Water Resources Research
Supporting Information for
Effects of alongshore morphology on groundwater flow and solute transport in a
nearshore aquifer
Ying Zhang1, Ling Li1, Dirk V.Erler2, Isaac Santos2, David Lockington1
1School
2Centre
of Civil Engineering, The University of Queensland, St. Lucia, QLD, Australia
for Coastal Biogeochemistry, School of Environmental Science and Management, Southern Cross
University, Lismore, NSW, Australia
Contents of this file
Text S1 to S2
Figures S1 to S5
Tables S1
Introduction
Six additional simulations were conducted to assess (1) whether the agreement with data
could be improved by adjusting parameters other than hydraulic conductivity and
allowing heterogeneity, and (2) correspondingly the effects of these factors on
groundwater flow and solute transport in the studied system. In addition, the 3D
simulation results were compared with the predictions of a 2D model to identify the
limitations of the 2D model in simulating groundwater hydrodynamics in nearshore
aquifers where complex beach morphology and hydraulic conditions exist. Please note
these discussions are aimed to address how the parameters affect the agreement with data
and 3D features rather than conducting a rigorous sensitivity analysis (which is not the
purpose of the present paper).
1
Text S1.
Six additional simulations (Table S1) were conducted to examine the effects of six
parameters on the groundwater level fluctuations and salt distribution. The six parameters
included four properties of the aquifer (porosity, dispersivity, capillary fringe and
heterogeneity), the inland hydraulic head and tidal oscillations. The simulated water table
fluctuations were compared with the recorded data (Figure S1). The averaged salinity
distribution given by each model was used to explore the effects of these factors on threedimensional saline plumes in the aquifer (Figure S2).
The value for aquifer porosity was reduced to 0.3 (Case 1), which is a typical value for
sands [Das, 2008]. The smaller porosity slightly intensified groundwater table
fluctuations in landward piezometers (i.e. R1 and R3 in Figures S1a and S1b). This is
because a decrease in porosity reduces the value of wave number and weakens the
dampening of tide-induced groundwater wave (water table fluctuations). The porosity
value also affected the development of relatively salty plumes (Figure S2a). The saline
plume bounded by the 30 ppt isoconcentration line penetrated deeper compared to its
counterpart in the base case. As a smaller porosity increases the pore water flow velocity,
the advective solute transport is enhanced.
Simulation results (Case 2) showed that the parameter a in the saturation-capillary
pressure relationship had negligible effects on the water table fluctuations and salinity
distribution. When the value of a was decreased from 14.5 m-1 to 4 m-1, the amplitude of
groundwater table fluctuation had a minor increase due to capillary rise (around 1 cm,
Figure S1). The averaged salinity distribution was very similar to the one in the base case
(Figure S2b).
The longitudinal dispersivity affected the sea water intrusion and the development of
saline plumes. The longitudinal dispersivity was decreased to 0.25 m (Case 3) and 0.1 m
(Case 7). Compared with the salinity distribution in the base case, small dispersivity
value resulted in a further intrusion of the cross-shore saltwater wedge (2 ~ 3 m
landward, Figures S2c and S2g). The saline plume bounded by the 30 ppt
isoconcentration line had a three-dimensional expansion, indicating an intensified
2
density-dependent flow locally. The averaged salinity distributions simulated with 𝛼𝐿 =
0.25 m and 𝛼𝐿 = 0.1 m were very similar to the base case results except that the saline
plume bounded by the 30 ppt isoconcentration line expanded further downward.
To capture the effects of instantaneous surface waves, the measured tidal data was
specified as seaward boundary instead of the synthetic signal in Case 4. For the
piezometer in the intertidal zone (R7), the model could replicate the rapid surface
pressure changes (Figure S1e). However, the under-prediction of water levels at R6 and
R13 remained. In addition, the averaged salinity distribution under measured tidal data
was very similar to the one under the synthetic signal.
As the base case over-predicted the mean water table in the landward zone, the inland
head was decreased to 0.8 m to achieve a better agreement with the measured water
levels at R1 and R3 (Case 5). With a smaller inland head, the saltwater wedge intruded
further (Figure S2e). However, the creek-originated saline plume did not change
significantly.
We also considered the possible heterogeneity of the studied aquifer. A two-layer soil
stratigraphy was considered with the lower layer (below z = -6 m) set to be more
conductive (5 times larger than the calibrated conductivity) (Case 6). Simulation results
showed that the amplitudes of water level fluctuations increased because the overall
conductivity was increased (Figure S1). As a constant inland head was specified, the
increased conductivity caused an increase in the inland flux. Consequently, the saltwater
wedge retreated (seaward) and the creek-originated saline plume was reduced (Figure
S2f). It can be inferred that the creek-originated saline plume expands and the saltwater
wedge intrudes further inland when the lower layer is less conductive.
In summary, the agreement between the simulation results and data could not be
approved significantly by adjusting other model parameters than hydraulic conductivity
and allowing a two-layer soil structure to represent the aquifer heterogeneity. In
particular, all the simulated cases under-predicted the water level fluctuations at R6 and
R13 and could not replicate the salinity value at the sampling ports. However, the
3
simulated results provided useful information about the effects of these factors on flow
and solute transport processes in this nearshore groundwater system. More detailed
geological survey is required. A flux-controlled creek boundary may also match the
reality better. The additional work is beyond the scope of the present paper but points to
directions for future research.
Text S2.
The 3D simulations showed that the alongshore morphological variations, including the
creek, influenced the tide-induced pore water flow in the modelled system significantly.
The effects are not represented and cannot be simulated by the 2D model. Here, we
explore further the differences between the simulation results given by the 3D and 2D
models. The comparisons were made with a focus on pore water flow and water fluxes
across the interface along the 2011 and 2012 transects, respectively. The setup of the 2D
model was based on cross-vertical sections along these two transects with the same
parameter values employed in the 3D model.
The phase-averaged salinity distributions and flow fields along the 2011 transect were
computed over a spring tidal cycle (Figure S3). The differences between salinity
distributions predicted by the 2D model and 3D model were mainly in following three
aspects: in the 3D model (1) the groundwater beneath the upper saline plume was more
saline; (2) the width of the saltwater wedge was enlarged and (3) the groundwater
discharging zone was wider. Although both the 2D model and the 3D model predicted
upper saline plumes of a similar extent and around a similar location, the salinity of
discharging groundwater beneath the upper saline plume was around 5 ppt in 2D model,
while in the 3D model the salinity of discharging groundwater reached 15 ppt, even in the
area landward of the high tide mark (x < -5 m). The creek-driven alongshore flow
appeared to inhibit the cross-shore seawater intrusion at the bottom of the aquifer,
resulting in a more seaward wedge toe in the 3D model than that in the 2D model. Local
net fluxes were also compared between the 2D and 3D simulation results (Figure S4).
Some similar trends were predicted by both models: tidally driven seawater infiltration
occurred landward of x = 0 m, peak outflow around x = 2.5 m and fresh groundwater
outflow of the same order of magnitude in the 2D and 3D models. However, the amount
4
of seawater outflow simulated by the 3D model was much larger despite a similar amount
of local seawater inflow predicted (again indicating contribution of alongshore flow and
salt transport).
The comparison along the 2012 transect (Figure S5) showed very different flow
behaviors and salinity distributions. In the 3D model, the seaward dispersion zone was
much wider and the saltwater wedge toe intruded the aquifer 30 m further landward. The
increased saltwater intrusion was due to the alongshore salt transport caused by the
combined effects of the tide and creek. A further evidence of the alongshore salt transport
was a high concentration saline plume formed within the saltwater wedge around x = -5
m. This feature could not be explained on the basis of cross-shore processes and thus was
not simulated by the 2D model. In the 2D model, the discharging groundwater above the
SW was largely originated the landward boundary. In the 3D model, the discharge
location was shifted seaward, and the sources of discharging water were from both the
freshwater input through the landward boundary and saltwater that infiltrated the beach
surface and the creek boundary. As the setup of the 2D model was based on the
assumption of a perpendicular transect to the shoreline with the neglect of alongshore
variability, it is not capable of simulating any features associated with the alongshore
flow and solute transport, especially in the near-creek area where the 2012 transect was
located.
References
Das, B. M. (2008), Advanced soil mechanics, 3rd ed., xxvi, 567 p. pp., Taylor & Francis,
London ; New York.
5
Figure S1. Comparison of simulated heads and measured heads.
6
Figure S2. Averaged salinity distribution over a spring-neap tidal cycle with iso-surfaces of specified
salinities shown.
7
Figure S3. Compasison of phase-averaged salinity distributions and flow fields between the 2D model
and 3D model along the 2011 transect. Vectors show the mean flow fields. Dash lines indicate the
mean sea level.
8
Figure S4. Simulated phase-averaged fluxes across the beach surface by the 2D and 3D models.
Figure S5 Comparison of phase-averaged salinity distributions and flow fields between the 2D model
and 3D model along the 2012 transect. Vectors show the mean flow fields.
Table S1. Summary of additional simulation cases.
Considered parameter
Value
Value in base case
Case 1
Porosity
0.3
0.65
Case 2
Parameter a in equation (4b)
4 m-1
14.5 m-1
Case 3
Longitudinal dispersivity
0.25 m
0.5 m
Case 4
Tidal signal
Measured tides
Synthetic tides
Case 5
Inland head
-0.6 m
0.2 m
Case 6
Heterogeneity
Two layers with different
Homogeneity
hydraulic conductivities
Case 7
Longitudinal dispersivity
0.1 m
0.5 m
9