Vol 436|25 August 2005|doi:10.1038/nature03935
LETTERS
Seasonal oscillations in water exchange between
aquifers and the coastal ocean
Holly A. Michael1, Ann E. Mulligan2 & Charles F. Harvey1
Ground water of both terrestrial and marine origin flows into
coastal surface waters as submarine groundwater discharge, and
constitutes an important source of nutrients, contaminants and
trace elements to the coastal ocean1–5. Large saline discharges have
been observed by direct measurements3,6–10 and inferred from
geochemical tracers11–13, but sufficient seawater inflow has not
been observed to balance this outflow. Geochemical tracers also
suggest a time lag between changes in submarine groundwater
discharge rates12,14 and the seasonal oscillations of inland recharge
that drive groundwater flow towards the coast. Here we use
measurements of hydraulic gradients and offshore fluxes taken
at Waquoit Bay, Massachusetts, together with a modelling study of
a generalized coastal groundwater system to show that a shift in
the freshwater–saltwater interface—controlled by seasonal
changes in water table elevation—can explain large saline discharges that lag inland recharge cycles. We find that sea water is
drawn into aquifers as the freshwater–saltwater interface moves
landward during winter, and discharges back into coastal waters as
the interface moves seaward in summer. Our results demonstrate
the connection between the seasonal hydrologic cycle inland and
the saline groundwater system in coastal aquifers, and suggest a
potentially important seasonality in the chemical loading of
coastal waters.
Studies that have directly measured submarine groundwater discharge (SGD) with networks of seepage meters show that much of the
discharging water has salinity near that of sea water6,7,9,10, yet seawater
inflow to coastal aquifers has not been observed in sufficient quantity
to explain the large saline outflow. Moore12 used natural radium as a
tracer of groundwater discharge to coastal waters and estimated that
SGD was ,100 m3 d21 per m length of shoreline along the southeastern US coast, equivalent to 40% of river discharge. From
consideration of the regional freshwater balance, Moore and
Church13 conclude that most of this discharge is seawater circulation.
Several studies that infer SGD from natural radium measured in
coastal waters reveal a seasonal pattern that is out of phase with the
recharge cycle. Radium measurements over several years along the
South Atlantic Bight indicate that discharge is larger in the summer
than in the winter and spring12, and monthly groundwater fluxes
estimated from radium measurements in Rhode Island exhibit a
distinct pattern that also peaks in the summer14. Recharge is lowest in
the summer where these studies were conducted along the eastern US
coast because the strong seasonal oscillation of evapotranspiration
peaks in the summer, dominating a weaker seasonal signal in
precipitation. Along the Ganges Delta, radium transported by SGD
is also out of phase with the recharge cycle, although the seasonal
pattern is shifted. There, radium flux to ocean water is largest in the
winter15, but recharge is highest in the summer because monsoonal
rains dominate evapotranspiration. However, natural tracers
measured in sea water provide little insight into the groundwater
dynamics that drive SGD, and most direct SGD measurements by
seepage meters or hydraulic gradients have been conducted in
temperate regions during the summer only7,9,10.
Both the missing source of saline ground water to coastal aquifers
and the seasonal pattern in total SGD can be explained by seasonal
exchange of saline water between aquifers and the coastal ocean. In
most coastal aquifers, freshwater discharge occurs throughout the
year because the water table remains above sea level (Fig. 1). In such
aquifers, the well-known Ghyben–Herzberg approximation16 predicts that the depth to the freshwater–saltwater interface below mean
sea level is 40 times the water table elevation above sea level, a factor
that results from the density difference between salt water and fresh
water. Thus, changes in the water table elevation could theoretically
be amplified by a factor of 40 in the interface depth, potentially
driving large fluxes of saline water between the subsurface and coastal
waters.
The Ghyben–Herzberg relation is only approximate for dynamic
groundwater systems because it assumes local hydrostatic conditions
and no mixing of salt and fresh water, and the theoretical amplification in the interface movement will decrease as dynamic equilibrium
is established. However, it remains true that motion of the water table
will drive interface movement. As recharge lifts the water table, the
interface moves seaward, fresh water replaces salt water, more fresh
water is drawn from inland, and saltwater discharge is driven into the
Figure 1 | Saline groundwater circulation in a simple coastal groundwater
system with a Ghyben–Herzberg interface. Mechanisms include: (1) tidal
pumping, (2) nearshore circulation due to tides and waves, (3) saline
circulation driven by dispersive entrainment and brackish discharge, and (4)
seasonal exchange.
1
Department of Civil and Environmental Engineering, MIT, Cambridge, Massachusetts 02139, USA. 2Marine Policy Center, Woods Hole Oceanographic Institution, Woods Hole,
Massachusetts 02543, USA.
© 2005 Nature Publishing Group
1145
LETTERS
NATURE|Vol 436|25 August 2005
coastal ocean; the opposite set of fluxes occurs when the water table
falls. Our numerical simulations of density-coupled groundwater
flow and salt transport in unconfined coastal aquifers relax the
assumptions of the Ghyben-Herzberg relation, and predict that
seasonal recharge oscillations drive a saltwater flux in and out of
the aquifer that can be greater than the freshwater discharge (Fig. 2).
Sensitivity analysis confirms that saline discharge lags inland fresh
recharge under a variety of hydrogeologic conditions (Supplementary Figs 4 and 5). The hydraulic head in aquifers has been shown to
lag recharge in the field by 0–3 months (refs 17, 18) because recharging water takes time to percolate through the unsaturated zone, and
because the water table will continue to rise in response to positive
recharge past the time of peak recharge. Our numerical simulations
of the full dynamic system show that peak saline discharge may lag 1–
5 months behind peak recharge (Fig. 2, Supplementary Figs 4 and 5).
Seasonal exchange dominates dispersive circulation in every simulation, except when dispersivity and hydraulic conductivity are
largest. Thus, simulated saline inflow during the winter can explain
both a decrease in total wintertime discharge and the observed net
saline discharge in the summer. Seasonal hydrologic cycles occur in
many regions of the world19, and seasonal seawater exchange is
predicted for every set of parameters in our simulations of homogeneous unconfined coastal aquifers (Fig. 2 and Supplementary
Fig. 2). Furthermore, layered aquifer systems may exhibit enhanced
seasonal exchange due to an increase in the length of the freshwater–
saltwater interface (Supplementary Fig. 6). Thus, seasonal saline
exchange probably exists in a wide range of coastal systems.
To investigate the hypothesis of seasonal saltwater exchange,
vertical hydraulic gradients were measured during the winter (February 2004) in the sediments beneath Waquoit Bay, Massachusetts,
Figure 2 | Simulated total fresh and saline fluxes across the sea floor per
metre length of shoreline. Seasons are approximate for a typical yearly
recharge cycle within the US. Results are presented over one simulated year
for a fixed parameter set (aquifer thickness ¼ 100 m, annual average
recharge ¼ 0.001 m d21, longitudinal dispersivity ¼ 2 m and transverse
dispersivity ¼ 0.1 m) and two values of hydraulic conductivity, K:
a, K ¼ 5 £ 1024 m s21, and b, K ¼ 1 £ 1024 m s21. Dispersive circulation is
apparent as saline outflow during times of net inflow. Additional simulations
(see also Supplementary Figs 2–5) indicate that increasing either hydraulic
conductivity or aquifer thickness results in an increase in both seasonal
outflow and dispersive entrainment but reduces lag, increasing average
recharge reduces seasonal outflow compared to fresh outflow and decreases
lag, and increasing dispersivity increases dispersive circulation, slightly
decreases seasonal outflow, and has no effect on lag. Seasonal changes in
freshwater discharge are much less than in saltwater discharge or inland
recharge.
1146
when ice cover enabled piezometer installation but prevented the use
of seepage meters. Saline discharge (,70% of total discharge in
August 2003) has been extensively characterized by direct seepage
meter measurement during each summer from 1999 to 20017 and in
August 2002 and 2003 (this study) in the same location. The field
results (Fig. 3) indicate that the direction of saline flow is reversed
between the winter and summer. Net downward hydraulic gradients
from the bay into the aquifer, over a complete tidal cycle, were
measured at five locations corresponding to the band of highest
saline discharge observed during the summer.
Both the fresh and saline discharge observed at Waquoit Bay in the
summer are probably a result of flow in the unconfined aquifer, and
seasonal interface motion explains the saline component of summer
discharge as well as winter inflow. The timing is shown through
analysis of meteorological data from Long Pond in Falmouth,
Massachusetts20, which indicates positive recharge from late autumn
to early spring during the study period (1999–2004), and greater
evapotranspiration than precipitation from late spring through to
early autumn (Supplementary Fig. 7). Water levels in wells screened
at varying depths in the unconfined aquifer within 6.5 km of the
study site21 over the same period follow a yearly cycle, peaking in
April and dropping to a minimum in December. This time lag
between the recharge maximum and the hydraulic head maximum
is consistent with the numerical results.
Several mechanisms other than seasonal exchange have been
hypothesized to drive saltwater circulation2,5,10,22–24, including tides,
wave run-up on the beach, and dispersive entrainment of saline water
into fresh discharge. These processes will not drive the seasonal shifts
from net outflow to inflow that we observe because they produce
inflows and outflows that balance over shorter timescales. Furthermore, the circulation produced by these processes (labelled 1, 2 and 3
in Fig. 1) appears to be of secondary importance to seasonal upland
forcing at our field site. Tidal pumping (process 1 in Fig. 1) drives
water into the aquifer at high tide and out at low tide. Because elastic
storage contributes little water over the timescale of tides, significant
tidal pumping occurs only near enough to shore such that water table
movement can accommodate inflows and outflows. In Waquoit Bay,
tidal pumping occurs in a zone less than 15 m from shore, but net
inflow has not been observed in this zone over a tidal cycle, and the
Figure 3 | Submarine groundwater discharge into Waquoit Bay,
Massachusetts. August seepage meter measurements reveal primarily
saline discharge over a tidal cycle in both 2002 and 2003. In 2003, net saline
discharge was 7.5 m3 d21, and net fresh discharge was 3.5 m3 d21 per m
length of shoreline. February hydraulic gradient measurements indicate
saline inflow at the location where peak outflow was measured during
summer. An upward hydraulic gradient and fresh pore water were observed
beyond 50 m from shore beneath a low-permeability muck cap and are
discussed in Supplementary Information (Supplementary Fig. 12).
© 2005 Nature Publishing Group
LETTERS
NATURE|Vol 436|25 August 2005
magnitude of exchange is much less than the saline discharge
observed farther from shore.
Wave run-up and tides (process 2 in Fig. 1) create inflow of sea
water in the intertidal zone that discharges offshore23. This circulation was investigated in Waquoit Bay with novel seepage meters that
operate in the shallow intertidal zone (Supplementary Fig. 8), and
also with a sodium bromide tracer injected near the high tide mark.
The measured intertidal inflow was much less than that needed to
balance saline outflow, and the tracer test results show that water that
enters the sediment at high tide discharges between 2 and 3 m from
the position of high tide, much closer to shore than the bulk of saline
discharge (Supplementary Fig. 9). Furthermore, bromide tracer was
never detected at depths greater than 1.2 m, indicating that saline
circulation due to tides and waves is confined to the shallow intertidal
zone.
Finally, salt dispersion along the freshwater–saltwater interface can
drive large-scale saltwater circulation22 (process 3 in Fig. 1). Kohout24
estimated this flux to be roughly 10% of the seaward flow of ground
water at a Florida field site. In Waquoit Bay, however, net inflow
seaward of the high discharge zone was observed in only one location
during each August experiment, both times within measurement
error of zero flow, and most saline water discharges far offshore of the
freshwater interface.
Our characterization of the location and magnitude of the four
mechanisms in Waquoit Bay (Supplementary Fig. 10) indicates that
saline discharge due to the seasonal exchange mechanism
(,3.7 m3 d21 per m length of shore) is similar to or greater than
the total saline discharge due to other circulation mechanisms
(,3.3 m3 d21 m21) during the summer. In summary, three previously known mechanisms for saline circulation result in zero net
saline outflow over a tidal cycle and none can explain the large net
saline discharge observed in Waquoit Bay during the summer, leaving
seasonally-induced saline cycling as the likely explanation for our
observations.
Seasonal exchange from coastal aquifers may account for a large
component of SGD where recharge cycles are significant, whether
natural or anthropogenic in origin. The global extent of seasonal
saline exchange is currently unknown, but yearly recharge cycles
appear to be widespread, suggesting that seasonal interface movement could be a potentially important driver of saline water exchange
along many coastlines. SGD is known to transport nutrients and
contaminants that affect coastal ecosystems1,25, and the solutes
delivered by saline discharge can be as important (or more important) as those delivered by fresh SGD3,6. Seasonal exchange may affect
sediment chemistry by subjecting coastal sediments to flushing by
typically oxic sea water from above during inflow periods, and
flushing with often anoxic water from below during discharge
periods. In areas along the eastern US coast, the greatest saltwater
discharge may occur in the summer when biological activity is
maximum and river flow is minimum, so input of nutrients may
be of particular importance. Furthermore, the chemistry of saltwater
discharge may vary seasonally because the first saline water to
discharge has most recently entered the aquifer, and the last saline
water to discharge in the yearly cycle has had the longest subsurface
residence time. This work demonstrates the connection between the
inland seasonal hydrologic cycle and the saline groundwater system
in coastal aquifers, proposes a mechanism for saltwater circulation
within aquifers that results in net saline outflow during several
months each year, and suggests the potential for seasonality in
chemical loading of coastal waters.
METHODS
Numerical simulations. Two-dimensional variable-density simulations were
performed using the finite element model FEFLOW26. The simulated domain
extended 500 m landward and 200 m seaward from the shoreline for aquifers
20 m and 100 m thick (Supplementary Fig. 1), and the number of elements
ranged from 93,532 to 597,638. The aquifer was simulated as unconfined and
spatially homogeneous, with zero flow and zero mass transport boundary
conditions along the base and sides. The recharge boundary condition along
the landward model top varied sinusoidally in time, with an average value of
0.002 or 0.001 m d21 and amplitude of 0.0025 m d21. The sea-floor boundary
was a constant head with constant concentration where flow was inward, zero
concentration gradient where flow was outward. Eight simulations were run: two
are shown in Fig. 2, and six are presented in Supplementary Information
(Supplementary Table 1, Supplementary Figs 1–5), with varying hydraulic
conductivity, dispersivity, aquifer thickness and average recharge.
Seepage meters. During August 2002 and 2003, 20 conventional seepage
meters27, enough to overcome local variability7, were installed in Waquoit Bay.
The meters were aligned in two transects, 1 m apart, and extended 50 m into
Waquoit Bay perpendicular to the shoreline in the same location where
40 seepage meters were used in 1999 and 20007. Eight novel intertidal seepage
meters (Supplementary Fig. 8) were placed in the nearshore zone of this transect,
the locations varying with the position of the tide. Unlike conventional seepage
meters, these intertidal meters are not submerged, enabling measurement of
groundwater inflow and outflow in very shallow water depths. Seepage meter
bags were pre-filled with sufficient bay water to allow detection of either inflow
or outflow, and collected at least every two hours over one tidal cycle. Discharge
salinity was calculated from conductivity probe measurements in samples from
the conventional seepage meter bags before and after deployment, and by
refractometer measurements from porewater samples 3 cm below the sediment
surface next to each intertidal seepage meter. Groundwater flux and salinity
measurements agree well between adjacent conventional and intertidal seepage
meters.
Piezometers. In February 2004, 11 piezometers were placed through bay ice
along the summer seepage meter transect to depths of 0.6–0.9 m. The water levels
in each piezometer and the bay were measured with an electronic water level
meter approximately every 1–2 h during daylight hours. The salinity of piezometer samples and baywater profiles were measured every few hours with a
conductivity probe. The average vertical hydraulic gradient over a tidal cycle
from two days of data with similar tidal variation was calculated by correcting for
density differences in the bay and piezometer water columns. Slug test data along
this transect were converted to estimates of hydraulic conductivity (Supplementary Fig. 11). The density-corrected hydraulic gradients and interpolated
hydraulic conductivities were converted to flux estimates by Darcy’s law.
Measured seepage rates, both inflow and outflow, from seepage meters
correlate closely to hydraulic gradients measured in adjacent piezometers
when tested concurrently, so these summer seepage meter measurements and
winter piezometer results can be compared directly.
Tracer test. On 27 August 2001, a sodium bromide solution was injected into the
beach at the head of Waquoit Bay near the high tide mark, during high tide. The
0.243 M injection solution, with the density of seawater, 1.025 kg l21, was used to
track the movement of the infiltrating bay water. Twenty-one piezometers were
driven to depths of 0.3–1.4 m, nested in groups placed approximately every 0.6 m
to a distance 4.3 m bayward from the injection point. Small-volume porewater
samples were extracted every 1–2 h during daylight for four consecutive days, 32
sample times in all. A conductivity probe and a multi-meter connected to a
bromide electrode were used to measure the total conductivity (in mS cm21) and
Br2 concentration (in mV calibrated to mol l21) of each sample.
Received 25 March; accepted 9 August 2005.
1.
2.
3.
4.
5.
6.
7.
8.
Johannes, R. E. The ecological significance of the submarine discharge of
groundwater. Mar. Ecol. Prog. Ser. 3, 365–-373 (1980).
Moore, W. S. The subterranean estuary: a reaction zone of ground water and
sea water. Mar. Chem. 65, 111–-125 (1999).
Simmons, G. M. Importance of submarine groundwater discharge (Sgwd) and
seawater cycling to material flux across sediment water interfaces in marine
environments. Mar. Ecol. Prog. Ser. 84, 173–-184 (1992).
Slomp, C. P. & Van Cappellen, P. Nutrient inputs to the coastal ocean through
submarine groundwater discharge: controls and potential impact. J. Hydrol.
295, 64–-86 (2004).
Burnett, W. C., Bokuniewicz, H., Huettel, M., Moore, W. S. & Taniguchi, M.
Groundwater and pore water inputs to the coastal zone. Biogeochemistry 66,
3–-33 (2003).
Kim, G., Lee, K. K., Park, K. S., Hwang, D. W. & Yang, H. S. Large submarine
groundwater discharge (SGD) from a volcanic island. Geophys. Res. Lett. 30
doi:10.1029/2003GL018378 (2003).
Michael, H. A., Lubetsky, J. S. & Harvey, C. F. Characterizing submarine
groundwater discharge: a seepage meter study in Waquoit Bay,
Massachusetts. Geophys. Res. Lett. 30 doi:10.1029/GL016000 (2003).
Robinson, M., Gallagher, D. & Reay, W. Field observations of tidal and seasonal
variations in ground water discharge to tidal estuarine surface water. Ground
Wat. Monit. Remediat. 18, 83–-92 (1998).
© 2005 Nature Publishing Group
1147
LETTERS
9.
10.
11.
12.
13.
14.
15.
16.
17.
18.
19.
20.
NATURE|Vol 436|25 August 2005
Smith, L. & Zawadzki, W. A hydrogeologic model of submarine groundwater
discharge: Florida intercomparison experiment. Biogeochemistry 66, 95–-110
(2003).
Taniguchi, M., Ishitobi, T. & Saeki, K. Evaluation of time-space distributions of
submarine ground water discharge. Ground Wat. 43, 336–-342 (2005).
Crotwell, A. M. & Moore, W. S. Nutrient and radium fluxes from submarine
groundwater discharge to Port Royal Sound, South Carolina. Aquat. Geochem. 9,
191–-208 (2003).
Moore, W. S. Large groundwater inputs to coastal waters revealed by Ra-226
enrichments. Nature 380, 612–-614 (1996).
Moore, W. S. & Church, T. M. Submarine groundwater discharge — Reply.
Nature 382, 122–-122 (1996).
Kelly, R. P. & Moran, S. B. Seasonal changes in groundwater input to a wellmixed estuary estimated using radium isotopes and implications for coastal
nutrient budgets. Limnol. Oceanogr. 47, 1796–-1807 (2002).
Moore, W. S. High fluxes of radium and barium from the mouth of the GangesBrahmaputra river during low river discharge suggest a large groundwater
source. Earth Planet. Sci. Lett. 150, 141–-150 (1997).
Bear, J., Cheng, A. H.-D., Sorek, S., Ouazar, D. & Herrera, I. in Seawater Intrusion
in Coastal Aquifers — Concepts, Methods and Practices (ed. Bear, J.) (Kluwer
Academic, Dordrecht, 1999).
Eltahir, E. A. B. & Yeh, P. A. J. F. On the asymmetric response of aquifer water
level to floods and droughts in Illinois. Wat. Resour. Res. 35, 1199–-1217 (1999).
Changnon, S. A., Huff, F. A. & Hsu, C.-F. Relations between precipitation and
shallow groundwater in Illinois. J. Clim. 1, 1239–-1250 (1988).
Tapley, B. D., Bettadpur, S., Ries, J. C., Thompson, P. F. & Watkins, M. M.
GRACE measurements of mass variability in the Earth system. Science 305,
503–-505 (2004).
Payne, R. Falmouth monthly climate reports. khttp://cis.whoi.edu/science/PO/
climate/index.cfml (accessed 30 July 2004).
1148
21. US Geological Survey National Water Information System (NWISWeb).
khttp://nwis.waterdata.usgs.gov/usa/nwis/gwlevelsl (accessed 30 July
2004).
22. Cooper, H. H. A hypothesis concerning the dynamic balance of fresh water and
salt water in a coastal aquifer. J. Geophys. Res. 64, 461–-467 (1959).
23. Li, L., Barry, D. A., Stagnitti, F. & Parlange, J. Y. Submarine groundwater
discharge and associated chemical input to a coastal sea. Wat. Resour. Res. 35,
3253–-3259 (1999).
24. Kohout, F. Cyclic flow of salt water in the Biscayne aquifer of southeastern
Florida. J. Geophys. Res. 65, 2133–-2141 (1960).
25. Valiela, I. et al. Transport of groundwater-borne nutrients from watersheds and
their effects on coastal waters. Biogeochemistry 10, 177–-197 (1990).
26. Diersch, H. J. G. Interactive, graphics-based finite-element simulation system
FEFLOW for modeling groundwater flow, contaminant mass, and heat
transport processes. (WASY Ltd., Berlin, Germany, 2002).
27. Lee, D. R. Device for measuring seepage flux in lakes and estuaries. Limnol.
Oceanogr. 22, 140–-147 (1977).
Supplementary Information is linked to the online version of the paper at
www.nature.com/nature.
Acknowledgements We thank the WBNERR staff and the many MIT and WHOI
faculty and students, USGS personnel, and others who assisted in the field. This
work was supported by a graduate research fellowship from the US National
Science Foundation.
Author Information Reprints and permissions information is available at
npg.nature.com/reprintsandpermissions. The authors declare no competing
financial interests. Correspondence and requests for materials should be
addressed to C.F.H. ([email protected]).
© 2005 Nature Publishing Group