Transcript

Characterisation of the hydrology of an estuarine wetland

Catherine E. Hughes, Philip Binning*, Garry R. Willgoose

Department of Civil, Surveying and Environmental Engineering, The University of Newcastle, Callaghan, N.S.W. 2308, Australia

Received 3 March 1998; received in revised form 24 June 1998; accepted 7 July 1998

Abstract

The intertidal zone of estuarine wetlands is characterised by a transition from a saline marine environment to a freshwaterenvironment with increasing distance from tidal streams. An experimental site has been established in an area of mangroveand salt marsh wetland in the Hunter River estuary, Australia, to characterise and provide data for a model of intertidal zonehydrology. The experimental site is designed to monitor water fluxes at a small scale (36 m). A weather station and ground-water monitoring wells have been installed and hydraulic head and tidal levels are monitored over a 10-week period along ashort one-dimensional transect covering the transition between the tidal and freshwater systems. Soil properties have beendetermined in the laboratory and the field. A two-dimensional finite element model of the site was developed using SEEP/W toanalyse saturated and unsaturated pore water movement. Modification of the water retention function to model crab holemacropores was found necessary to reproduce the observed aquifer response. Groundwater response to tidal fluctuations wasobserved to be almost uniform beyond the intertidal zone, due to the presence of highly permeable subsurface sedimentsbelow the less permeable surface sediments. Over the 36 m transect, tidal forcing was found to generate incoming fluxes in theorder of 0.22 m3/day per metre width of creek bank during dry periods, partially balanced by evaporative fluxes of about0.13 m3/day per metre width. During heavy rainfall periods, rainfall fluxes were about 0.61 m3/day per metre width, dom-inating the water balance. Evapotranspiration rates were greater for the salt marsh dominated intertidal zone than the non-tidalzone. Hypersalinity and salt encrustation observed show that evapotranspiration fluxes are very important during non-rainfallperiods and are believed to significantly influence salt concentration both in the surface soil matrix and the underlying aquifer.q 1998 Elsevier Science B.V. All rights reserved.

Keywords:Wetlands; Salt marshes; Tidal flats; Water balance; Unsaturated zone; Hunter Valley

1. Introduction

Estuarine wetlands are characterised by complexinteractions between vegetation type, surface waterfluxes and porewater movement. The hydrology oftidal wetlands is very sensitive spatially to smallchanges in topography and associated tidal regime.In addition to tidal fluxes, several other factors play

an important role in the hydrology of wetlands. Theseinclude vegetation, rainfall, seasonal variations inevapotranspiration, extreme tidal or flood events,and variations in regional groundwater flow.

Mitsch and Gosselink (1993, p. 68) have noted that‘‘ hydrology is probably the single most importantdeterminant of the establishment and maintenanceof specific types of wetland’’ (their italics). In parti-cular, hydrology is a key determinant in species dis-tribution, in wetland productivity (biomass producedper unit time), and nutrient cycling and availability.

Journal of Hydrology 211 (1998) 34–49

0022-1694/98/$ - see front matterq 1998 Elsevier Science B.V. All rights reserved.PII: S0022-1694(98)00194-2

* Corresponding author. Fax: +61-2-4921-6991; e-mail:[email protected]

To be able to understand the ecology of these envir-onments it is crucial to understand the hydrology. Thereverse is also true, with ecology being critical todetermining the hydrological balance. Because ofthe intimate relationship between estuarine wetlandecology and hydrology, knowledge of wetland hydrol-ogy is critical if we are to predict and manage changein wetland environments. These include both long-term gradual changes such as climate change andprojected sea-level rise, and sudden changes resultingfrom human interference e.g. hydraulic modificationof tidal flow.

The bulk of research on salt marsh hydrology hasconcentrated on theSpartina alternifloradominatedmid and high latitude salt marshes of the NorthernHemisphere, which typically occupy the entire inter-tidal zone. Groundwater and porewater fluxes havebeen found to be important factors affecting wetlandproductivity, through their influence on accumulationand removal of chloride, nutrients and toxins, hyper-salinisation, sediment oxidation potential, pH and soilmoisture content (Chalmers, 1982; Howes et al.,1986; Nuttle and Harvey, 1988). Studies of porewaterflow in salt marsh sediments have focused on solute

fluxes (e.g. Yelverton and Hackney, 1986; Howes andGoehringer, 1994; Nuttle and Harvey, 1995), creekbank drainage following tidal inundation (e.g. Nuttleand Harvey, 1988; Harvey et al., 1987), porewatertransport mechanisms (e.g. Harvey and Nuttle, 1995;Harvey et al., 1995), and regional groundwater dis-charge (e.g. Nuttle and Harvey, 1995).

In the Hunter Region of Australia, salt marsh isoften found in the upper intertidal zone in conjunctionwith mangroves, which dominate the lower intertidalzone. To date research in Australian tidal wetlands hastended to concentrate on mangal or mangrove com-munities and has largely neglected the adjacent saltmarsh zone. Consequently, little is known about thesimilarities and differences in hydrology betweenAustralian and Northern Hemisphere salt marshenvironments.

This paper presents the results of a field investi-gation of the hydrogeology at a site at TomagoSouth in the Hunter River estuary, Newcastle,Australia (Fig. 1). Measurements of water tableresponse to tidal forcing from salt marsh creekshave generally found a rapid decline in water tablemovement with increasing distance from the creek,

Fig. 1. Contour and vegetation map of study site with well locations; map of Tomago South rehabilitation site, Hunter River, Newcastle(inset).

35C.E. Hughes et al. / Journal of Hydrology 211 (1998) 34–49

with vertical movement becoming negligible within 5to 10 m (Howes and Goehringer, 1994; Nuttle andHarvey, 1988). In this study, the groundwaterresponse to tidal inundation extends over a greaterdistance because of the characteristics of the under-lying sand aquifer and the location of the study site inrelation to the creek. In contrast to previous studies,groundwater response beyond the normal intertidalzone has been observed to remain almost uniformwith increasing distance from the creek. To under-stand the soil water and salt balance beyond theimmediate creek bank zone it is therefore necessaryto be able to predict water table response to tidalforcing as well as to rainfall and evapotranspiration.

In this study measurements of hydraulic head,climate, and soil and aquifer properties were made,aquifer response to tidal forcing is analysed andtemporal and spatial scale issues relating to ground-water in a wetland are discussed. A finite elementmodel was developed in order to further understandthe relative contribution of tidal forcing, evapo-transpiration and rainfall to saturated and unsaturatedflow.

2. Site description

The Tomago South wetlands are located in an inter-barrier depression overlying an Inner Barrier sand unit(Thom et al., 1992). The aquifer flowing through thesand unit, known as the Tomago Sandbeds, is animportant groundwater resource for the area. Duringthe past 3000 years mud flats have developed on topof the sand unit so that the surface soils compriseestuarine muds upon which the tidal wetlands of thearea have established (Thom et al., 1992). Located atthe edge of this estuarine mud unit in the floodplain ofthe Hunter River, the study site has a surface layer ofonly 0.5 to 1.0 m of muds overlying the Inner Barriersand unit which is estimated from regional bore holedata to be approximately 20 m thick (Woolley et al.,1995). The stratigraphy of the study site is shown inFig. 2.

Salt marsh and mangrove wetlands in the HunterRiver estuary have suffered varying degrees of distur-bance since European settlement in the early 1800s.Wetlands at Tomago South have been largelyreclaimed for agriculture through the excavation of

Fig. 2. Stratigraphic cross-section and well locations at the study site.

36 C.E. Hughes et al. / Journal of Hydrology 211 (1998) 34–49

a series of drains and construction of a levee bank withone-way floodgates. A portion of the original wetland,fringing the Hunter River, has been allowed to remaintidal but has also been modified by past grazing,mangrove clearance and drain construction. Approxi-mately half of the area is managed by the NSWNational Parks and Wildlife Service and the remain-der is part of an industrial buffer zone owned andmanaged by the Tomago Aluminium Company.Under the auspices of the Kooragang WetlandRehabilitation Project, the entire area of approxi-mately 11 km2 has been identified for wetland restora-tion and rehabilitation, with emphasis on habitatcreation for migratory water birds and fisheries, com-bined with recreational and educational opportunities.

The study site is located approximately 500 m fromthe Hunter River (Fig. 1, inset) which experiencessemidiurnal and mixed tidal fluctuations with anormal maximum range of about 2 m. Tidal range atthe study site is attenuated significantly and is notsinusoidal, due primarily to a flow obstruction in thetidal creek. Only tides above 0.37 m AustralianHeight Datum (AHD) exceed the blockage and aretransmitted to the site, which is therefore often subjectto only a single tidal peak each day. Maximum eleva-tion at the site is 0.65 m above the minimum tidallevel and the normal tidal range is approximately0.5 m. Such a degree of tidal attenuation appears tobe typical of the salt marsh areas at Tomago.

Storm surges in the Hunter River have beenobserved to have a significant effect on tidal fluctua-tions resulting in tidal levels considerably exceedingnormal king tide levels. Historical reports of majorflood events indicate inundation of the entire flood-plain area at Tomago occurs at approximately 50-yearintervals with local flooding occurring at 10-yearintervals (PWD, 1994).

Vegetation at the site includes salt marsh speciesSporobolus virginicus, Sarcocornia quinqueflora,Triglochin striataandSuaeda australisand mangrovespecies Aegiceras corniculatum and Avicenniamarina. Salt tolerant grass and weed species arefound at the fringe of the intertidal zone. Salt marshand mangrove species distribution correspondsbroadly with degree of tidal inundation (Fig. 1) withmangroves adjacent to waterways.

Animal burrows, particularly crab holes, are adominant feature of soils in the intertidal zone.

These have been found to dramatically increase sur-face infiltration rates in other mangrove and salt marshareas (Clarke and Hannon, 1967; Harvey and Nuttle,1995), increase accumulation of porewater solutes inthe soil matrix (Harvey et al., 1995) and form a sig-nificant pathway for tidal flow (Ridd, 1996; Wolanskiet al., 1992).

3. Data collection

Groundwater behaviour at the site is controlled by acombination of periodic tidal and evapotranspirationfluctuations, irregular rainfall events and possiblyregional groundwater flow.

In order to characterise groundwater response toshort term tidal and climatic forcing, a series ofwells was installed at the site at locations shown inFigs. 1 and 2. Six small 56 or 70 mm diameter wells(A1–A6) were placed to depths ranging from 0.8 to1.4 m from the surface along a one-dimensional trans-ect from the creek across the highest part of the site toa shallow tidal depression, a nested set of 22 mmwells (P2–P4) was installed at 2, 3 and 4 m depthsadjacent to well A3, and two 22 mm wells (A3B,A6B) were placed perpendicular to the 1D transectat 3 m distance from, and at similar depths to, A3and A6, respectively. Wells were screened over thebottom 30 cm and located in the more permeable sandlayers in order to ensure a rapid borehole response tochanging groundwater conditions. The vertical loca-tion of the wells is shown in Fig. 2.

An automatic climate station was placed approxi-mately 300 m from the study site. The stationmeasured rainfall, temperature and humidity, windspeed and direction and incoming short wave solarradiation data averaged on an hourly basis. Potentialevaporation was calculated using the Penman com-bination equation (Maidment, 1993, p. 4.16).Measurements of energy balance components madeusing eddy correlation apparatus during August,September and October 1997 enabled the actualevapotranspiration of three vegetation types at thestudy site to be estimated. Daily pan evaporationand rainfall values from Williamtown AirportMeteorological Office (AMO), approximately 10 kmto the north-east, were also acquired for comparisonand for a longer-term record.

37C.E. Hughes et al. / Journal of Hydrology 211 (1998) 34–49

Continuous monitoring of head levels in the wellsA1, A3 to A6 was carried out in a maximum of threewells at any one time over two periods from 16/1/97 to4/3/97 and 11/3/97 to 1/4/97. Measurements weretaken at 15 min intervals using pressure transducerspowered by rechargeable batteries and recorded witha variety of dataloggers. During these periods waterlevel in the creek adjacent to the site was alsomonitored. Electronic measurements were calibratedusing manual data collected at intervals of 3 days to1 week with standard errors in the order of 1 to 3 mmcalculated.

Hydraulic conductivity was estimated using a vari-ety of methods. Slug tests were performed in four ofthe six large wells. Measurements of infiltration andhydraulic conductivity of surface sediments weremade in the field using double ring infiltrometersand a Guelph permeameter. In the laboratory, fallinghead tests were performed on samples from the samelocations, bulk density and moisture content weredetermined gravimetrically, particle size analysiswas undertaken using the hydrometer method, organicmatter content was determined by ignition at 4508Cand porosity was estimated based on the assumptionthat organic matter density is 0.224 g/cm3 and mineraldensity is 2.65 g/cm3 (Maidment, 1993, Eq. 5.1.1 andp. 5.35).

4. Results and discussion

4.1. Soil properties

The basal sediments are medium grained silty sandswith clay lenses and thin layers of shell, and thesurface sediments are highly organic fine silt sandsor ‘‘estuarine mud’’. Due to capillary rise in the fine

mud layer and regular tidal inundation, the surfacemoisture content remains at or near saturation in thelower intertidal zone. Above the lower intertidalzone the water table is further from the surface,therefore, the surface moisture content is more vari-able and is influenced by the action of rainfall andevapotranspiration.

Hydraulic conductivity determined from slug testresults ranged from 0.54 to 26.7 m/day, and averaged16 m/day for the silty sand and 0.7 m/day for the claysand layer underlying the muds in the low marsh.Results from Guelph permeameter, falling head anddouble ring infiltrometer tests at seven locations werein good agreement and are summarised in Table 1.Based on these results, surface hydraulic conductivitywas estimated to be approximately 0.01 m/day for theestuarine mud matrix and 1 m/day for non-tidal top-soils. The presence of crab holes and smaller macro-pores appears to increase the overall surfaceinfiltration rate to a range of 0.1 to 1 m/day, whichis 1 to 2 orders of magnitude larger than the matrixhydraulic conductivity. Individual crab hole infiltra-tion rates average 11 m/day. Ponded areas in thelower intertidal zone have virtually no measurableinfiltration.

Particle size analysis revealed that the surface sedi-ments are predominantly silty sands with clay contentranging from 6 to 12%. The bulk density of thesesediments ranged from 0.63 to 1.35 g/cm3, organicmatter comprised 8 to 22% and porosity varied from0.45 to 0.7. Sediment properties are summarised inTable 1.

Frequent tidal and soil moisture content fluctua-tions would be expected to induce a shrink–swellresponse of 10 to 20 mm in the highly porous surfacemuds, particularly in the less frequently inundatedzones (Fityus and Welbourne, 1996).

Table 1Soil property results for the four major sediment zones

Material location Sat hydraulicactivity

Initialinfiltration

Porosity Claycontent

Organicmatter

Bulkdensity

(m/day) (m/day) (%) (%) (g/cm3)

Subsurface sands 0.5–27a – 0.45 – – –Lower intertidal zone 0.01–0.06b 0–1.7 0.5–0.7 6–9 8–22 0.68–1.24Upper intertidal zone 0.01–0.17b 0.17–0.28 0.45–0.61 6–8 8–14 0.9–1.35Non-tidal zone 0.7–3.4b 4.5 0.71 12 19 0.63

a Before slug tests.bGuelph permeameter and falling head tests.

38 C.E. Hughes et al. / Journal of Hydrology 211 (1998) 34–49

The concentration of crab holes in the intertidalzone is approximately 38/m2 (1% of surface area)with diameters of 2–75 mm (mean diameter17 mm). Each of these crab holes is likely to havemultiple exits and they extend in depth from the sur-face to the lowest tidal levels of the water table. Asmall rim containing excavated material is commonlyfound encircling the entrance to each crab hole.Infiltration measurements at the study site haveshown infiltration rates as high as 1 m/day for asingle crab hole located in a low permeability(0.01 m/day) soil matrix. Crab hole and othermacropore flow is a major contributor torainwater and tidal infiltration while matrix flowdominates the process of water table drainage andevapotranspiration.

The salinity of the groundwater at the siteincreases with depth, and has been measured to beup to 50 ppt or 1.5 times that of seawater at a depthof 4 m.

4.2. Water table response to tidal forcing

As can be seen in the typical examples of thehydraulic head data given in Fig. 3, hydraulic headmeasurements showed a large rapid response to tidalfluctuations across the entire site.

In Fig. 3, the ground surface is inundated by hightides at locations A4 and A6, whereas the ground sur-face at A3 is not. The hydraulic heads measured inpiezometers A4 and A6 are different from the surfacewater level as the piezometer is screened in the deepersand layer and not at the surface. The measured headsare similar in all piezometers. These observationssuggest that during tidal inundation there is littlehydraulic response through the low permeabilitymud to surface flooding, and that most of the observedhead change is due to tidally induced lateral flow ofwater in the underlying high permeability sand layer.

Vertical head gradients measured using the nestedwells P2 to P4 (10−3 to 10−2) were up to an order of

Fig. 3. Typical hydraulic head response to (A) spring tide forcing (8/2/97 to 10/2/97); tidal forcing and rainfall (10/2/97 to 12/2/97); and (B)neap tide forcing and rainfall (28/1/97 to 2/2/97). The legend shows the distance of each piezometer from the creek.

39C.E. Hughes et al. / Journal of Hydrology 211 (1998) 34–49

magnitude greater than horizontal gradients aroundA3 with maximum gradients occurring in a shorttime period following the tidal peak. This impliesthat vertical flow in the top 4 m of the sediment issignificant at the site. Because of these verticalgradients, water table location differs slightly frommeasured head. Horizontal gradients perpendicularto the transect, from A3 to A3B and A6 to A6B,were commonly less (10−3) than horizontal gradientsalong the transect and perpendicular to the creek (10−3

to 10−2), suggesting that the wetland system may beadequately represented by a two-dimensional verticaltransect of the field site.

4.3. Rainfall

In the 6 months prior to the study period, the studysite at Tomago South experienced typical climaticconditions. Pan evaporation and rainfall measure-ments from Williamtown AMO were similar tolong-term averages. During the period modelled inJanuary and February 1997, Tomago South experi-enced two significant storm events of 64 and111 mm in 72 h. February rainfall recorded atWilliamtown was 34 mm higher than the average of152 mm.

Significant rainfall events induced a similarly largeresponse in the groundwater, but both the peak anddecay of the rainfall response lag behind the rainfall.This is due to the low conductivity of surfacesediments, which limit the rate of infiltration and

drainage. Comparison of Periods A and B (Fig. 3)indicate that during the neap tide period the rainfallresponse is slower. This can be explained by the rela-tively low initial soil moisture content expected in theabsence of significant tidal groundwater fluxes.

The groundwater response to rainfall was observedto be significant across the whole site with similar lagtime and drainage rates in both tidally inundated andnon-tidal zones. Very small rainfall events of about4–5 mm are sufficient to produce a groundwaterresponse. The maximum hydraulic head observedfor two separate major rainfall events was similarindicating a possible maximum soil storage capacityat the point where the rate of sub-surface drainage tothe creek equals the infiltration rate of the soil. Initialrainfall in the order of 10 to 20 mm (over the first 2 to4 h) is required to produce this response.

4.4. Evapotranspiration

For the study period, the Penman potential evapora-tion calculated using on-site weather station data froma fixed reference location averaged 3.8 mm/day with amaximum of 7.4 mm/day. During the same period,pan evaporation rates of up to 11.2 mm/day andwith an average of 5.6 mm/day were observed atWilliamtown AMO. Average monthly rainfall andpan evaporation at Williamtown AMO for the 49and 22 years of record are given in Fig. 4.

Actual evapotranspiration (AET) was determinedat three sites, with different dominant vegetation

Fig. 4. Average monthly rainfall and pan evaporation, Williamtown AMO.

40 C.E. Hughes et al. / Journal of Hydrology 211 (1998) 34–49

types during the spring of 1997, using eddy correla-tion measurements. Actual evapotranspiration wascompared with Penman potential evaporation andPenman-Monteith potential evapotranspiration(PET) at the fixed reference location. AET/Penmanopen water evaporation ratios were averaged fromfour daily measurements for each salt marsh commu-nity and three for the kikuyu pasture, with values of0.6 (60.1), 0.72 (60.03) and 0.85 (60.13) for kikuyu,Sporobolus virginicusand Sarcocornia quinquefloradominated sites respectively (standard deviations ofmeasurements shown in brackets). CorrespondingAET/PET ratios were 0.87 (60.04), 1.01 (60.13)and 1.05 (60.08). The higher evapotranspirationcorresponds to vegetation found in the intertidalzone, where soil-water saturation values are higher.

The limited data available suggest that on an aver-age yearly basis, rainfall at Williamtown and evapo-transpiration at Tomago South are similar, withrainfall exceeding evapotranspiration in the wintermonths, and evapotranspiration exceeding rainfall inthe summer months. Therefore, after allowing forrainfall loss due to surface runoff, it is likely thatevapotranspiration exceeds rainfall infiltration mostof the time. The extra water lost to evapotranspirationis supplied by tidal flooding and soil water movementfrom the tidal creeks to the soil surface.

Evapotranspiration is an important factor in the soilmoisture and salt balance of tidal wetlands and insome studies has been found to be a dominant factorin groundwater fluctuations (Harvey and Nuttle, 1995;

Dacey and Howes, 1984; Hemond and Fifield, 1982).Inspection of the well data from the study site to date,however, provides no indication of underlying watertable sensitivity to evaporative forcing. Given thattidally driven head fluctuations in the order of 100to 250 mm occur daily in the groundwater table, itseems reasonable that an evaporative response in theorder of 5 to 10 mm/day would be undetectable inhead measurements.

4.5. Analytical solution for groundwater response totidal forcing

Several analytical methods have been developed toanalyse the response of an aquifer to periodic forcing.Such solutions may be used to determine the proper-ties of tidally influenced aquifers (e.g. Millham andHowes, 1995). Townley (1995) recast Williams’(1982) solution for periodic flow in a homogeneousone-dimensional unconfined aquifer bounded on oneside by sinusoidally varying tides, in terms of com-plex variables as follows:

h(x, t) =Hs +ReHp cosh(bx=l)(cos(qt) + i sin(qt)

cosh(b)

� �(1)

whereh(x,t) is the head at timet and distancex fromthe tidal fluctuation,Hs is the steady component ofhead, Hp is the periodic component of head orthe tidal amplitude,b2 =2pi(L2S=TP) whereL is thelength of the aquifer,S is the specific yield,T is the

Fig. 5. Analytical solution for groundwater response withHs = 20,S= 0.35,T = 1000,P = 0.5 to (A) 0.4 m tidal forcing adjacent to the studysite (Hp = 0.2,L = 50); (B) 1.4 m tidal forcing downstream of the creek obstruction 150 m from the study site (Hp = 0.7,L = 150); and (C) 2 mtidal forcing at the Hunter River 500 m from the study site (Hp = 1.0, L = 500).

41C.E. Hughes et al. / Journal of Hydrology 211 (1998) 34–49

aquifer transmissivity,P is the tidal period and theangular frequency of fluctuations,q =2p=P.

Townley’s analytical solution can be used to esti-mate the response of groundwater at the Tomago siteto various scales of tidal influence. Specifically it canbe used to determine whether groundwater behaviourat the study site is due to the 0.4 m tidal range in theadjacent creek, the 1.4 m tides in the creek 100 mdownstream or the 2 m tides in the Hunter River500 m away.

Fig. 5 shows Townley’s analytical solution evalu-ated for these three scales of tidal forcing in an idea-lised homogeneous, one-dimensional aquifer, usinghydraulic parameters estimated for our site. Theresults show clearly that the larger tidal forcing inthe river is not transmitted over the long distance tothe field site, while the smaller tidal fluctuationsobserved in the adjacent creek are. Therefore, thelocal groundwater movement is dominated by tidalforcing directly adjacent to the site. These resultsare consistent with observed head data from thestudy site, suggesting that the Townley solution

broadly represents the behaviour of the groundwaterin the sand unit at the study site.

The scale of the study site is small in comparisonwith the range of scales encountered at the TomagoSouth wetlands, where distance between tidal water-ways varies from approximately 40 m as seen at thestudy site, up to 300 m. Tides vary from small attenu-ated fluctuations such as those encountered at thestudy site to the full 2 m range of the Hunter River.Most of the wetland is close enough to a tidal creek orchannel to experience tidal forcing of groundwater.

5. Two-dimensional finite element analysis

Conditions at the study site are far more complexthan the simple assumptions required for Townley’ssolution. The Townley model predicts a diminishingtidal response with distance from the tidal forcing,even at the small scales of the field site. Such a fieldresponse has been observed by Howes and Goehringer(1994) and Nuttle and Harvey (1988). Several factors

Fig. 6. SEEP/W and Townley solutions (Hs = 6, Hp = 0.2,P = 0.5) for (A) sand (S= 0.35,T = 1000,K s = 26 m/d,Qs = 0.45) and (B) mud (S=0.1,T = 0.35,Ks = 0.05 m/d,Q s = 0.7). Grey shading represents the amplitude of the Townley solution at any distance and lines represent theSEEP/W solution at specific times.

42 C.E. Hughes et al. / Journal of Hydrology 211 (1998) 34–49

distinguish our study site from the idealised situationdescribed by Townley and in some instances fromthe other sites. These include the presence of over-bank flow, observed three-dimensional ground waterflows, aquifer heterogeneity, and non-sinusoidal tidalbehaviour. SEEP/W, a commercial two-dimensionalfinite element model was used to model the site(GEO-SLOPE, 1994). A two-dimensional modelwas used rather than a three-dimensional modelbecause the observed fluxes in the third dimensionare smaller than those in the modelled dimensions.

SEEP/W is based on a mass balance statement andDarcy’s Law applied to both saturated and unsatu-rated flow (GEO-SLOPE, 1994). The governingdifferential equation used by SEEP/W is:

]

]xKx

]H]x

� �+

]

]zKz

]H]z

� �Q=

]v

]t(2)

whereH is total head,Kx andKz are hydraulic con-ductivity in the x and z directions respectively,Q isthe applied boundary flux,v is the volumetric watercontent andt is time.

No examples of SEEP/W being used to model atidally forced aquifer were found, so confidence inthe ability of SEEP/W to model a sinusoidally varyingtidal head boundary was gained by successful replica-tion of Townley’s analytical solutions for a 0.4 m tidalrange in sand and mud. This comparison is validdespite the linearisation assumption in the solutionof Townley because the head changes are small com-pared to the aquifer thickness. Consequently, the line-arised solution of Townley is close to the solution ofthe fully non-linear phreatic aquifer equations. A

comparison of the results from the SEEP/W modeland the Townley analysis is given in Fig. 6.

Not only is the amplitude of groundwater responseto tidal forcing modelled accurately by SEEP/W, asseen in Fig. 6, the shape of the tidal wave at anyspecific stage of the tidal cycle is also comparable(not shown in Fig. 6).

To model saturated and unsaturated flow at thestudy site, the finite element mesh shown in Fig. 7was constructed. The full mesh extends down to anelevation of−20 m. Element layers and material typeswere based on the site stratigraphy shown in Fig. 2.

5.1. Soil water retention and conductivity curves

Application of SEEP/W to model unsaturated flowrequires detailed functions of hydraulic conductivityand volumetric moisture content versus pore pressureto define the behaviour of each material. Saturatedconductivity and saturated moisture content (usingporosity) were based on the values in Table 1. Satu-rated moisture content was fixed at 0.45, 0.7, 0.6 and0.7 for the silty sand, estuarine mud, clayey sand andsurface layer, respectively. Initial saturated hydraulicconductivity values were 26, 0.06, 1 and 0.04 m/dayfor the silty sand, estuarine mud, clayey sand and sur-face layer, respectively; these values were modified to20, 0.15, 2 and 0.043 m/day during calibration. Theforms of the functions were not measured for thematerials at the site. Functions were selected foreach material type from Appendix A.5 of the SEEP/W manual (GEO-SLOPE, 1994) then modified duringcalibration to those shown in Fig. 8. By calibrating

Fig. 7. SEEP/W mesh for the top 5 m of the study site (grid goes to 20 m).

43C.E. Hughes et al. / Journal of Hydrology 211 (1998) 34–49

these functions and the 2D model to measured hydrau-lic head for a variety of different scenarios, an opera-tional model was created that would apply to anyplausible values of recharge, tidal flux and evapo-transpiration within the ranges of calibration.

During the calibration process it became apparentthat the water table response to tidal forcing is deter-mined largely by the hydraulic conductivity of thesand and mud units. The response to rainfall wasfound to be determined primarily by near-surfacecharacteristics. Calibration of the model for both rain-fall and tidal forcing demanded conflicting materialcharacteristics for the mud layer. To produce theobserved rapid response to tidal inundation, a highair entry pressure of−15 kPa was set for the estuarinemud in Fig. 8, so the mud remains saturated and theconductivity does not decrease rapidly when the watertable drops. In contrast, to produce the slow watertable decline observed after a rainfall event, a surfacelayer was constructed with a low air entry pressure toensure a rapid decrease of conductivity with satura-tion. The material type of the top layer of elements inFig. 7 is given the hydraulic functions of the surfacelayer.

At the study site, a significant component of bothsurface infiltration and subsurface transport isbelieved to occur in macropores rather than throughthe sediment matrix. Use of Darcy’s Law and relativehydraulic conductivity functions is not generallyconsidered appropriate in modelling macropore flow

(J. Long, personal communication). However, consid-erable research has been completed on measuring soilwater retention and hydraulic conductivity curves insoils where two or multi-domain macropore flow isobserved (Jarvis and Messing, 1995; Timlin et al.,1994; Harvey, 1993). Large macropores, such as thecrab holes found at the study site, have been observedto drain at matric potentials of less than 0.1 kPa(Beven and Germann, 1982). When the soil is nearlysaturated, macropores dominate the flow and the con-ductivity is high, however, when the pressure dropsslightly, the macropores de-saturate and the conduc-tivity rapidly drops to that of the soil matrix.

In order to better reflect the behaviour of macro-pores in the near surface region of the estuarine mud, asurface layer was created as described above. Theupper soil layer at the site is characterised by a highconcentration of crab holes in the intertidal zone andby loose highly organic topsoils above the intertidalzone. This contrasts with the relatively undisturbedestuarine mud matrix beneath. The air entry pressureis low (−1 to −3 kPa) and the slope of the moisturecontent function of this surface layer is very steep,which is consistent with curves for macropore soilsreported in the literature.

The soils were modelled to be anisotropic, withanisotropy ratios (Kz:Kx) equal to 0.25, 0.5, 1, 1.5 inthe silty sand, estuarine mud, clayey sand and surfacelayer, respectively, whereKx is taken from the con-ductivity function. The higher anisotropy value

Fig. 8. Volumetric water content and hydraulic conductivity versus pore pressure for the four material types used in the SEEP/W model.

44 C.E. Hughes et al. / Journal of Hydrology 211 (1998) 34–49

employed in the surface layer reflects the influence ofthe macropores, while the lower values in the siltysand and estuarine mud reflect horizontal stratificationin the deposit. The model was quite sensitive tovariations in anisotropy and saturated hydraulic con-ductivity of all layers except the clayey sand unit.

In addition, retention and conductivity functionsvary with changes in the pore water salt concentration.Bresler (1981) showed that solute composition(Na:Ca ratio) and concentration affect the unsaturatedhydraulic conductivity and water retention of clayeysoils. Hydraulic conductivity increases and the slopeof the water retention function becomes steeper withincreasing solute concentration in soils with a highNa:Ca ratio. The implications of this on the near sur-face water and solute balance at the study site may besignificant, making it a topic for further investigationoutside the scope of this paper.

5.2. Boundary functions

Three types of boundary conditions were applied inthe model at various nodes and times (Fig. 9).

Tidal head (m) was applied at all surface nodes oflower elevation than the maximum tidal height as avarying head boundary function. When the headspecified is less than the elevation of the node theboundary condition is set toQ = 0.

Rainfall was applied as a flux (m/day) to the surfacelayer of nodes above the zone influenced by tidalinundation. Once the surface is saturated, thisboundary condition approximates infiltration bysetting the head boundary condition so that the headequals node elevation. At such times, the applied rain-fall flux exceeds the saturated hydraulic conductivity.This is equivalent to infiltration excess or surfaceponding.

Potential evaporation was applied as a nodal flux(m3/day) equally to the top three layers of nodes in themesh (approximately 0 mm, 130 mm and 260 mmbelow the surface) in order to approximate the actualvertical distribution of moisture extraction throughevapotranspiration. During periods of tidal inundationand rainfall, the surface evaporation is taken from thesurface ponding store rather than the soil matrix so asnot to affect soil moisture.

Fig. 9. Boundary condition input data.

45C.E. Hughes et al. / Journal of Hydrology 211 (1998) 34–49

The SEEP/W software does not allow two bound-ary conditions to be applied to the same node, nordoes it allow for automatic exchange of boundaryconditions. However, rainfall, tidal inundation andevapotranspiration occur simultaneously at surfacenodes. Therefore, when conflicting boundary require-ments occur, the dominant process at that time isgiven priority, in the order tidal inundation, rainfall,evaporation. The tidal boundary condition wasapplied to the highest node inundated during eachperiod for the whole of the period.

Due to limitations in the number of time stepsallowed in SEEP/W, the model was run in ninesections. The boundary conditions applied at eachnode were changed at the beginning of each sectionand at the commencement or end of a rainfall period(more frequent altering of boundary conditions doesnot significantly alter results). The final time step ofthe previous section was used as the initial conditionfor the following section. The length of time step wasvaried from 15 min to 2 h. The shorter intervals wereused for periods where more rapid change wasexpected, e.g. the onset of rainfall and during therapid rise and fall of the tide.

SEEP/W is not capable of modelling the evapo-transpiration process directly; however, a modifierfunction may be applied to scale potential evaporationdata according to the pore pressure at each node. Anevaporation modifier based broadly on the stressindex principle (i.e. modifier= SI = (v − vwilting point)/(vfield capacity− vwilting point)) was used in the model. Thepotential evaporation for each time step is multipliedby the modifier value calculated from the matrixpressure at that node to calculate the evaporativeboundary flux in the model. The modifier decreasesthe magnitude of the evaporation boundary flux as thematrix pressure, and correspondingly the soil moistureavailable for evaporation, decreases at any evapora-tion node. In the absence of actual wilting point andfield capacity data for the study site, the modifier wasdeveloped using estimates within normal ranges thenadjusted during calibration to remove the sharpdecreases in water table during low tide.

5.3. Model results

The SEEP/W model was calibrated for the two datasets in Fig. 3: first, a 2-day rainfall event during neap

Fig. 10. Comparison of model results and observations (dashes represent observations, solid line is SEEP/W output).

46 C.E. Hughes et al. / Journal of Hydrology 211 (1998) 34–49

tides; second, a spring tide period with significantevapotranspiration followed by a rainfall event duringthe end of the spring tide period. Model parameters foreach soil type were determined by the relative balanceof fit between wet and dry periods. The model wasthen applied to a continuous period of 26 days, asshown in Fig. 10.

The average deviation of model heads fromobserved heads was in the order of 5 mm for A3,15 mm for A4 and 2 mm for A6. Maximum deviationsof up to 61 mm occurred when extreme drops in headwere predicted by the model at the commencement ofrainfall, e.g. on 25/1/97 and 17/2/97. A4 shows a sig-nificant ‘‘under prediction’’ for high rainfall periodsfrom 30/1/97 to 2/2/97 and 12/2/97 to 15/2/97 whichis believed to be an artefact of the field data due todisturbance during bore hole installation. SEEP/Wunderpredicts the peak tidal response during the dryperiod from 6/2/97 to 10/2/97.

Model results are highly sensitive to variations insaturated hydraulic conductivity, anisotropy ratio,retention and conductivity function slope in the sur-face layer and estuarine mud units. Predictions of tidalforcing are much less sensitive to these variations thanare predictions of rainfall. Reviewing and changingthe surface boundary condition between tidal andrainfall and evaporation more frequently (e.g.between each time step in the model run) was notfound to significantly alter head response.

5.4. Relative magnitude of porewater flows

Boundary fluxes for ‘‘representative’’ tidal andrainfall periods were compiled from SEEP/W output(Fig. 11). The fluxes presented in the Figure are thetotal flux in a given direction over 1 day, summedover the length of the boundary to which they areapplied. Initial and final times with similar observedhydraulic heads were chosen for each figure tofacilitate direct comparison.

Peak modelled saturated specific discharge is indi-cative of the maximum zone of groundwater move-ment. For the tidal period (A), peak specific dischargewas in the order of 0.38 m/day below the creek bedand 0 to 0.1 m/day in the centre of the site. Specificdischarge for the rainfall period (B) peaked at around0.43 m/day below the creek bed, and was generally inthe range 0 to 0.1 m/day in the centre of the site,decreasing to around 0.05 in the unsaturated zone.Flow paths through the sand unit to and from thecreek are dominant during both periods. The tidalperiod is characterised by a strong cyclic responseto tidal fluctuations in both the sand and mud units,with an equally significant evaporation loss. Surfacetidal inundation provides a smaller but substantialcontribution to the water balance. The rainfall periodis completely dominated by subsurface drainage ofinfiltrated rainfall, which is only slowed, not reversed,during high tides. The maximum rainfall fluxes occur

Fig. 11. Hydrologic budget and generalised flow paths for (A) tidal and evaporative forcing 7:00 9/2/97 to 10:00 10/2/97; and (B) rainfall, tidaland evaporative forcing 21:30 11/2/97 to 16:30 12/2/97. Flux units are m3/day per metre width of creek bank.

47C.E. Hughes et al. / Journal of Hydrology 211 (1998) 34–49

during a neap tide period when there is a minimuminitial storage and maximum hydraulic gradient forrecharge.

The fluxes above can be examined and used toexplain the high salt concentrations in the piezo-meters, which were observed to be at concentrationsup to 1.5 times that of seawater. Water salinity inestuarine surface water at the site has been observedto range from 6 to 35 ppt, so concentration of saltsthrough evapotranspiration must be occurring. Indeedsalt encrustation is often observed at the site. Theseconcentrations can be explained by the climate data,which suggest that annual average evapotranspirationexceeds rainfall infiltration. Furthermore, the zone oftidal infiltration into the soil matrix is small, as seenfrom the above fluxes (a peak flux of 0.38 m/day wasobserved to occur at the creek edge), so that mixing ofsoil water with estuarine water is minimal. Ultimatelyit is the balance between the concentrating of salts dueto evapotranspiration and the diluting effect due torainfall and tidal mixing that gives the final soil saltconcentration.

6. Conclusion

This study has found that at the Tomago South site,in contrast to previous studies at other sites (Howesand Goehringer, 1994; Nuttle and Harvey, 1988), tidalforcing is a dominant mechanism of porewater move-ment in the saturated and intertidal zones, with thelargest fluxes due to subsurface drainage to thecreek. Major rainfall events are observed to have ahighly significant but short-term effect on watertable levels due to rapid drainage through the under-lying sand aquifer. Evapotranspiration was estimatedto be very important during dry periods and is antici-pated to be highly significant in determining the saltbalance in the soil matrix.

The range of influence of tidal forcing extends farbeyond the zone of surface inundation and it was notpossible to observe a significant attenuation ingroundwater response over the small scale studied.Analytical methods have enabled us to estimate thedecay of the tidal response for a variety of scenariosencountered at Tomago. From these results, we canconclude that tidal fluctuations will influence thewater table throughout a large part of the Tomago

South wetlands, due to the density of the tidaldrainage network.

Finite element modelling of the site has providedinsight into the path of water movement and thepartitioning of rainfall and tidal effects within thesediments. Future work will couple a two-dimensionalsolute transport model with the flow model.

The anticipated application of this research at theTomago site is in predicting groundwater and surfacewater balance response to alterations in the tidal andsurface water regimes arising from proposed wetlandrestoration works.

Acknowledgements

The first author on this paper was supported by anAPA(I) grant funded by the Australian ResearchCouncil and Tomago Aluminium Pty Ltd. The projectwas also supported by the Kooragang WetlandRehabilitation Project. We thank Brendan Harrodfor allowing us to use his results from laboratorysoil testing and field measurement of infiltration andMarika Vertzonis and Jetse Kalma for their eddy cor-relation measurements. We would also like toacknowledge the useful comments provided by thereviewers.

References

Beven, K., Germann, P., 1982. Macropores and water flow in soils,Water Resources Research, 18, 1311–1325.

Bresler, E., 1981. Transport of salts in soils and subsoils. In:Holmes, J.W., Talsma, T. (Eds.), Land and Stream Salinity.Elsevier, Amsterdam, pp. 35–62.

Chalmers, A.G., 1982. Soil dynamics and the productivity ofspartina alterniflora. In: Kennedy, V.S. (Ed.), Estuarine Com-parisons. Academic Press, New York, pp. 231–242.

Clarke, L.D., Hannon, N.J., 1967. The mangrove swamp and saltmarsh communities of the Sydney district: I. Vegetation, soilsand climate, Journal of Ecology, 55, 753–771.

Dacey, J.W.H., Howes, B.L., 1984. Water uptake by roots controlswater table movement and sediment oxidation in shortSpartinamarsh. Science 4 May, 487–489.

Fityus, S.G., Welbourne, J.C., 1996. Trends in shrink–swell testresults in the Newcastle region. In: Jaksa, M.B, Kaggwa, W.S.,Cameron, D.A. (Eds.), Proceedings of the 7th Australia NewZealand Conference on Geomechanics, Adelaide SA, 1–5 July1996. IEAust, Canberra, pp. 394–399.

GEO-SLOPE International Ltd, 1994. SEEP/W Users Guide. GEO-SLOPE, Calgary.

48 C.E. Hughes et al. / Journal of Hydrology 211 (1998) 34–49

Harvey, J.W., 1993. Measurement of variation in soil solute tracerconcentration across a range of effective pore sizes, WaterResources Research, 29, 1831–1837.

Harvey, J.W., Chambers, R.M., Hoelscher, J.R., 1995. Preferentialflow and segregation of porewater solutes in wetland sediment,Estuaries, 18, 568–578.

Harvey, J.W., Germann, P.F., Odum, W.E., 1987. Geomorpholo-gical control of subsurface hydrology in the creekbank zone oftidal marshes, Estuarine, Coastal and Shelf Science, 25, 677–691.

Harvey, J.W., Nuttle, W.K., 1995. Fluxes of water and solute in acoastal wetland sediment. 2. Effect of macropores on soluteexchange with surface water, Journal of Hydrology, 164,109–125.

Hemond, H.F., Fifield, J.L., 1982. Subsurface flow in salt marshpeat: a model and field study, Limnology and Oceanography,27, 126–136.

Howes, B.L., Dacey, J.W.H., Goehringer, D.D., 1986. Factors con-trolling the growth form ofSpartina alterniflora: feedbacksbetween above-ground production, sediment oxidation, nitro-gen and salinity, Journal of Ecology, 74, 881–898.

Howes, B.L., Goehringer, D.D., 1994. Porewater drainage anddissolved organic carbon and nutrient losses through the inter-tidal creekbanks of a New England salt marsh, Marine EcologyProgress Series, 114, 289–301.

Jarvis, N.J., Messing, I., 1995. Near-saturated hydraulic conduc-tivity in soils of contrasting texture measured by tension infilt-rometers, Soil Science Society of America Journal, 59, 27–34.

Maidment, D.R. (Ed.), 1993. Handbook of Hydrology. McGraw-Hill, New York.

Millham, N.P., Howes, B.L., 1995. A comparison of methods todetermine K in a shallow coastal aquifer, Groundwater, 33, 49–57.

Mitsch, W.J., Gosselink, J.G., 1993. Wetlands, 2nd ed. VanNostrand Reinhold, New York.

Nuttle, W.K., Harvey, J.W., 1988. Geomorphological controls onsubsurface transport in two salt marshes. In: Kustler, J.A.,Brooks, G. (Eds.), Proceedings of the National WetlandSymposium: Wetland Hydrology. Association of State WetlandManagers, New York, pp. 253–261.

Nuttle, W.K., Harvey, J.W., 1995. Fluxes of water and solute in acoastal wetland sediment. 1. The contribution of regionalgroundwater discharge, Journal of Hydrology, 164, 89–107.

Public Works Department (PWD), 1994. Lower Hunter RiverFlood Study: Green Rocks to Newcastle. PWD Report No.91077, PWD, Sydney.

Ridd, P., 1996. Flow through animal burrows in mangrove creeks,Estuarine, Coastal and Shelf Science, 43, 617–625.

Thom, B.G., Shepherd, M., Ly, C.K., Roy, P.S., Bowman, G.M.,Hesp, P.A., 1992. Coastal geomorphology and quaternarygeology of the Port Stephens-Myall Lakes area. Departmentof Biogeography and Geomorphology, ANU, Canberra.

Timlin, D.J., Ahuja, L.R., Ankeny, M.D., 1994. Comparison of threefield methods to characterize apparent macropore conductivity,Soil Science Society of America Journal, 58, 278–284.

Townley, L.R., 1995. The response of aquifers to periodic forcing,Advances in Water Resources, 18, 125–146.

Williams, J.A., 1982. Diffusivity, storativity and the Dupuitassumptions for periodic flow in a vertical Hele–Shawmodel, Water Resources Research, 18, 925–930.

Wolanski, E., Mazda, Y., Ridd, R. 1992. Mangrove hydro-dynamics. In: Robertson, A.I., Alongi, D.M., (Eds.), TropicalMangrove Ecosystems. AGU, Washington DC, pp. 43–62.

Woolley, D., Mount, T., Gill, J., 1995. Technical Review of theTomago/Tomaree/Stockton Groundwater. NSW Department ofWater Resources, Sydney.

Yelverton, G.F., Hackney, C.T., 1986. Flux of dissolved organiccarbon and pore water through the substrate of aSpartina alter-niflora marsh in North Carolina, Estuarine, Coastal and ShelfScience, 22, 255–267.

49C.E. Hughes et al. / Journal of Hydrology 211 (1998) 34–49


Recommended