Z., Wang, Z., Helbig, J., Chen, N. and Yong, B.
2008
.
Seasonal variability of the Labrador current and shelf circulation off Newfoundland
.
Journal of Geophysical Research
,
113
:
C10013
doi: 10.1029/2007JC004376
He, H.
1986
.
Analysis of variation of tide of the Pearl River Delta
.
Renmin Zhujiang
,
1
:
22
29
.
Ji, Z., Hu, G., Shen, J. and Wan, Y.
2007
.
Three-dimensional modeling of hydrodynamic processes in the St. Lucie Estuary
.
Estuarine, Coastal and Shelf Science
,
73
:
188
200
.
Kuo, A. Y., Shen, J. and Hamrick, J. M.
1996
.
The effect of acceleration on bottom shear stress in tidal estuaries
.
ASCE Journal of Waterways, Port, Coastal and Ocean Engineering
,
122
:
75
83
.
Li, S. and Ao, D.
2000
.
Rise of sea level and change of saline water intrusion at the Pearl River Estuary
.
Renmin Zhujiang
,
6
:
42
44
.
Lin, J., Xie, L., Pietrafesa, L. J., Ramus, J. S. and Paerl, H. W.
2007
.
Water quality gradients across Albemarle-Pamlico Estuarine system: seasonal variations and model applications
.
Journal of Coastal Research
,
23
(
1
):
213
229
.
Mao, Q., Shi, P., Yin, K., Gan, J. and Qi, Y.
2004
.
Tides and tidal current in the Pearl River Estuary
.
Continental Shelf Research
,
24
:
1797
1808
.
Mark, F. M. and John, M. W.
2002
.
Sea level is rising: Do we know why?
.
PANS
,
99
:
6524
6526
.
Mellor, G. and Yamada, T.
1982
.
Development of a turbulent closure model for geophysical fluid problems
.
Review of Geophysics and Space Physics
,
20
(
4
):
851
875
.
Shi, X., Chen, T. and Yu, K.
2008
.
Sea-level changes in Zhjjiang Estuary over last 40 years
.
Marine Geology & Quaternary Geology
,
28
:
127
134
.
Wong, L., Chen, J., Xue, H., Dong, L., Su, J. and Heinke, G.
2003
.
A model study of the circulation in the Pearl River Estuary (PRE) and its adjacent coastal waters: 1. Simulations and comparison with observations
.
Journal of Geophysical Research
,
108
(
C5
):
3156
doi:10.1029/2002JC001451
Wong, L., Chen, J. and Dong, L.
2004
.
A model of the plume front of the Pearl River Estuary, China and adjacent coastal waters in the winter dry season
.
Continental Shelf Research
,
24
:
1779
1795
.
Xu, H., Lin, J. and Wang, D.
2008
.
Numerical study on salinity stratification in the Pamlico River Estuary
.
Estuarine, Coastal and Shelf Science
,
80
:
74
84
.
Yang, Q. and Luo, X.
1999
.
The secular trends of sea level change in LingdingYang Estuary
.
Scientia Geographica Sinica
,
19
(
2
):
125
127
.
Ye, J., He, Z. and Zhou, Z.
1986
.
Numerical model of salt transport under tidal current of the Pearl River Estuary
.
Renmin Zhujiang
,
6
:
7
15
.