Salt marsh ecohydrological zonation due to heterogeneous ...accounting for intertidal inﬁltration, reaction, transport, discharge, and mixing of salt marsh groundwater and tidal

Salt marsh ecohydrological zonation due to heterogeneousvegetation–groundwater–surface water interactions

Kevan B. Moffett,1 Steven M. Gorelick,1 Robert G. McLaren,2 and Edward A. Sudicky2

Received 4 May 2011; revised 9 November 2011; accepted 9 January 2012; published 15 February 2012.

[1] Vegetation zonation and tidal hydrology are basic attributes of intertidal salt marshes,but specific links among vegetation zonation, plant water use, and spatiotemporallydynamic hydrology have eluded thorough characterization. We developed a quantitativemodel of an intensively studied salt marsh field site, integrating coupled 2-D surface waterand 3-D groundwater flow and zonal plant water use. Comparison of model scenarios withand without heterogeneity in (1) evapotranspiration rates and rooting depths, according tomapped vegetation zonation, and (2) sediment hydraulic properties from inferred geologicalheterogeneity revealed the coupled importance of both sources of ecohydrologicalvariability at the site. Complex spatial variations in root zone pressure heads, saturations,and vertical groundwater velocities emerged in the model but only when both sources ofecohydrological variability were represented together and with tidal dynamics. Theseregions of distinctive root zone hydraulic conditions, caused by the intersection ofvegetation and sediment spatial patterns, were termed ‘‘ecohydrological zones’’ (EHZ).Five EHZ emerged from different combinations of sediment hydraulic properties andevapotranspiration rates, and two EHZ emerged from local topography. Simulated pressureheads and groundwater dynamics among the EHZ were validated with field data. The modeland data showed that hydraulic differences between EHZ were masked shortly after aflooding tide but again became prominent during prolonged marsh exposure. We suggestthat ecohydrological zones, which reflect the combined influences of topographic, sediment,and vegetation heterogeneity and do not emphasize one influence over the others, are thefundamental spatial habitat units comprising the salt marsh ecosystem.

Citation: Moffett, K. B., S. M. Gorelick, R. G. McLaren, and E. A. Sudicky (2012), Salt marsh ecohydrological zonation due to

heterogeneous vegetation–groundwater–surface water interactions, Water Resour. Res., 48, W02516, doi:10.1029/2011WR010874.

1. Introduction[2] Coastal salt marshes serve critical functions valuable

to both natural and human systems. Marshes’ complex bio-geochemical role in the coastal zone has inspired particu-larly active research in recent years, since the nutrientoutwelling hypothesis by Odum [1980] (as discussedby Childers et al. [2000]). However, it has been difficultto generalize dynamic land-ocean interactions, whethermarshes are a source or sink for coastal zone constituents,because of the lack of a comprehensive hydrologic modelaccounting for intertidal infiltration, reaction, transport,discharge, and mixing of salt marsh groundwater and tidalsurface water. A better understanding of salt marsh ecohy-drological system dynamics is required before connectionswith other fields such as coastal biogeochemistry can bestrengthened [Rodriguez-Iturbe et al., 2007].

[3] Wetland surface water and groundwater dynamicsare inextricably linked, although often studied separately[Winter et al., 1998]. Intertidal marsh surface water hydrol-ogy is characterized by tidal flood and ebb in the channelnetwork [Fagherazzi et al., 1999; Marani et al., 2002,2003]. During especially high tides, the channels fill beyondbankfull capacity and tidal waters flood across the marshplain, influenced by vegetation roughness [Temmermanet al., 2005; van Proosdij et al., 2006].

[4] Salt marsh groundwater flow is represented in the lit-erature by four somewhat conflicting conceptual models:(1) Flow in interior marsh sediments is entirely verticalbecause of alternating phases of evapotranspiration andinfiltration governed by the tides [Hemond and Fifield,1982; Dacey and Howes, 1984]. (2) Vertical flow in themarsh interior feeds deep, slow flow paths beneath themarsh that contribute to submarine groundwater dischargein the coastal zone [Wilson and Gardner, 2006]. (3) Flow ishorizontal and restricted to the few meters near the channelbanks affected by rapid drainage; there is no lateral flow inthe marsh interior because of low sediment permeabilityand zero groundwater head gradient [Harvey et al., 1987;Nuttle, 1988; Montalto et al., 2007]. (4) Plant root wateruptake near channel banks controls local water table posi-tion and unsaturated flow [Howes and Goehringer, 1994;

1Department of Environmental Earth System Science, Stanford Univer-sity, Stanford, California, USA.

2Department of Earth and Environmental Sciences, University ofWaterloo, Waterloo, Ontario, Canada.

Copyright 2012 by the American Geophysical Union0043-1397/12/2011WR010874

W02516 1 of 22

WATER RESOURCES RESEARCH, VOL. 48, W02516, doi:10.1029/2011WR010874, 2012

http://dx.doi.org/10.1029/2011WR010874

Ursino et al., 2004; Wilson and Gardner, 2005; Li et al.,2005; Marani et al., 2006; Tosatto et al., 2009]. Each ofthese conceptual models clearly addresses marsh ground-water hydrology at a different scale, yet each discountsthe possibility of lateral groundwater flow in the marsh in-terior and leaves open the question of the effects of differ-ent plant species on salt marsh groundwater flow [Silvestriet al., 2005].

[5] Expanding on the last of the above conceptualmodels, there are three notable features of salt marsh plant-water interactions: (1) Plant water use comprises a signifi-cant portion of the soil water balance and partially controlswater table position [Dacey and Howes, 1984]. (2) Plantspecies take up water from the root zone at different ratesbecause of differences in energy balance, photosyntheticmetabolism, and water use efficiency [Teal and Kanwisher,1970; Mahall and Park, 1976a, 1976b; Antlfinger andDunn, 1979; Giurgevich and Dunn, 1982]. (3) Intertidalsalt marshes exhibit pronounced vegetation zonation, withplant assemblages configured into visually obvious spatialpatterns. This vegetation zonation is partially related to pat-terns of soil water availability or excess [Chapman, 1938a,1938b; Mahall and Park, 1976b; Cooper, 1982; Penningsand Callaway, 1992; Silvestri et al., 2005; Varty andZedler, 2008], among other causes including interspecificinteractions and variations in nutrients, soils, salinity, tidalexposure, and disturbance [Bertness et al., 1992; Penningsand Callaway, 1992; Emery et al., 2001; Pennings et al.,2005; Forbes and Dunton, 2006]. Interestingly, these threefeatures of salt marsh plant-water relations have not yetbeen combined nor integrated with physics-based modelsof intertidal hydrology.

[6] This study integrated salt marsh vegetation zonation,interspecific differences in plant water use, three-dimen-sional variably saturated groundwater hydrology, sedimentheterogeneity, and tidal flooding into a numerical model ofthe salt marsh ecohydrological system. Our goal was toquantify the relative and combined effects of these multiplesystem components on marsh hydrology by comparing acomplex model including all these components to simplermodels that included spatial variations in only sedimentproperties or evapotranspiration, or neither. We also con-trasted flooding and nonflooding tidal regimes.

2. Review of Literature on Intertidal Salt MarshGroundwater Modeling

[7] Models of intertidal salt marsh groundwater flowhave not reconciled the contrasting scales of tidal andgroundwater flow and plant-water interactions thus far,since they have not accounted for the natural three-dimensionality of the intertidal system nor for spatiallyvariable plant-water interactions. Intertidal salt marshgroundwater numerical modeling studies are summarizedin Table 1. Zero-dimensional water balance models cannotrepresent spatial variability and multiscale hydrologicalprocesses. One-dimensional models have focused either onvertical infiltration (1-D-Z models) or on the extent of theperpendicular propagation of the harmonic tidal signal awayfrom major tidal creeks (1-D-X models). Most salt marshgroundwater modeling has examined two-dimensionaldomains that are vertical slices perpendicular to major

tidal channels (2-D-XZ models). Empirical investigationsof groundwater flow and transport have also focused onthis 2-D-XZ geometry [Chapman, 1938a; Agosta, 1985;Harvey and Odum, 1990; Howes and Goehringer, 1994;Hayden et al., 1995; Osgood and Zieman, 1998; Montaltoet al., 2006]. Few models or experiments have examinedlateral (2-D-XY) or three-dimensional (3-D) salt marshgroundwater hydrology. Notable exceptions are the ground-water tracer experiments by Tobias et al. [2001] andJordan and Correll [1985] in 2-D-XY and 3-D, respec-tively, and the 3-D archetypal marsh-and-channel model byXin et al. [2011].

[8] Many marsh groundwater models have included un-saturated flow; some have included sediment layering; fewhave included sediment heterogeneity or macroporosity(see Table 1). Evapotranspiration has been omitted frommany models or represented as a constant, spatially uniformsurface flux. This flux was based on field or laboratory datain only a few cases (see Table 1). Very few models havedistributed evapotranspiration demand through a finite root-ing depth [Hemond and Fifield, 1982; Hughes et al., 1998;Marani et al., 2006] and none have considered spatiallyvariable evapotranspiration, despite the characteristic zona-tion and water use variations of salt marsh plant species.

[9] No salt marsh groundwater model has yet integratedinterspecific differences in plant water use, salt marsh vege-tation zonation, intertidal plant-water interactions, 3-D vari-ably saturated groundwater hydrology, and tidal floodinginto a more complete ecohydrological framework for theintertidal salt marsh system. In this study, we develop sucha framework via numerical simulations of an intensivelystudied field site.

3. Numerical Model Governing Equations[10] A numerical model of fully coupled surface water

and groundwater dynamics was developed using Hydro-GeoSphere [Therrien and Sudicky, 1996; Therrien et al.,2008]. Variably saturated groundwater flow was simulatedas a 3-D boundary value problem. At the same time, 2-D,depth-averaged surface water flow was simulated in a sur-face domain spatially coincident with the top (ground) sur-face of the groundwater flow domain. Exchange betweenthe two domains was solved for, driven by the differencebetween surface water and groundwater heads at the sedi-ment surface. Capturing rapid tidal dynamics required veryfine, adaptive model time steps. The computational cost offine time steps was justified by the ability to precisely solvefor the transient positions of surface water bodies andgroundwater seepage faces in three dimensions, a uniquefeature of this salt marsh modeling study.

[11] Variably saturated groundwater flow was repre-sented by Richards’ equation:

�rðqÞ þ Go 6 Q ¼ SwSs@

@tþ � @Sw

@t: (1)

Variables are the 3-D gradient (!) of groundwater flux (q),exchange flux with the surface (Go), and boundary fluxes(Q). Fluid mass was conserved by temporal (t) changes insediment saturation (Sw) and fluid pressure ( ), accordingto the sediment specific storage (Ss) and porosity (�).

W02516 MOFFETT ET AL.: SALT MARSH ECOHYDROLOGICAL ZONATION MODELING W02516

2 of 22

Tab

le1.

Sum

mar

yof

Inte

rtid

alS

alt

Mar

shG

roun

dwat

erF

low

Num

eric

alM

odel

s

Sou

rce

Mod

elD

imen

sion

aT

ime

Dom

ain

Sat

urat

ion

Reg

ime

Sed

imen

ts,

Mac

ropo

rosi

tyP

aram

eter

izat

ion

Eva

potr

ansp

irat

ion

Nut

tle

and

Har

vey

[199

5]w

ater

bala

nce

0-D

tran

sien

tsa

tura

ted

hom

ogen

ous;

mac

ropo

rosi

tyse

nsit

ivit

yan

alys

istr

ansi

ent

Pri

estl

ey-T

aylo

rE

T,d

ay-

tim

eon

lyM

onta

lto

etal

.[20

07]

Bou

ssin

esq

wit

hD

upui

t0-

D,1

-D-X

tran

sien

tsa

tura

ted

hom

ogen

ous;

appr

oxim

ate

pref

eren

tial

flow

sens

itiv

ity

anal

ysis

cons

tant

unif

orm

Pen

man

-Mon

teit

hP

ET

atlo

wti

deon

ly

Har

vey

etal

.[19

87]

Bou

ssin

esq

wit

hD

upui

t1-

D-X

tran

sien

tsa

tura

ted

hom

ogen

ous

from

fiel

dda

ta;

sens

itiv

ity

anal

ysis

none

Nut

tle

[198

8]B

ouss

ines

qw

ith

Dup

uit

1-D

-Xtr

ansi

ent

satu

rate

dho

mog

enou

sca

libr

ated

cons

tant

unif

orm

ET¼

3.2

mm

d�1

from

Pri

estl

ey-T

aylo

rT

obia

set

al.[

2001

]0D

wat

erba

lanc

e;

1DD

arcy

1-D

-Xst

eady

stat

esa

tura

ted

hom

ogen

ous

from

fiel

dda

ta;

sens

itiv

ity

anal

ysis

tran

sien

tP

ET

mod

el

Hem

ond

and

Fifi

eld

[198

2]1D

unsa

tura

ted

Dar

cyw

ith

com

pres

sion

1-D

-Ztr

ansi

ent

unsa

tura

ted

hom

ogen

ous

sens

itiv

ity

anal

ysis

cons

tant

unif

orm

ET¼

6m

md�

1

Hem

ond

etal

.[19

84]

vert

ical

grou

ndw

ater

flow

1-D

-Ztr

ansi

ent

satu

rate

dho

mog

enou

sfr

omfi

eld

data

none

Har

vey

and

Odu

m[1

990]

0Dw

ater

bala

nce

;1D

Dar

cy1-

D-Z

stea

dyst

ate

satu

rate

dla

yere

dfr

omfi

eld

data

cons

tant

unif

orm

PE

T(v

alue

from

map

)H

ughe

set

al.[

1998

]S

EE

P/W

2-D

-XZ

tran

sien

tun

satu

rate

dla

yere

d;

mac

ropo

rosi

tyca

libr

ated

tran

sien

tP

enm

anP

ET

atlo

wti

de/

nora

inon

lyU

rsin

oet

al.[

2004

],M

aran

iet

al.[

2006

]R

icha

rds

equa

tion

2-D

-XZ

tran

sien

tun

satu

rate

dho

mog

enou

sfr

omli

tera

ture

cons

tant

unif

orm

ET¼

0or

ET¼

4m

md�

1or

8m

md�

1

Gar

dner

[200

5a,2

005b

]B

IEM

2-D

-XZ

tran

sien

tsa

tura

ted

hom

ogen

ous

sens

itiv

ity

anal

ysis

none

Li

etal

.[20

05]

TO

UG

H2

2-D

-XZ

tran

sien

tun

satu

rate

dþ

air

hom

ogen

ous

from

Urs

ino

etal

.[2

004]

cons

tant

unif

orm

ET¼

0or

ET¼

4m

md�

1at

low

tide

only

Wil

son

and

Gar

dner

[200

5]S

UT

RA

2-D

-XZ

tran

sien

tun

satu

rate

dho

mog

enou

sfr

omU

rsin

oet

al.

[200

4]no

ne

Gar

dner

and

Wil

son

[200

6]B

IEM

and

SU

TR

Ava

riat

ions

2-D

-XZ

tran

sien

tsa

tura

ted

orun

satu

rate

dho

mog

enou

sfr

omfi

eld

data

none

Wil

son

and

Gar

dner

[200

6]S

UT

RA

2-D

-XZ

tran

sien

tun

satu

rate

dho

mog

enou

sse

nsit

ivit

yan

alys

isno

neG

ardn

er[2

007]

Fle

xPD

Efi

nite

elem

ent

2-D

-XZ

tran

sien

tun

satu

rate

dla

yere

dse

nsit

ivit

yan

alys

isno

neC

arte

ret

al.[

2008

]S

UT

RA

2-D

-XZ

stea

dyst

ate

satu

rate

dho

mog

enou

sca

libr

ated

none

Col

aet

al.[

2008

]th

erm

ohyd

rom

echa

nica

lm

odel

2-D

-XZ

tran

sien

tun

satu

rate

dþ

air

hom

ogen

ous

cali

brat

edco

nsta

nt,c

alcu

late

dba

sed

on75

%am

bien

tai

rhu

mid

ity

Xin

etal

.[20

09]

SU

TR

A3-

D-X

Z(Y

)tr

ansi

ent

unsa

tura

ted

laye

red

;m

acro

poro

sity

sens

itiv

ity

anal

ysis

none

Xin

etal

.[20

11]

EL

CIR

Cþ

SU

TR

A3-

D-X

YZ

tran

sien

tun

satu

rate

dho

mog

eneo

usfr

omli

tera

ture

none

Thi

sst

udy

Hyd

roG

eoS

pher

e3-

D-X

YZ

tran

sien

tun

satu

rate

dhe

tero

gene

ous

from

fiel

dda

ta;

sens

itiv

ity

anal

ysis

spat

iall

yva

riab

le:

from

ET

fiel

dda

tafr

omm

ajor

vege

tati

onzo

nes

a Dim

ensi

ons:

X,c

reek

perp

endi

cula

r;Y

,cre

ekpa

rall

el;

Z,v

erti

cal.


3 of 22

[12] Constant fluid density, fluid viscosity, and sedimentporosity were assumed. Intertidal groundwater and surfacewater at the field site that was the focus of this investigationwere both highly saline and of comparable density. Fluiddensity, viscosity, and saturated hydraulic conductivity val-ues were representative of saline fluid (Table 2).

[13] Groundwater flux (q) was calculated as the gradientof pressure ( ) and elevation (z) heads, scaled by the unsat-urated sediment hydraulic conductivity (K � kr) :

q ¼ �K � krrð þ zÞ; (2)

where K was the saturated hydraulic conductivity tensorand kr the relative permeability given the degree of watersaturation.

[14] Saturation-pressure (Sw( )) and relative permeability–saturation (kr(Sw)) relationships were described using theMualem–van Genuchten method [van Genuchten, 1980]:

Sw ¼Swr þ ð1� SwrÞð1þ j� j�Þ��; < 0

1; � 0

((3)

kr ¼ S0:5e ½1� ð1� S��e Þ

��2: (4)

Also,

� ¼ 1� 1=�; (5)

Se ¼ ðSw � SwrÞ=ð1� SwrÞ: (6)

[15] Variables are saturation (Sw), residual saturation(Swr), pressure head ( ), effective saturation (Se), and em-pirical parameters (� and � ; see Table 2 and Figure 1).

[16] Surface water flow was represented by the 2-D dif-fusion wave approximation of the Saint Venant equationfor unsteady shallow water flow, assuming depth-averagedflow velocities, hydrostatic surface water heads, and negli-gible inertial terms:

@

@xdoKo

@ho

@x

� �þ @

@ydoKo

@ho

@y

� �� doGo6 Qo ¼ �

@ð�ohoÞ@t

:

(7)

Variables are fluid sources and sinks (Qo), exchange withthe subsurface (Go), a rill storage factor (�o : 0 at surfaceand 1 at top of rills), isotropic surface conductance (Ko),and the surface water head (ho) equal to the depth of water(do) above the local surface elevation (zo). Surface conduct-ance (Ko) was defined by Manning’s formula, via the

Table 2. Simulation Parameters Common to All Model Scenarios

Parameter Value Source

Manning roughness coefficient 6 � 10�7 d m�1/3 smooth surface, isotropic in x and y directions (equivalent to 0.05 s m�1/3)[Tsihrintzis and Madiedo, 2000]

Surface-subsurface exchange coefficient lex 10�3 Ebel et al. [2009]Fluid density � 1027 kg m�3 saline waterFluid viscosity � 90.72 kg m�1 d�1 saline waterSediment specific storage (Ss) 10�3 m�1 sensitivity analysis (Appendix A)Sediment porosity � 0.5 clay sediment [Schaap and Leij, 2000]Unsaturated sediment constitutive relationships Mualem–van Genuchten [van Genuchten, 1980]Sediment residual saturation �r 0.098 clay sediment [Schaap and Leij, 2000]van Genuchten � 1.78 m�1 clay sediment [Schaap and Leij, 2000]van Genuchten � 1.3 clay sediment [Schaap and Leij, 2000]Initial condition simulated heads following 7 days of gravity drainage, as if a nonflooding

(neap) tide period without tidal oscillationsNontidal boundary conditions zero flux through model base and model sides AB-BC (Figure 2)

Figure 1. Unsaturated clay sediment constitutive relation-ships used in all model simulations, after van Genuchten[1980] and Schaap and Leij [2000].


4 of 22

direction of maximum surface water velocity gradient (s)and a friction coefficient (n, Table 2):

Ko ¼ d2=3o

@ho

@s

� ��1=2

n�1: (8)

[17] The 2-D surface water domain was identical to thetop (ground) surface of the 3-D unsaturated groundwaterdomain. The two domains were coupled by the first-orderexchange coefficient (dual-node) approach. The volumetricgroundwater–surface water exchange flux (Go) is

Go ¼ðkrsKszÞðh� hoÞ=ðlexdoÞ: (9)

Variables are vertical saturated subsurface hydraulic con-ductivity (Ksz), groundwater head (h), surface water head(ho), an exchange coefficient (lex, Table 2), and surfacerelative permeability (krs). The relative permeability (krs)for flow from the subsurface to the surface was given byequation 4. The relative permeability (krs) for flow from thesurface to the subsurface was equal to one, permitting freeinfiltration.

4. Field Site and Model Setup4.1. Hydrologic Setting

[18] An intensively studied field site guided the develop-ment of the numerical models. The site was in the PaloAlto Baylands Nature Preserve (37�2705400N, 122�605800W),San Francisco Bay, California, USA. The central field sitethat was the focus of the study consists of a flat marsh plainbounded by one intertidal channel and bisected by a smallerchannel (Figure 2). Local tidal harmonics are mixed semi-diurnal. Low tides recede beyond the marsh to the mudflatsand subtidal San Francisquito Creek channel. The higher oftwo daily spring tides generally overtops the tidal channelsand floods the marsh plain. Neap tides may not exceedbankfull channel capacity and so may not flood the marshfor a few days at a time.

[19] The near-surface sediments at the site are texturallyclay, with 61.82% clay, 35.54% silt, and 2.64% sand, onaverage (based on 23 core samples 0–30 cm depth; onlinesupplement to Moffett et al. [2010]). Regional sedimentsconsist of at least 3 m of low-permeability estuarine mudoverlying the uppermost of a series of alluvial aquifersinterbedded with marine clay [Hamlin, 1983].

[20] Unlike in many conceptual models of coastalhydrogeology, there is no fresh groundwater dischargefrom inland to the salt marshes of Palo Alto, California.Helley and Lajoie [1979] described four mechanisms ofsaline intrusion into local shallow aquifer systems, but nogroundwater discharge to the near surface. At our field site,a levee separated the high intertidal marsh plain from thesubsided, nontidal inland area occupied by the Palo AltoAirport. A well installed on the inland (airport) side of thislevee perpetually recorded total heads lower than those inthe salt marsh over a period of three years, supporting aconceptualization of the site in which there is no shallowregional groundwater flow to the salt marsh from inland.

4.2. Model Domain

[21] The numerical model’s finite element surface meshwas constructed of �1 m wide triangular elements. Ele-ments in locations of greater than 15� local topographic

slope (channel banks and levee edges) were refined to�0.5 m wide elements. Nodes were assigned interpolatedvalues of surface elevations. Surface topography withinthe central field site was represented by a kriged modelof 742 surveyed surface elevations of centimeter accuracy[Moffett, 2010, Figure 6–15]. The central field site, enclosedby tidal channels and levees, had an average elevation of1.02 6 0.06 m above mean sea level (� 6 1) over a 0.96hectare area. Tidal channels around the central field sitewere located by walking the banks with a continuously re-cording GPS unit and channel bank heights were measuredat regular intervals.

[22] To enable logical prescription of boundary condi-tions, the model domain was extended beyond the centralfield site to cover a total area of 2.2 hectares. The topogra-phy in the areas external to the central field site (levees,mudflat, and adjacent marsh) was based on 1 m lidar dataof coarse vertical accuracy (10–20 cm [TerraPointUSA,2005]), resulting in a rougher appearance in Figure 2. TheLIDAR data were registered to the local mean sea leveldatum via regression with collocated survey data.

[23] The 3-D finite element mesh for groundwater simu-lation was extended from the surface to 10 m below meansea level for a total depth of approximately 11 m beneaththe marsh plain. This large model depth was chosen toensure the influence of the bottom boundary on near-surfacehydraulic dynamics would be negligible. Refined detail in

Figure 2. Field site map. Analysis focused on the centralfield site, within the dashed lines. The full extent of themodel domain (A-B-C-D) is outlined by the solid lines.Black circles mark in situ and modeled piezometers. Insetsshow the location within northern California and southernSan Francisco Bay (37�2705400N, 122�605800W).


5 of 22

the near-surface depths of interest was provided by nodesheets parallel to local topography at depths of 0.1, 0.2, 0.3,0.5, and 1.0 m. The model totaled 6 layers, 43,443 nodes,and 86,242 elements.

[24] The surface and subsurface flow equations weresolved using adaptive model time steps. Nodal saturationchanges greater than 5% and nodal head changes greaterthan 1 cm between time steps were prohibited to avoidovershooting the nonlinear hydraulic dynamics in the un-saturated zone during a time step. The time step durationwas decreased until these incremental changes were withinallowed levels.

4.3. Initial Conditions and Fixed Boundary Conditions

[25] Initial conditions for all simulations were the hy-draulic heads resulting from a simulation of 7 days of grav-ity drainage initiated with the following conditions: a fullysaturated domain, a spatially uniform specified head of3 m, and a specified head boundary of �0.25 m alongmodel sides CD-DA (see Figure 2). The initial head of 3 mabove mean sea level placed about 2 m of water on themarsh surface and so ensured complete flooding and satura-tion prior to the development of partially drained condi-tions. This approach to developing initial conditions isconsistent with established groundwater modeling practice[e.g., Loague and VanderKwaak, 2002; Mirus et al., 2011].This initial simulation period may be intuitively thought ofin this case as if simulating a simplified (nonoscillatory),neap tide period immediately prior to testing subsequentmodel scenarios. This initial simulation was conducted withhomogeneous, isotropic saturated hydraulic conductivity andspecific storage values, within the ranges of values appropri-ate for clay sediments (K ¼ 0.204 m d�1 and Ss ¼ 10�4

m�1 [Schaap and Leij, 2000]).[26] Starting from this partially drained initial condition,

we developed a suite of model scenarios to test the relativeinfluences of tidal regime, sediment hydraulic properties,and zonally distributed variations in plant water use on thegroundwater and surface water hydrology of the field site.In each simulation, no vertical groundwater flow was per-mitted across the model base nor horizontal flow across themodel edge symmetry boundaries AB-BC (Figure 2). Othersimulation parameters common to all the model scenariosare listed in Table 2.

4.4. Lateral Boundary Conditions: Tidal Scenarios

[27] Lateral model faces CD-DA (Figure 2) were simu-lated as subsurface and surface specified head boundaries.Head values were interpolated to match the adaptive timesteps of the model from water levels recorded every 10 minat a temporary tide gauge at the field site on 23–24 December2007 (Figure 3). We compared scenarios simulating aflooding or nonflooding tide because both occur frequentlyat the field site because of the mixed semidiurnal tidal sig-nal. The tidal boundary condition for the flooding scenariowas given by 12 h of data that included a tide that exceededbankfull stage and flooded the marsh surface. The tidalboundary condition for the nonflooding scenario was givenby 12 h of data that included a maximum tidal stage justbelow bankfull tidal channel capacity.

[28] Both tidal scenarios permitted 4.5 h of drainage fol-lowing the bankfull stage of the ebb tide prior to the end of

the simulation (Figure 3), at which time results wereevaluated. During low-tide periods, the total heads at thetidal boundaries (locations along edges CD-DA, all withsurface elevation �0.26 cm or lower) were maintained at�0.25 m to simulate at least 1 cm of persistent surfacewater along the boundary, as observed in the field.

4.5. Porous Media Hydraulic Properties:Sediment Scenarios

[29] Most salt marsh groundwater models to date havesimulated a marsh of homogeneous sediments (see Table 1).In this study, we compared a simple homogeneous, isotropicsediment scenario to a scenario of heterogeneous sedimentstructures underlying the field site that were determined viapiezometer data and numerical model sensitivity analysis.The sediments of the homogeneous scenario were isotropicwith hydraulic conductivity K ¼ 7 � 10�2 m d�1. Sensitivityanalysis suggested this value was most suitable for locationsthat displayed substantial hydraulic responses to tidal forcing(see Appendix A). The sediments of the heterogeneous sce-nario were arrayed in a layered system of spatially variablehydraulic conductivity values as depicted in Figure 4a.

[30] The development of the heterogeneous sedimentscenario shown in Figure 4a is detailed in Appendix A. Inbrief, groundwater head responses to tidal forcing wererecorded at 14 field locations in piezometer pairs screenedat depths of 0.5 m or 1.0 m below the ground surface(Figure 2). Model sensitivity analysis was conducted usingmultiple realizations of saturated hydraulic conductivity,specific storage, porosity, and layering, with all parametersin ranges appropriate for the clay sediments of the fieldsite. The simulated response of groundwater heads to aflooding tide was found to be sensitive to the specific storageand the vertical hydraulic conductivity of the clay sedimentsbut insensitive to layering, horizontal hydraulic conductivity,and porosity. The sensitivity analysis led to a conceptualmodel of sediment heterogeneity at the field site (Figure 4a),which was partially based on historical field site geomor-phology (Figure 4b), as described further in Appendix A.

4.6. Surface Boundary Conditions:Evapotranspiration Scenarios

[31] The surface conditions of the model were varied totest the effects of evapotranspiration patterns on marsh

Figure 3. Flooding and nonflooding tidal boundary con-ditions based on tidal data recorded at the field site andapplied to model faces CD-DA for 12 h simulations.


6 of 22

hydrologic response. Two scenarios used common methodsfrom other modeling studies and a third accounted for thedistinctive vegetation zonation of salt marshes.

[32] 1. For the zero ET scenario, the surface boundarycondition was determined by the simulated surface waterhydraulics as either a positive pressure head or a seepageface, with zero specified ET flux.

[33] 2. In the uniform ET scenario, a spatially uniform,constant upward flux of 4 mm d�1 [Ursino et al., 2004](see also Dacey and Howes [1984] as discussed by Li et al.[2005]) was distributed linearly with depth through the top0.2 m (two model layers) of sediments. The linear weight-ing approximated the decline in root biomass with depthobserved in salt marshes, to an extinction depth [Chapman,1938a; Howes et al., 1981; Ellison et al., 1986; Steinkeet al., 1996; Boyer et al., 2000], and the greater effect of soilevaporation at the surface. In locations of standing water, thespecified flux was met first by surface evaporation.

[34] 3. In the zonal ET scenario, various evapotranspira-tion rates were distributed linearly over various rootingdepths according to the evapotranspiration rates and vege-tation zonation of dominant salt marsh plant species deter-mined from the field site, as depicted in Figure 5. Thedevelopment of Figure 5, evapotranspiration rates, androoting depths is detailed in Appendix B.

5. Model Results5.1. Salt Marsh Ecohydrological Zones

[35] Intricate spatial variations in simulated root zone hy-drology were produced upon combining in one model, inter-specific differences in plant water use, vegetation zonation,plant-water interactions, 3-D variably saturated groundwaterhydrology, and variable degrees of 2-D tidal flooding. Theintricate spatial variations in root zone hydrology did notappear in simpler model scenarios that lacked some of thesecomponents. The multifaceted model, intended to representconditions as close as possible to those at the field site, werefer to in the remainder of this paper as the complex modelscenario: it included heterogeneous sediment properties asin Figure 4a and spatially variable evapotranspiration androoting depths as in Figure 5 for flooding or nonfloodingtidal conditions as in Figure 3.

[36] Comparison of the complex model to simpler scenar-ios showed that approximately seven different root zonehydrological environments occurred at the same time indifferent regions of the complex model and that these differ-ent environments were caused by different local combina-tions of hydraulic and vegetative influences, not by eitherinfluence alone (Table 3). We termed these distinct rootzone hydrological environments ecohydrological zones.

Figure 4. (a) Modeled sediment heterogeneity, developed from sensitivity analysis of simulated headscompared to data from 28 piezometers (Appendix A, with examples from labeled locations L, M, X, andQ). Saturated hydraulic conductivity values (K, m d�1) were prescribed at two depths: shallow K valuesbetween the surface and about 0.5 m depth and deep K values between about 0.5 and 1.0 m depth. Tidalchannels and the approximate extent of levee compaction are outlined. The curved shape of sedimentzones was partly based on the infilling of historical tidal channels that occurred since 1921. (b) Historicaltidal channels identified from 1921 aerial photography and a 1857 coastal survey [Hermstad et al.[2009], displayed over 2004 aerial photography [National Geospatial-Intelligence Agency, 2004].


7 of 22

Ecohydrological zones are two or more adjacent subregionsof an ecosystem exhibiting substantially different hydraulicconditions in their root zones because the subregions are sep-arated by contrasts in plant water use, hydraulic properties,or both. The hydraulic conditions noted in this study that dif-fered substantially between ecohydrological zones were root

zone pressure head, saturation, and vertical groundwatervelocity. Contrasts in hydraulic properties that contributed toecohydrological zonation in this study were provided bysediment hydraulic conductivity heterogeneity and variationsin topography and tidal influence. Contrasts in vegetationwater use in this study were provided by spatially variable,species-specific evapotranspiration rates and rooting depths.The remainder of this section presents the ecohydrologicalzonation in the models and field data due to these vegetativeand geological sources of heterogeneity. Additional potentialinfluences on ecohydrological zonation are discussed insection 6.1.

[37] The first of two topographically influenced ecohy-drological zones (EHZ) was indicated by large negativepressure heads in the salt marsh channel banks (abbreviatedEHZ 1). According to comparisons of different model sce-narios after a nonflooding tide, the extent of this zone awayfrom the tidal channels was sensitive to multiple local hy-draulic influences. The width of EHZ 1 surrounding thecentral field site channel was narrow when the channel wassurrounded by the relatively high conductivity sediments ofthe homogeneous sediment scenario (Figures 6c and 6d)and wider when surrounded by the lower conductivitysediments of the heterogeneous sediment scenario (Figures6a and 6b). The width of EHZ 1 also varied with evapo-transpiration regime: it was narrow given spatially uniformevapotranspiration (Figures 6b and 6d) and wider with spa-tially variable, zonal evapotranspiration and rooting depths(Figures 6a and 6c).

[38] A second topographically influenced ecohydrologi-cal zone was indicated by ponded infiltration in surfacedepressions (EHZ 2), characterized by positive pressureheads, saturated sediments, and downward groundwaterflow (Table 3). The extent of this zone again appeared sen-sitive to both sediment and evapotranspiration variations.Ponds were less extensive in the heterogeneous sedimentscenario, which included a large central area of low hy-draulic conductivity (Figures 6a and 6b), compared to the

Table 3. Schematic of Causes and Consequences of Seven Ecohydrological Zones and an Eighth Hydrological State Exhibited for aLimited Time After Tidal Flooding

Salt Marsh Ecohydrological Zone CausesaConsequences

(in Surficial Sediments)

1: Channel bank topography: channel bank very large � , low S, 6Vzb

2: Pond topography: surface depression þ , S 1, small �Vz

3: High K, high ET heterogeneity in sediment hydraulics, evapotranspiration, and rootingdepth: high K, high ET

large � , medium S, large þVz

4: High K, low ET heterogeneity in sediment hydraulics, evapotranspiration, and rootingdepth: high K, low ET

medium � , medium S, small þVz

5: Low K, high ET heterogeneity in sediment hydraulics, evapotranspiration, and rootingdepth: low K, high ET

very large � , very low S, small þVz

6: Low K, medium ET heterogeneity in sediment hydraulics, evapotranspiration, and rootingdepth: low K, medium ET

large � , medium S, �0 Vz

7: Low K, low ET heterogeneity in sediment hydraulics, evapotranspiration, and rootingdepth: low K, low ET

small � , high S, small �Vz

Post-flooding hydrological statec recent flooding 6 0, very high S, small 6Vz

aCauses tested in this study were tidal regime, local topography, evapotranspiration (ET) rate and rooting depth, and saturated sediment hydraulic con-ductivity (K). Consequences observed were pressure head ( ), saturation (S), and vertical groundwater velocity (Vz ; plus is upward, and minus isdownward).

bDirection of near-surface groundwater flow locally depends on relative influence of upward flow by ET extraction and downward flow by drainage.cAn eighth hydrological state occurs shortly after a flood event; as a temporal, not spatial, condition and attributable to tidal flooding alone, it is not

defined as an ecohydrological zone.

Figure 5. Field site vegetation zonation. Major vegeta-tion zones are shown by dominant plant species in differentcolors. Evapotranspiration rates and rooting depths used inthe spatially variable evapotranspiration models are indi-cated in the legend (also see Appendix B).


8 of 22

broad flooded areas that remained after a nonfloodingtide in the homogeneous-sediment scenario (Figures 6c and6d). Although the low hydraulic conductivity region inthe heterogeneous-sediment scenario might logically havesupported perched surface water, evaporative water lossreduced the amount of water available at the surface,accounting for these results.

[39] Five distinct ecohydrological zones appeared dueto the combined influence of sediment and vegetationheterogeneity in the complex model scenario during theperiod of marsh exposure after a nonflooding tide (seeTable 3). Two of these EHZ emerged among sediments ofrelatively high hydraulic conductivity. In some areas, highevapotranspiration rates superimposed on high hydraulicconductivity sediments resulted in very large negative sur-ficial pressure head values (Figure 6a), moderate saturationvalues (Figure 7a), and relatively rapid, upward unsatu-rated flow (Figure 7b) (EHZ 3, Figure 7c). In other areas,low ET rates among high K sediments produced smallerpressure heads (Figure 6a) and groundwater velocities(Figure 7b) (EHZ 4, Figure 7c).

[40] Among low hydraulic conductivity sediments, dif-ferent evapotranspiration rates distinguished three ecohy-drological zones. High ET among low K sediments resultedin little groundwater movement (Figure 7b) but very largenegative pressure heads (Figure 6a) and saturation valuesbelow field capacity (Figure 7a) (EHZ 5, Figure 7c). Theespecially low pressure heads simulated in this zone were

facilitated by the shape of the unsaturated characteristiccurves modeled for the clay marsh sediments (Figure 1).These constitutive relationships caused a relatively largechange in pressure head for a small change in saturation:the soil remained 99% saturated at a pressure head of ¼�0.05 m, 98% saturated at ¼ �0.1 m, and 94% saturatedat ¼ �0.25 m. The hydraulic influences of low K sedi-ments and moderate ET offset each other in some locations,resulting in large negative pressure heads (Figure 6a), mod-erate soil saturations (Figure 7a), and root zone stagnationwith approximately zero vertical root zone fluid velocity(black in Figure 7b) (EHZ 6, Figure 7c). Finally, very lowET among low K sediments induced only small negativepressure heads (Figure 6a) and hardly changed the high sat-uration and low downward groundwater infiltration ratesremaining from prior saturated conditions (cyan in Figure7b) (EHZ 7, Figure 7c).

[41] The differentiation of these distinct ecohydrologicalzones in the complex model scenario after a nonfloodingtide was qualitatively validated by tensiometer data col-lected in the field. Prior to the modeling study, tensiometersattached to a stabilizing stake were pushed into the rootzone to 10 cm depth (ceramic cup spanning 5 to 15 cm;locations in Figure 2) and monitored manually during thecontrasting hydrological conditions of neap and springtides. At the end of a neap tide period, during which themarsh was not flooded for 8 days, tensiometer pressureheads were lowest in EHZ 3, 5, and 6 (Figure 8a).

Figure 6. Effects of spatially variable sediment hydraulic properties and evapotranspiration after anonflooding tide: surficial pressure heads ( , in m) 4.5 h after bankfull (maximum) tidal stage. Blackareas indicate a pressure head of zero at the land surface, i.e., conditions of incipient ponding due to asurficial water table. Positive pressure heads at the surface (in blue) indicate the depth of standing water.


9 of 22

Tensiometer pressure heads after this prolonged, neap tideexposure were slightly higher in EHZ 4, and highest inEHZ 7. These field data corroborated the ecohydrologicalzonation framework summarized in Table 3.

5.2. Postflooding Hydrological Conditions

[42] In addition to the seven ecohydrological zones iden-tified from nonflooding tidal conditions, an eighth type ofhydrological environment temporarily prevailed in the saltmarsh for a limited time after tidal flooding. According tothe tensiometer field data, spatial variations in near-surfacepressure heads were nearly eliminated after a flooding tide(Figure 8b). According to the model simulations, tidalflooding temporarily overwhelmed any significant spatialvariations in root zone hydraulic conditions that might havebeen induced by variations in sediment hydraulic conduc-tivity and evapotranspiration, leaving the marsh surface ina relatively uniform state (Figure 9), as indicated by thetensiometer data. In simulations, soon (4.5 h) after a flood-ing tide, variations in sediment hydraulic conductivity had

only a slight effect on spatial variations in surficial pressureheads in the marsh interior and variations in evapotranspi-ration had very little effect (Figure 9).

5.3. Temporal Development of Ecohydrological ZonesFollowing Flooding

[43] Since tidal flooding masked underlying ecohydro-logical zonation: How do the various root zone hydraulichabitats associated with ecohydrological zonation reappearfollowing a flooding tide? To better understand the tempo-ral development of EHZ after flooding, we conducted asimulation of extended marsh exposure during neap tide.This extended model was based on the previous complexmodel scenario: it included heterogeneous sediment hy-draulic properties and spatially variable evapotranspirationand rooting depths (as in Figures 4a and 5). The modelspanned a 6 day period beginning with a flooding tide, fol-lowed by 5.5 days of nonflooding tides: the neap tide pe-riod recorded at the field site on 16–21 October 2007(Figure 10a). Root zone hydraulics at six locations within

Figure 7. Effects of spatially variable sediment hydraulic properties and evapotranspiration after anonflooding tide: (a) surficial saturation and (b) vertical groundwater velocity (in m d�1) 4.5 h afterbankfull (maximum) tidal stage. In Figure 7b, black areas indicate near-surface vertical groundwaterflow reversal or stagnation (Vz 0). The results in Figures 7a and 7b correspond to the pressure headresults in Figure 6a. (c) Seven ecohydrological zones, constituting different root zone hydraulic habitats,emerge from local combinations of heterogeneous soil hydraulic properties and spatially variable evapo-transpiration rates and rooting depths.


10 of 22

each of the spatially extensive EHZ (2–7) were monitoredfor the duration of the simulation. Model element discreti-zation was coarsened away from these observation regionsto preserve model runtime efficiency during the extendedsimulation period.

[44] During the neap tide simulation, each of the EHZfollowed a distinct trajectory in developing its characteristichydraulic conditions. In topographic depressions (EHZ 2),the root zone remained saturated for at least two days afterflooding (2.5–3 days into the simulation), despite ongoingevaporation (black lines in Figure 10). The rest of themarsh quickly developed unsaturated conditions, althoughvia various trajectories. High K sediments were saturatedby the flooding tide and then exhibited moderate rates ofpressure head and saturation decline at 10 cm depth (cyanand magenta lines, EHZ 3 and 4, Figures 10b and 10d). Incontrast, low K sediments maintained unsaturated condi-tions at 10 cm depth during surface flooding (red, blue, andgreen lines, EHZ 5–7, inset in Figure 10d) but eventually

developed the lowest pressure heads and saturations, withtheir relative conditions influenced by the high, medium, orlow imposed ET (Figures 10b and 10d). Deeper in the rootzone, the relative conditions in low K and high K regionswere reversed, with greater desaturation occurring in highK regions than in low K regions at 20 cm depth over theneap tide period (Figures 10c and 10e).

[45] Each EHZ’s distinctive unsaturated conditions wereaccompanied by different shallow vertical groundwaterflow regimes. In high K regions and surface depressions,there was a short spike of infiltration during the initialflooding tide (EHZ 2–4, inset in Figure 10f) but the verticalspecific discharge then increased precipitously to meet theET demand. In low K regions, the initial flooding tide didnot perturb the slight upward flux to the surface induced bymedium to high ET rates (EHZ 5–6, inset in Figure 10f)nor the slight downward flux at the surface in the low K,low ET zone (EHZ 7, inset in Figure 10f), though the fluxdirection in the latter zone (EHZ 7) reversed about half aday after flooding.

5.4. Cumulative Surface Water–GroundwaterExchange

[46] Since spatially variable sediment properties and evap-otranspiration greatly influenced local marsh hydraulics:Did they also influence cumulative groundwater–surfacewater exchange? To answer this question we compared twolimiting case scenarios: the complex model with heteroge-neous sediments and spatially variable evapotranspiration(HetK/VarET, as in Figures 4a and 5), and simple modelwith homogeneous sediments and no evapotranspiration(HomK/NoET). Absolute model results may have beeninfluenced by the overestimation of infiltration by Rich-ards’ equation [Li et al., 2005; Wilson and Gardner, 2006;Tosatto et al., 2009], but comparison between models wasstill informative.

[47] Cumulative simulated surface water–groundwaterexchange rates were primarily influenced by sediment hy-draulic properties, with lower infiltration and exfiltrationrates during high tides in the complex HetK/VarET sce-nario (Figures 11c–11f); evapotranspiration had no appa-rent effect. In contrast to the exchange rate, both sedimenthydraulic properties and evapotranspiration influenced theexchanged volume and the volumetric magnitudes of infil-tration and pore water flushing (Figures 11g–11j). The sim-ple HomK/NoET scenario showed considerable pore waterflushing, indicated by the high maximum infiltrated volumereached and the large net exfiltration at the end of thesimulation. In contrast, the complex HetK/VarET scenarioshowed only modest infiltration, yet some of that infiltratedwater was retained as a net positive infiltration volume atsome marsh locations at the end of the tidal period, replac-ing water removed by evapotranspiration. Continued exfil-tration after the start of tidal infiltration is expected in saltmarshes if 3-D marsh geometry is accounted for [Harveyet al., 1987; Xin et al., 2011]. The rates of change ofvolumetric exchange were more gradual in simulations of anonflooding tide than in simulations of a flooding tide(Figures 11h and 11j versus 11g and 11i).

[48] Vertical profiles of groundwater head and dischargetaken from the same two model scenarios provided moredetailed insight into the distribution of, and controls on,

Figure 8. Tensiometer pressure head statistics (median,first and third quartiles, maximum, and minimum) for tensi-ometers located among the five sediment- and vegetation-controlled ecohydrological zones (EHZ 3–7) at 10 cm depth(locations in Figure 2). (a) After a nonflooding neaptide period (19 November 2007) and (b) after a flood tide(7 December 2007). Number (n) of tensiometer measure-ments per EHZ is indicated.


11 of 22

groundwater–surface water exchange (Figure 12). Ground-water discharge from the channel banks appeared anoma-lously large in the simple HomK/NoET scenario but moresubtle in the complex HetK/VarET scenario. Four notableaspects of the shallow groundwater flow paths and ground-water discharge in the complex HetK/VarET scenario aredescribed in the following list and labeled in Figure 12 bycorresponding Roman numerals (i–iv).

[49] In the locations labeled i in Figure 12, depressionstorage (EHZ 2) supplied water to adjacent high K, high ETzones (EHZ 3).

[50] Groundwater discharge to tidal channels was sup-pressed in the locations labeled ii in Figure 12 by locallyhigh ET, which reduced the heads around the channel (inEHZ 1, Figure 12b).

[51] Groundwater discharge was suppressed in the loca-tions labeled iii in Figure 12 by low K outcrops. For example,discharge was suppressed around an island in the main tidalchannel because the channel beds intersected low K sedi-ments below 0.5 m elevation (Figure 12f). At the same time,the ET ‘‘discharge’’ from the top of the island was enhancedin an effect similar to that in location i in Figure 12, only inthis case the water ‘‘source’’ was suppressed drainage. Dis-charge was also suppressed where low K sediments out-cropped around the levee and at the shoreline in Figure 12h.

[52] In addition to the effects of heterogeneous sedimentsand ET, the influence of rooting depth on root zone hydraul-ics was apparent in the locations labeled iv in Figure 12. For

example, strong vertical head gradients were induced in theroot zone by regions of high ET and shallow (10 cm) rootsamong low K sediments (EHZ 5) in Figures 12d and 12h.

6. Discussion6.1. Ecohydrological Zonation

[53] This modeling study demonstrated how ecologicaland hydrogeological heterogeneity can combine to creatediverse root zone hydraulic habitats in even a small saltmarsh area, despite the largely homogeneous influence oftidal flooding on a flat marsh plain. Considering heteroge-neity in either vegetation or sediments, alone, resulted invery little variability among root zone hydraulic conditions.The intricate mosaic of root zone hydraulic conditions thatemerged in the complex model due to the intersection ofheterogeneous vegetative and hydrological influences con-stituted ecohydrological zonation. Although coastal hydrol-ogy and vegetation have separately been recognized asinfluencing salt marsh groundwater dynamics for nearly acentury [e.g., Johnson and York, 1915; Chapman, 1938a;Mahall and Park, 1976c; Hemond and Fifield, 1982;Dacey and Howes, 1984; Harvey et al., 1987; Nuttle,1988; Howes and Goehringer, 1994], a spatially explicitconceptual model that integrates both hydrogeological andecological influences on salt marsh groundwater dynamicshas not previously been proposed, which is what is cap-tured in the concept of ecohydrological zonation.

Figure 9. Effects of spatially variable sediment hydraulic properties and evapotranspiration after aflooding tide: surficial pressure heads ( , in m) depicted 4.5 h after bankfull ebb tidal stage. Black areasindicate a pressure head of zero at the land surface, i.e., conditions of incipient ponding due to a surficialwater table. Positive pressure heads at the surface (in blue) indicate the depth of standing water.


12 of 22

[54] Ecohydrological zonation encapsulates both visuallyobvious, above-ground wetland patterning and hidden,below-ground hydraulic patterning, which are each integralparts of overall wetland ecohydrology. Since this studyexamined only spatial variations in evapotranspiration, root-ing depth, sediment hydraulic conductivity, and topography,

it is likely that our definition of ecohydrological zonation(see section 5.1) captured only some components of abroader suite of intersecting spatial patterns that contributeto wetland organization. Additional relevant system attrib-utes, the overlay of which might produce a richer picture ofthe three dimensional basis for salt marsh habitat complexity,

Figure 10. Development of ecohydrological zones’ distinct root zone hydraulics over a neap tideperiod. (a) The neap period began with a flooding tide, followed by 5.5 days of nonflooding conditions.(b–g) Hydraulic response simulated at 10 and 20 cm soil depths within ecohydrological zones 2–7 (linesshow median of six locations monitored per zone).


13 of 22

include: heterogeneity in sediment storage properties, mapsof macroporosity or bioturbation, root zone biogeochemicalgradients (e.g., inorganic or microbial or due to root exu-dates), and spatial patterns in meteorological influences (e.g.,prevailing wind direction, average solar angle). Whateverintersecting system attributes ultimately contribute to a fulldefinition, we suggest that ecohydrological zones are the fun-damental habitat units comprising the salt marsh ecosystem.This perspective contrasts with a century-long focus on visu-ally obvious vegetation patterns as the major unit of habitatvariation in salt marshes. The ecohydrological zonation con-cept also contrasts with previous conceptual models of wet-lands that treat vegetation as a consequence of, not also acontributing cause of, spatial and temporal variations in thephysical variables used to define the wetland hydrologicalsystem [e.g., Brinson, 1993].

[55] The idea of a salt marsh being segmented into manyecohydrological zones remains to be tested further withnew field studies. Of the seven ecohydrological zones weidentified, only one has been studied in detail : the channelbank environment (EHZ 1) [e.g., Howes et al., 1981;

Howes and Goehringer, 1994; Ursino et al., 2004; Maraniet al., 2006; Tosatto et al., 2009]. Howes and Goehringer[1994] illustrated this zone’s principle characteristics: sub-stantial groundwater drainage, relatively high groundwaterflow rates, and comparatively low soil saturation betweenhigh tides. Although this zone is generally thought toextend 1–2 m from the channel bank in muddy sediments[Hughes et al., 1998], our models showed that its widthdepends on multiple local factors including near-channelsediment hydraulic properties and evapotranspiration mag-nitude and distribution.

6.2. Limitations of the Models

[56] Although we believe ecohydrological zonation to bea generally useful concept, the details of the zonation pro-posed by this study are subject to some limitations. To cap-ture the topographic, geologic, and vegetative complexityof a natural salt marsh, our models were necessarily site-specific. Much of the ecohydrological zonation that mightbe expected in intertidal high-marsh settings may havebeen represented among the many combinations of

Figure 11. Cumulative marsh groundwater balances simulated during (a, c, e, g, i) flooding or (b, d, f,h, j) nonflooding tidal conditions for limiting-case models of homogeneous sediments (HomK) or hetero-geneous sediments (HetK) and no evapotranspiration (NoET) or spatially variable evapotranspiration(VarET). Periods of rising and falling tidal stage are set apart by vertical dashed lines; results are pre-sented from the start of the rising tide. Note that the y axis scales in Figures 11h and 11j are half those inFigures 11g and 11i.


14 of 22

evapotranspiration rates, rooting depths, and sedimentproperties we tested, but the number of zones and their spe-cific characteristics are likely to vary from site to site.

[57] In general, salt marshes that are slowly accreting orsubsiding, such as low-energy lagoon or estuarine systems,are likely to share the fine sediment texture and low topo-graphic relief of our field site. Reported saturated hydraulicconductivity values similar to our 0.07–0.0007 m d�1 valuesinclude an Australian estuarine marsh, 0.01–0.17 m d�1

[Hughes et al., 1998], and a Venice lagoon marsh,0.00017 m d�1 (core) and 0.086–0.00086 m d�1 (models)[Cola et al., 2008]. Regarding topography, the use of akriged surface in the central portion of our models did notcapture microtopographic variations in surface elevation

that might affect very local hydraulics. However, suchmicrotopography is generally considered to occur at meterscales or less [Gray and Scott, 1977; Tessier et al., 2002;Morzaria-Luna et al., 2004; Varty and Zedler, 2008] and sowould have been below the resolution of our finite elementmesh even if such fine scale data had been available. Thecentral field site, on which we focused, was less than onehectare in area and was physically surveyed at 742 pointswith centimeter accuracy. The remaining outer portions ofthe model that were represented using lower-accuracy lidardata were physically and hydraulically separated from thecentral site by large tidal channels or levees. These outerportions of the model were not used to draw conclusionsand were only included to permit logical prescription of

Figure 12. Vertical sediment profiles showing groundwater head and discharge for two limiting-casemodels: (a, c, e, g) homogeneous sediments with no evapotranspiration (HomK/NoET) and (b, d, f, h)heterogeneous sediments with spatially variable evapotranspiration (HetK/VarET). Results shown arefrom 4.5 h after the bankfull stage of a nonflooding tide (as in Figures 6a and 7). Sets of nodes on profilesare located at 0, 10, 20, 30, 50, and 100 cm below ground surface and qualitative discharge vectors beginat each node. The bottom edge of each profile shown is at 0.75 m above sea level. Scale bars and legendin Figure 12a apply to all profiles. The inset in Figure 12c shows profile locations (background map as inFigure 5). Arabic numerals (1–7) mark ecohydrological zones. Roman numerals (i–iv) mark locations ofinterest referred to in section 5.4.


15 of 22

boundary conditions away from the central field site.Therefore, it is unlikely that the surface topography in theseouter areas substantially influenced the results in the centralfield site area.

[58] Even our most complex model was not utterly eco-hydrologically comprehensive; not all potentially relevantprocesses were included. As a first approximation, we rep-resented plant water use as if the plants were anisohydric orhydrolabile: plants that continue using water at a relativelyconstant rate and do not exhibit water uptake limitation forextended periods. This assumption of temporally constantevapotranspiration is a good approximation for halophytesable to use osmotic regulation to extract water even at verylow soil water contents [Marani et al., 2006]; it was alsoconsistent with the constant rate of salt marsh plant wateruptake observed in lysimeters [Dacey and Howes, 1984]and with most previous modeling approaches (see Table 1).In reality, each species, and even each individual plant,may experience water and/or salt limitation at a differenttime, to a different degree, under different conditions. Thesimulation of such dynamic plant-water feedbacks wasbeyond the scope of our preliminary models. Most of ouranalysis was drawn from 12 h simulation periods includingonly 4.5 h following the bankfull stages of ebbing hightides, during which short time periods the lack of dynamicvegetation feedback is unlikely to have substantiallyaffected the results. However, the lack of dynamic plant-water feedbacks in the extended neap tide simulationmeans that the lowest pressure head and saturation valuesshown in Figure 10 might not actually be achieved in thefield since the curves would flatten out if water limitationwere reached. Still, the extended simulation, as is, providesa bound on the maximum divergence in behavior amongEHZ, given no limitation of evapotranspiration.

[59] The likely effects of processes not accounted for inthe model, such as plant water use limitation, can be qualita-tively interpreted by comparing the model results in Figure10b to the tensiometer data in Figure 8. A short time after aflooding tide, almost all the pressure head measurements inthe field were in the relatively narrow 20 cm range of �40to �60 cm (Figure 8b). Simulated pressure heads (in EHZ3–7) remained within 20 cm of each other until about2.5 days after the flood tide, at which time they also spannedthe range �40 to �60 cm. The longer time required by themodel to reach the �40 to �60 cm range might be attrib-uted to the lack of air entrapment processes in the model:persistent entrapped air is known to help maintain low pres-sure heads in the field [Chapman, 1938a]. If so, the lowpressure heads and saturations in our nonflooding modelscenarios (Figures 6 and 7) are conservative results, as addi-tional entrapped air would further lower these values. A lon-ger time after a flooding tide, the pressure heads modeled inhigh K zones (EHZ 3–4) remained roughly consistent withfield measurements, for at least a few days. Among low Ksediments, however, the model developed more pronouncedunsaturated conditions faster than were indicated by thefield data (EHZ 5–7). The omission of plant water uptake li-mitation from the model likely explains the more extremepressure head declines in the low K regions of the model,since the ability of plants to extract water from low K sedi-ments likely does become limited in the field sometime afterflooding.

6.3. Significance of Ecohydrological Zonation inCoastal Hydrology

[60] Are the transient and heterogeneous plant-waterinteractions that we have categorized as ecohydrologicalzones ecologically and hydrologically significant? Weexamine this question from three angles: first, in terms ofroot zone aeration; second, in terms of marsh groundwaterdynamics; and third, in terms of marsh-estuary exchange.

6.3.1. Root Zone Aeration[61] The potential significance of ecohydrological zonation

for salt marsh root zone aeration and vegetation productivitymay be understood by examining the most prominent ecohy-drological zone that emerged from our simulations. Thiszone, EHZ 5, was characterized by very large negative surfi-cial pressure heads, very low root zone saturations, and smallupward groundwater flow in low K, high ET regions. EHZ 5reached negative pressure heads of up to 25 cm at the marshsurface a short time after a nonflooding tide (Figure 6a), andmore over a longer, neap tide period (Figure 10b), barringsubstantial plant water limitation. However, given the unsatu-rated hydraulic response of the clay soils (Figure 1), even apressure head of just �25 cm locally lowered the marsh sur-face to 94% saturation, providing 6% aeration of the near-surface root zone pore volume.

[62] Although 6% aeration might be insignificant in ter-restrial environments of drier, coarser soils, it is very sig-nificant in intertidal settings. Salt marsh soils may have afield capacity of at least 97.1% [e.g., Bradley and Morris,1990], so only about 3% soil aeration could be caused bygravity drainage over a few days. The field capacity of oursimulated clay soils was over 98%, so soil aeration totaling6% of the pore volume indicates at least two times moreloss of water from the root zone by evapotranspiration(�4%) than by drainage (2%). The capacity for physicaldrainage is even smaller in some marshes: e.g., a minimumresidual saturation of 97.3% to 98.5% (from specific yieldof 2–3% and porosity of 75–90% [Dacey and Howes,1984]). Thus, the 6% aeration we simulated in regions ofhigh evapotranspiration and low soil hydraulic conductivity(EHZ 5) is quite significant relative to prior studies, espe-cially given that infiltration may have been overestimatedby Richards’ equation [Li et al., 2005; Wilson and Gard-ner, 2006; Tosatto et al., 2009].

[63] A related, interesting result from our simulationswas that the plant species Spartina foliosa and Salicorniavirginica caused the same degree of surficial soil aerationdespite being simulated with different evapotranspirationrates (5.6 and 3.6 mm d�1) and different rooting depths (20and 10 cm). At the marsh surface, the simulated hydraulicsof this grass and succulent were indistinguishable, leading tothem being combined in the same ecohydrological zone(EHZ 3 if in low K sediments; EHZ 5 if in high K sedi-ments). As much as 22% of total soil aeration has been attrib-uted to the grass species Spartina alterniflora growing inSouth Carolina salt marshes [Morris and Whiting, 1985], butthe same phenomenon has not previously been suggested tooccur in the root zone of the succulent Salicornia virginica.

6.3.2. Marsh Groundwater Dynamics[64] The potential significance of ecohydrological zona-

tion for salt marsh groundwater movement, and possible


16 of 22

redistribution of nutrients within the marsh, is notablefrom the surprisingly complex groundwater flow patternsinduced by the EHZ mosaic. In some areas, the spatial jux-taposition of EHZ’s contrasting root zone hydraulic condi-tions appeared to induce novel horizontal water movement.Because our marsh plain was topographically nearly flat,the spatial variation in surface pressure head in Figure 6awas almost identical to the spatial variation in total hydrau-lic head (not shown). The largest potential for horizontalflow between EHZ occurred between EHZ 3 and 4 (Figure6a, complex model, nonflooding scenario), where the in-stantaneous specific discharge rate at the end of the shortsimulation was as high as 0.04 cm d�1 near the marsh sur-face. Although small, this potential vegetation-induced flowwas not insignificant compared to potential horizontal flowinduced by topography. Topography induced 1–2 cm d�1

instantaneous specific discharge at the marsh surface at theedge of the ponded infiltration regions (edges of EHZ 2)and 8 cm d�1 near the channel banks (in EHZ 1). Evapo-transpiration-induced advection toward strongly transpiringvegetation was proposed by Mann and Wetzel [2000], whotested the concept in microcosms with partial success.Harvey et al. [1995], however, noted the opposite effect:infiltration was enhanced in, and pore water moved awayfrom, small hummocks occupied by vegetation, primarilybecause of hummock macropores. A consistent conceptualmodel remains to be developed, but our results suggest thatthe relative applicability of these two prior conclusionsdepends on the local balance of the influences of evapo-transpiration, sediment hydraulic properties, and microto-pography. Any vegetation-induced horizontal groundwaterflow in clay sediments is bound to be small and furthercomplicated by transient, oscillatory tidal influences. How-ever, we hypothesize that even a small amount of net fluidtransport across vegetation zone boundaries (e.g., betweenthe high ET Salicornia or Spartina of EHZ 3 and the lowET Distichlis of EHZ 4) could be biogeochemically andecologically significant in the long term, at the temporalscale of stable salt marsh vegetation configurations some-times spanning decades or centuries.

[65] The models revealed a second phenomenon on themarsh plain that may have important biogeochemical con-sequences: local stagnation of vertical groundwater flow inthe root zone. Stagnation was induced in low K, moderateET regions by the balance between downward drainagepotential and upward ET potential (EHZ 6; Figures 7b and7c). Stagnation was also induced along boundaries betweenvegetation zones simulated with different water uptakerates, if those boundaries were among low conductivitysediments (black lines in Figure 7b). Although such stagna-tion is likely to be a transient condition, even oscillatoryroot zone pore water flow exhibiting occasional stagnationperiods would lessen the local groundwater–surface waterexchange magnitude and might strongly influence rootzone oxygen and nutrient balances.

6.3.3. Marsh-Estuary Exchange[66] The potential significance of ecohydrological zona-

tion in terms of marsh-estuary exchange is based on thepremise that marsh groundwater is biogeochemically dis-tinct from coastal surface waters. Salt marshes have longbeen conceptualized as net suppliers of dissolved and

particulate constituents to coastal waters, with perhaps upto half of the supply from groundwater seepage [Jordanand Correll, 1985; Childers et al., 2000]. In our complexmodel simulations, which included sediment and evapo-transpiration heterogeneity, groundwater dischargeoccurred mainly via the tidal channel network, not via themarsh-estuary shoreline. However, this phenomenon of dis-charge focused in tidal channels may not apply to all set-tings: some coastlines include much greater inlandgroundwater heads than at our field site, or more hydrauli-cally conductive mudflat sediments, which may permit sub-stantially more direct shoreline seepage.

7. Conclusion[67] This study combined salt marsh vegetation zonation,

interspecific differences in plant water use, 3-D variablysaturated groundwater hydrology, sediment heterogeneity,and transient tidal flooding of different magnitudes in com-plex models of the salt marsh ecohydrological system. Themodels demonstrated that superimposed patterns of differ-ential plant water use and sediment hydraulic propertiescan produce a surprisingly complex mosaic of salt marshroot zone hydraulic environments, despite the largely ho-mogeneous influence of tidal flooding on a flat marsh plain.These substantially different root zone hydraulic environ-ments, or ecohydrological zones, are caused by spatial dif-ferences in plant water use, sediment hydraulic properties,and possibly by additional system attributes not examinedin this study. On the basis of the ecohydrological zonationobserved in our models and field data, we highlight threeconclusions.

[68] 1. The well-studied tidal channel bank environmentis hydrologically distinct from the less well understoodmarsh interior. Even so, groundwater dynamics in channelbanks may be overemphasized in models not accountingfor sediment heterogeneity.

[69] 2. The marsh interior is not a large homogeneouszone of predominantly stagnant groundwater : it may beconceptually divided into distinct ecohydrological zones,each with particular ecohydrological characteristics. Thiszonation also creates potential for lateral, vegetation-induced exchange among zones.

[70] 3. Tidal flooding temporarily masks the ecohydro-logical zones’ mosaic of diverse root zone hydraulic habi-tats, but the zones, established during low and neap tides,may serve to define salt marsh ecosystem organization atleast as much as variations in high tides’ influence.

[71] The central theme of this study recognized that,although the tides, channels, and vegetation are the mostvisually prominent features of coastal salt marshes, the inter-actions between tides and vegetation occur most directlythrough the roots and root zone. The root zone sedimentsinfluence tidal infiltration and root water uptake and governthe mass transfer between these processes [Chapman,1938b]. This perspective, focused on the root zone and itsabiotic and biotic hydraulic properties, partially reconcilesexisting conceptual models of the salt marsh groundwatersystem as vertical flow due entirely to evapotranspirationand infiltration [Hemond and Fifield, 1982]; slow downwardflow in the marsh interior feeding deep submarine ground-water discharge [Wilson and Gardner, 2006]; horizontal


17 of 22

flow restricted to the channel banks, with no interior marshflow [Harvey et al., 1987; Nuttle, 1988; Montalto et al.,2007]; or plant water uptake near channel banks controllinglocal water table position and unsaturated flow [Howes andGoehringer, 1994; Ursino et al., 2004; Wilson and Gard-ner, 2005; Li et al., 2005; Marani et al., 2006; Tosattoet al., 2009]. On the basis of the results of our detailed 3-Dsimulations, we conclude that each of these mechanisms issimultaneously significant in situ in different spatial regionsof even a small marsh site and these different spatial regionscan be usefully thought of as distinct ecohydrological zones.

Appendix A: Development of HeterogeneousSediment Model

[72] Numerical model sensitivity analysis was used toestimate the sediment structures underlying the field site.Groundwater head responses to tidal forcing were recordedin pairs of piezometers installed to depths of 0.5 m and1.0 m at 14 field locations (Figure 2). Piezometers were2 inch plastic pipe with 15 cm of slotted screen above anend cap; casings were tall enough to not be submerged byhigh tides. Clean, well-sorted sand filled the auger holearound the screen, below native clay material backfill. Piez-ometers and the ground surface were surveyed by total sta-tion. Piezometer and tidal channel water levels were loggedby pressure transducers every 10 min over portions of4 years (Dataflow Systems Pty Ltd, Christchurch, NewZealand). The principle data used in this study were from atypical spring tide day, 23–24 December 2007, chosen toprovide 24 h of clean data at as many logged piezometersas possible.

[73] Simulated porous medium properties were initiallybased on literature values and laboratory measurementsand were then refined on the basis of field measurementsand sensitivity analysis. Falling-head laboratory tests oftwo clay cores from the field site yielded similar results :K ¼ 7 � 10�4 m d�1 [Moffett et al., 2008]. Sensitivity anal-ysis was conducted using multiple values of K, specificstorage (Ss), porosity (’), and layering, with parameters inranges appropriate for the clay sediments of the site. Obser-vation points simulated in the model corresponded to thelocations of the 28 field piezometers.

[74] The simulated response of groundwater heads to aflooding tide was found to be sensitive to Ss and to the ver-tical hydraulic conductivity (Kz) of the sediments. Theresponse was insensitive to layering, to the horizontal hy-draulic conductivity (Kh), and to ’. Example results areprovided in Figures A1a–A1c and additional results inchapter 6 by Moffett [2010]. Model sensitivities to Ss andKz were comparable (Figures A1a and A1b). Since theplausible values of K for clays span a larger relative rangethan Ss for clays (4 orders of magnitude compared to 2[Smith and Wheatcraft, 1993]), we adjusted values of Kz,rather than Ss, obtaining a scenario similar to field observa-tions by narrowing the constraints relative to the range ofplausible values.

[75] A conceptual model of a heterogeneous, layered Kfield (Figure 4a) was developed: (1) The best fit Kz valuefor each of the 28 observation points was identified andassigned locally as an isotropic K value (since simulationswere insensitive to Kh). (2) A three-layer system was

constructed from these 28 scattered K values and isdescribed as follows.

[76] 1. Shallow sediments, down to about 0.5 depth (to0.5 m above mean sea level, MSL), were inferred from the0.5 m deep piezometer data. Shallow piezometers with lowK values generally fell within a high-elevation area of thecentral site (Figure 2), which was used to guide the demar-cation of a near-surface low K region (Figure 4a).

[77] 2. Deeper sediments, about 0.5 to 1 m depth (0.5 to0 m above MSL), were inferred from the 1.0 m deep pie-zometer data. Deep piezometers with low K values sur-rounded the tidal channel bisecting the field site. It is likelythat this bisecting channel is the remnant of a previouslymuch larger tidal channel visible in geographic surveys andaerial imagery of the site from 1857, 1897, and 1921[Hermstad et al., 2009], but now mostly filled with estua-rine sediment (Figure 4b). The channel arc was used as aguide in delineating the shape of a inferred low K regionfrom about 0.5 to 1 m depth.

[78] 3. The bottom layer of sediments, to the base of themodel, was assigned a low K value (7 � 10�4 m d�1) onthe basis of a conceptual model of increasing sedimentcompaction and decreasing numbers of burrow and rootmacropores with depth. The levees surrounding two sidesof the central field site in the model were also assigned thislow K value to account for sediment compaction duringconstruction.

[79] Simulations using this heterogeneous sediment con-ceptual model (Figure 4a) successfully reproduced theobserved responses of most of the 0.5 m deep piezometersto tidal forcing (Figure A1d). To not overspecify the heter-ogeneity of the model to fit the field data, K values in a fewsmall areas were maintained as similar to their surround-ings, resulting in some small model data discrepancies,e.g., at location XS (Figure A1d).

Appendix B: Development of HeterogeneousEvapotranspiration Model

[80] To develop a conceptual model of zonally distributedevapotranspiration rates and rooting depths, we began witha map of the field site vegetation from a prior study[Moffett et al., 2010]. The relative magnitudes of evapo-transpiration (ET) of the dominant species under the condi-tions prevailing at the field site were obtained in a four-stepthermal remote sensing–based procedure [Moffett andGorelick, 2012]. (1) A diurnal series of time-lapse thermalimages of vegetation surface temperature was collectedover a patch of each dominant species (Spartina foliosa,Salicornia virginica, and Distichlis spicata). (2) Transpira-tion was estimated by applying the canopy-level model ofJarvis and McNaughton [1986] to the canopy temperaturedata, also using measurements of net radiation, ground heatflux, and meteorological conditions collected on the imag-ing days at the field site; only a portion of the net radiationwas partitioned to the canopy, according to a canopy radia-tion extinction factor. (3) Soil evaporation was estimatedusing the Priestley and Taylor [1972] model based on theremainder of the net radiation. (4) An average of the com-bined daytime evaporation and transpiration rates was cal-culated from each species’ simulated diurnal ET timeseries.


18 of 22

[81] The relative magnitudes of the average daytime ETrates of Spartina foliosa : Salicornia virginica : Distichlisspicata calculated using this method were 1.94 : 1.24 : 1.Since these three species accounted for most of the sitevegetation in roughly equal proportions, their relative ETmagnitudes were scaled to produce an average rate of4 mm d�1, to be comparable to our uniform ET scenario andto prior studies [e.g., Ursino et al., 2004] (see also Daceyand Howes [1984] as discussed by Li et al., [2005]). Theresulting ET rates were 5.59, 3.57, and 2.88 mm d�1 forSpartina foliosa, Salicornia virginica, and Distichlis spi-cata, respectively. These values were comparable to valuesreported by, or which we estimated from, existing labora-tory and field studies. For Spartina spp., expected ET ratesrange from about 2 to 10 mm d�1 [Giurgevich and Dunn,1982; Bradley and Morris, 1991; Howes and Goehringer,

1994; Maricle et al., 2007]. Rates for Salicornia and simi-lar Sarcocornia include 3.2 mm d�1 [Hughes et al., 2001]and 4.57 mm d�1 [Antlfinger and Dunn, 1979]. Ratesfor Distichlis spicata include 2.4 [Maricle et al., 2007],2.9 mm d�1 [Snyder et al. 2003], and 2.58–4.38 mm d�1

[Groeneveld and Warren, 1992].[82] Most of the field site was dominated by Spartina

foliosa, Salicornia virginica, or Distichlis spicata aloneand these vegetation zones were assigned their respectiveET rates. Some areas apparently codominated by Salicor-nia and Distichlis were assigned the average of the tworates; small adjacent zones dominated by Frankenia salinaor Jaumea carnosa were also included in this generic class.Other small, fragmented zones were amalgamated intolarger surrounding zones. The levees around the site werelargely covered in Distichlis growing in dry soils,

Figure A1. Sensitivity of simulated total groundwater head during a flooding tide to multiple values ofsediment: (a) specific storage Ss, (b) vertical saturated hydraulic conductivity Kz, and (c) horizontal satu-rated hydraulic conductivity Kh. (d) Results of simulation using heterogeneous hydraulic conductivityfield depicted in Figure 4a. In each graph, the gray line shows the tidal signal measured by the tide gaugeand used to drive the model; the black line shows the total head data measured in situ at the field siteduring this tide. Colored lines show simulated piezometer responses to the tide, given different materialproperties. The example piezometer locations (L, M, Q, and X) are labeled in Figure 4a. Tidal stage andtotal head (y axis) are in units of meters above mean sea level. Each x axis spans 12 h.


19 of 22

approximated as having half the marsh Distichlis ET rate.Remaining portions of the model external to the centralmarsh site (adjacent marsh, mudflat) were assigned thegeneric rate 4 mm d�1 (Figure 5). The ET rates were simu-lated as constant flux values.

[83] Published data on the depth distribution of salt marshevapotranspiration in the root zone are sparse. Many stud-ies have assumed rooting depths of 20 or 30 cm on the basisof data from Spartina alterniflora marshes [Howes et al.,1981; Ellison et al., 1986; Steinke et al., 1996; Maraniet al., 2006; Darby and Turner, 2008]. We used species-specific rooting depth estimates from the literature and dis-tributed the simulated ET demand linearly over thesedepths in our models: 10 cm for Salicornia virginica[Chapman, 1938a; Mahall and Park, 1976a; Justin andArmstrong, 1987; Seliskar, 1983], 20 cm for Spartinafoliosa [Mahall and Park, 1976b; Boyer et al., 2000], and30 cm for Distichlis spicata [Seliskar, 1983]. The linearweighting approximated the decline in root biomass withdepth observed in salt marshes, to an extinction depth[Chapman, 1938a; Howes et al., 1981; Ellison et al.,1986; Steinke et al., 1996; Boyer et al., 2000], and thegreater effect of soil evaporation at the surface. As with theET rate assignments, we used the average value for Salicor-nia and Distichlis in their mixed zone (20 cm), the Disti-chlis value on the levees (30 cm), and the generic averagevalue in the remaining marsh areas outside the central fieldsite (20 cm) (Figure 5). The top portion of the model, dis-cretized at 0, 10, 20, 30, 50, and 100 cm depths, easilyaccommodated these ET distribution depths.

[84] Acknowledgments. This work was supported by National Sci-ence Foundation grant EAR-0634709 to Stanford University. Any opin-ions, findings, and conclusions or recommendations expressed in thismaterial are those of the authors and do not necessarily reflect the views ofthe National Science Foundation. We thank the City of Palo Alto BaylandsNature Preserve for permission to conduct the field studies and numerouscolleagues for assistance with field installation, surveying, and monitoring.We thank the San Francisco Estuary Institute for sharing with us the histor-ical coastlines, aerial photography, and lidar data.

ReferencesAgosta, K. (1985), The effect of tidally induced changes in the creekbank

water table on pore water chemistry, Estuarine Coastal Shelf Sci., 21,389–400.

Antlfinger, A. E., and E. L. Dunn (1979), Seasonal patterns of CO2 and watervapor exchange of three salt marsh succulents, Oecologia, 43, 249–260.

Bertness, M. D., L. Gough, and S. W. Shumway (1992), Salt tolerances andthe distribution of fugitive salt marsh plants, Ecology, 73, 1842–1851.

Boyer, K. E., J. C. Callaway, and J. B. Zedler (2000), Evaluating the pro-gress of restored cordgrass (Spartina foliosa) marshes: Belowgroundbiomass and tissue nitrogen, Estuaries, 23, 711–721.

Bradley, P. M. and J. T. Morris (1990), Physical characteristics of saltmarsh sediments: ecological implications, Marine Ecology ProgressSeries, 61, 245–252.

Bradley, P. M., and J. T. Morris (1991), Relative importance of ion exclu-sions, secretion and accumulation, in Spartina alterniflora Loisel.,J. Exp. Bot., 42, 1525–1532.

Brinson, M. M. (1993), A hydrogeomorphic classification for wetlands,Tech. Rep. WRP-DE-4, U.S. Army Corps of Eng. Waterw. Exp. Stn.,Vicksburg, Miss.

Carter, E. S., S. M. White, and A. M. Wilson (2008), Variation in ground-water salinity in a tidal salt marsh basin, North Inlet Estuary, South Caro-lina, Estuarine Coastal Shelf Sci., 76(3), 543–552, doi:10.1016/j.ecss.2007.07.049.

Chapman, V. J. (1938a), Studies in salt-marsh ecology: Sections I to III,J. Ecol., 26, 144–179.

Chapman, V. J. (1938b), Studies in salt-marsh ecology: Sections IV and V,J. Ecol., 27, 160–201.

Childers, D. L., J. W. Day Jr., and H. N. McKellar Jr. (2000), Twenty moreyears of marsh and estuarine flux studies: Revisiting Nixon (1980), inConcepts and Controversies in Tidal Marsh Ecology, edited by M. P.Weinstein and D. A. Kreeger, pp. 391–423, Kluwer Acad., Dordrecht,Netherlands.

Cola, S., L. Sanavia, P. Simonini, and B. A. Schrefler (2008), Coupled ther-mohydromechanical analysis of Venice lagoon salt marshes, WaterResour. Res., 44, W00C05, doi:10.1029/2007WR006570.

Cooper, A. (1982), The effects of salinity and waterlogging on the growthand cation uptake of salt marsh plants, New Phytol., 90, 263–275.

Dacey, J. W. H., and B. L. Howes (1984), Water uptake by roots controlswater table movement and sediment oxidation in short Spartina marsh,Science, 224, 487–489.

Darby, F. A., and R. E. Turner (2008), Below- and aboveground Spartinaalterniflora production in a Louisiana salt marsh, Estuaries Coasts, 31,223–231, doi:10.1007/s12237-007-9014-7.

Ebel, B. A., B. B. Mirus, C. S. Heppner, J. E. VanderKwaak, and K. L.Loague (2009), First-order exchange coefficient coupling for simulatingsurface water-groundwater interactions: Parameter sensitivity and con-sistency with a physics-based approach, Hydrol. Processes, 23, 1949–1959, doi:10.1002/hyp.7279.

Ellison, A. M., M. D. Bertness, and T. Miller (1986), Seasonal patterns inthe belowground biomass of Spartina alterniflora (Gramineae) across atidal gradient, Am. J. Bot., 73, 1548–1554.

Emery, N. C., P. J. Ewanchuk, and M. D. Bertness (2001), Competition andsalt-marsh plant zonation: Stress tolerators may be dominant competi-tors, Ecology, 82, 2471–2485.

Fagherazzi, S., A. Bortoluzzi, W. E. Dietrich, A. Adami, S. Lanzoni,M. Marani, and A. Rinaldo (1999), Tidal networks: 1. Automaticnetwork extraction and preliminary scaling features from digital terrainmaps, Water Resour. Res., 35, 3891–3904, doi :10.1029/1999WR900236.

Forbes, M. G., and K. H. Dunton (2006), Response of a subtropical estua-rine marsh to local climatic change in the southwestern Gulf of Mexico,Estuaries Coasts, 29, 1242–1254.

Gardner, L. R. (2005a), A modeling study of the dynamics of pore waterseepage from intertidal marsh sediments, Estuarine Coastal Shelf Sci.,62, 691–695, doi:10.1016/j.ecss.2004.10.005.

Gardner, L. R. (2005b), Role of geomorphic and hydraulic parameters ingoverning pore water seepage from salt marsh sediments, Water Resour.Res., 41, W07010, doi:10.1029/2004WR003671.

Gardner, L. R. (2007), Role of stratigraphy in governing pore water seepagefrom salt marsh sediments, Water Resour. Res., 43, W07502, doi:10.1029/2006WR005338.

Gardner, L. R., and A. M. Wilson (2006), Comparison of four numericalmodels for simulating seepage from salt marsh sediments, EstuarineCoastal Shelf Sci., 69, 427–437, doi:10.1016/j.ecss.2006.05.009.

Giurgevich, J. R., and E. L. Dunn (1982), Seasonal patterns of daily netphotosynthesis, transpiration and net primary productivity of Juncusroemerianus and Spartina alterniflora in a Georgia salt marsh, Oecolo-gia, 52, 404–410.

Gray, A. J., and R. Scott (1977), The ecology of Morecambe Bay: VII. Thedistribution of Puccinellia maritima, Festuca rubra, and Agrostis stoloni-fera in the salt marshes, J. Appl. Ecol., 14, 229–241.

Groeneveld, D. P., and D. C. Warren (1992), Total transpiration from landareas estimated from hand-held porometer measurements, Chapter E ofEvapotranspiration measurements of native vegetation, Owens Valley,California, June 1986, edited by D. H. Wilson, R. J. Reginato, and K. J.Hollett, Water-Resources Investigations Report 91-4159, 49–59, U.S.Geol. Surv., Sacramento, Calif.

Hamlin, S. N. (1983), Injection of treated wastewater for ground-waterrecharge in the Palo Alto Baylands, California, Hydraulic and chemicalinteractions—Preliminary report, U.S. Geol. Surv. Water Resour. Invest.Rep., 82-4121, U.S. Geol. Surv., Sacramento, Calif.

Harvey, J. W., and W. E. Odum (1990), The influence of tidal marsheson upland groundwater discharge to estuaries, Biogeochemistry, 10(3),217–236.

Harvey, J. W., P. F. Germann, and W. E. Odum (1987), Geomorphologicalcontrol of subsurface hydrology in the creekbank zone of tidal marshes,Estuarine Coastal Shelf Sci., 25, 677–691.

Harvey, J. W., R. M. Chambers, and J. R. Hoelscher (1995), Preferentialflow and segregation of porewater solutes in wetland sediment, Estuaries,18(4), 568–578.


20 of 22

Hayden, B. P., M. C. F. V. Santos, G. Shao, and R. C. Kochel (1995), Geo-morphological controls on coastal vegetation at the Virginia CoastReserve, Geomorphology, 12, 283–300.

Helley, E. J., and K. R. Lajoie (1979), Flatland deposits of the San Fran-cisco Bay Region, California—Their geology and engineering proper-ties, and their importance to comprehensive planning, U.S. Geol. Surv.Prof. Pap., 943.

Hemond, H. F., and J. L. Fifield (1982), Subsurface flow in salt-marshpeat—A model and field-study, Limnol. Oceanogr., 27(1), 126–136.

Hemond, H. F., W. K. Nuttle, R. W. Burke, and K. D. Stolzenbach (1984),Surface infiltration in salt marshes: Theory, measurement, and biogeo-chemical implications, Water Resour. Res., 20, 591–600.

Hermstad, D., K. Cayce, and R. Grossinger (2009), Historical ecology oflower San Francisquito Creek, Phase 1, technical memorandum accom-panying project GIS data, Contrib. 579, Hist. Ecol. Program, San Fran-cisco Estuary Inst., Oakland, Calif. [available at http://www.sfei.org/projects/2866.]

Howes, B. L., and D. D. Goehringer (1994), Porewater drainage and dis-solved organic carbon and nutrient losses through the intertidal creek-banks of a New England salt marsh, Mar. Ecol. Prog. Ser., 114, 289–301.

Howes, B. L., R. W. Howarth, J. M. Teal, and I. Valiela (1981), Oxidation-reduction potentials in a salt marsh: Spatial patterns and interactionswith primary productivity, Limnol. Oceanogr., 26, 350–360.

Hughes, C. E., P. Binning, and G. R. Willgoose (1998), Characterization ofthe hydrology of an estuarine wetland, J. Hydrol., 211, 34–49.

Hughes, C. E., J. D. Kalma, P. Binning, G. R. Willgoose, and M. Vertzonis(2001), Estimating evapotranspiration for a temperate salt marsh, New-castle, Australia, Hydrol. Processes, 15, 957–975, doi:10.1002/hyp.189.

Jarvis, P. G., and K. G. McNaughton (1986), Stomatal control of transpira-tion: Scaling up from leaf to region, Adv. Ecol. Res., 15, 1–49.

Johnson, D. S., and H. H. York (1915), The relation of plants to tide-levels,Carnegie Inst. Washington Publ., 206.

Jordan, T. E., and D. L. Correll (1985), Nutrient chemistry and hydrologyof interstitial water in brackish tidal marshes of Chesapeake Bay,Estuarine Coastal Shelf Sci., 21, 45–55.

Justin, S. H. F. W., and W. Armstrong (1987), The anatomical characteristicsof roots and plant response to soil flooding, New Phytol., 106, 465–495.

Li, H. L., L. Li, and D. Lockington (2005), Aeration for plant root respira-tion in a tidal marsh, Water Resour. Res., 41, W06023, doi:10.1029/2004WR003759.

Loague, K., and J. E. VanderKwaak (2002), Simulating hydrologicalresponse for the R-5 catchment: Comparison of two models and theimpact of the roads, Hydrol. Processes, 16, 1015–1032.

Mahall, B. E., and R. B. Park (1976a), The ecotone between Spartinafoliosa Trin. and Salicornia virginica L. in salt marshes of northern SanFrancisco Bay: II. Soil water and salinity, J. Ecol., 64, 793–809.

Mahall, B. E., and R. B. Park (1976b), The ecotone between Spartinafoliosa Trin. and Salicornia virginica L. in salt marshes of northern SanFrancisco Bay: I. Biomass and production, J. Ecol., 64, 421–433.

Mahall, B. E., and R. B. Park (1976c), The ecotone between Spartinafoliosa Trin. and Salicornia virginica L. in salt marshes of northernSan Francisco Bay: III. Soil aeration and tidal immersion, J. Ecol., 64,811–819.

Mann, C. J., and R. G. Wetzel (2000), Hydrology of an impounded loticwetland—Subsurface hydrology, Wetlands, 20, 33–47.

Marani, M., S. Lanzoni, D. Zandolin, G. Seminara, and A. Rinaldo (2002),Tidal meanders, Water Resour. Res., 38(2), 1225, doi:10.1029/2001WR000404.

Marani, M., E. Belluco, A. D’Alpaos, A. Defina, S. Lanzoni, and A.Rinaldo (2003), On the drainage density of tidal networks, Water Resour.Res., 39(2), 1040, doi:10.1029/2001WR001051.

Marani, M., S. Silvestri, E. Belluco, N. Ursino, A. Comerlati, O. Tosatto,and M. Putti (2006), Spatial organization and ecohydrological interac-tions in oxygen-limited vegetation ecosystems, Water Resour. Res., 42,W06D06, doi:10.1029/2005WR004582.

Maricle, B. R., D. R. Cobos, and C. S. Campbell (2007), Biophysical andmorphological leaf adaptations to drought and salinity in salt marshgrasses, Environ. Exp. Bot., 60, 458–467, doi:10.1016/j.envexpbot.2007.01.001.

Mirus, B. B., B. A. Ebel, C. S. Heppner, and K. Loague (2011), Assessingthe detail needed to capture rainfall-runoff dynamics with physics-basedhydrologic response simulation, Water Resour. Res., 47, W00H10,doi:10.1029/2010WR009906.

Moffett, K. B. (2010), Groundwater-vegetation-atmosphere interactions inan intertidal salt marsh, PhD thesis, Dep. of Environ. Earth Syst. Sci.,

Stanford Univ., Stanford, Calif. [Available at http://purl.stanford.edu/nd883hd4098.]

Moffett, K. B., and S. M. Gorelick (2012), A method to calculate heteroge-neous evapotranspiration using submeter thermal infrared imagerycoupled to a stomatal resistance submodel, Water Resour. Res., 48,W01545, doi:10.1029/2011WR010407.

Moffett, K. B., S. W. Tyler, T. Torgersen, M. Menon, J. S. Selker, andS. M. Gorelick (2008), Processes controlling the thermal regime of salt-marsh channel beds, Environ. Sci. Technol., 42, 671–676, doi:10.1021/es071309m.

Moffett, K. B., D. A. Robinson, and S. M. Gorelick (2010), Relationship ofsalt marsh vegetation zonation to spatial patterns in soil moisture, salin-ity, and topography, Ecosystems, 13, 1287–1302, doi:10.1007/s10021-010-9385-7.

Montalto, F. A., T. S. Steenhuis, and J.-Y. Parlange (2006), The hydrologyof Piermont Marsh, a reference for tidal marsh restoration in the Hudsonriver estuary, New York, J. Hydrol., 316, 108–128, doi:10.1016/j.jhydrol.2005.03.043.

Montalto, F. A., J.-Y. Parlange, and T. S. Steenhuis (2007), A simple modelfor predicting water table fluctuations in a tidal marsh, Water Resour.Res., 43, W03439, doi:10.1029/2004WR003913.

Morris, J. T., and G. J. Whiting (1985), Gas advection in sediments of aSouth Carolina salt marsh, Mar. Ecol. Prog. Ser., 27, 187–194.

Morzaria-Luna, H., J. C. Callaway, G. Sullivan, and J. B. Zedler (2004),Relationship between topographic heterogeneity and vegetation patternsin a Californian salt marsh, J. Veg. Sci., 14, 523–530.

National Geospatial-Intelligence Agency (2004), USGS High ResolutionOrthoimage USNG 10SEG775460, San Francisco-Oakland, CA, providedto National Geospatial-Intelligence Agency (NGA) and U.S. GeologicalSurvey (USGS) by EarthData International of Maryland, LLC. [availableat http://atlas.ca.gov/casil/imageryBaseMapsLandCover/imagery/urban_imagery/metadata/SFBAY_Del1/Metadata/10SEG775460.xml.]

Nuttle, W. K. (1988), The extent of lateral water movement in the sedimentsof a New England salt marsh, Water Resour. Res., 24(12), 2077–2085.

Nuttle, W. K., and J. W. Harvey (1995), Fluxes of water and solute in acoastal wetland sediment. 1. The contribution of regional groundwaterdischarge, J. Hydrol., 164, 89–107.

Odum, E. P. ( 1980), The status of 3 ecosystem level hypotheses regardingsalt marsh estuaries tidal subsidy outwelling and detritus based foodchains, in Estuarine Perspectives, edited by V. Kennedy, pp. 485–495,Academic, New York.

Osgood, D. T., and J. C. Zieman (1998), The influence of subsurface hy-drology on nutrient supply and smooth cordgrass (Spartina alterniflora)production in a developing barrier island marsh, Estuaries, 21, 767–783.

Pennings, S. C., and R. M. Callaway (1992), Salt marsh plant zonation:The relative importance of competition and physical factors, Ecology,73, 681–690.

Pennings, S. C., M.-B. Grant, and M. D. Bertness (2005), Plant zonation inlow-latitude salt marshes: Disentangling the roles of flooding, salinityand competition, J. Ecol., 93, 159–167.

Priestley, C. H. B., and R. J. Taylor (1972), On the assessment of surfaceheat flux and evaporation using large-scale parameters, Mon. WeatherRev., 100, 81–92.

Rodriguez-Iturbe, I., P. D’Odorico, F. Laio, L. Ridolfi, and S. Tamea(2007), Challenges in human land ecohydrology: Interactions of watertable and unsaturated zone with climate, soil, and vegetation, WaterResour. Res., 43, W09301, doi:10.1029/2007WR006073.

Schaap, M. G., and F. L. Leij (2000), Improved prediction of unsaturatedhydraulic conductivity with the Mualem–van Genuchten model, Soil Sci.Soc. Am. J., 64, 843–851.

Seliskar, D. M. (1983), Root and rhizome distribution as an indicator ofupper salt marsh wetland limits, Hydrobiologia, 107, 231–236.

Silvestri, S., A. Defina, and M. Marani (2005), Tidal regime, salinity andsalt marsh plant zonation, Estuarine Coastal Shelf Sci., 62, 119–130,doi:10.1016/j.ecss.2004.08.010.

Smith, L., and S. W. Wheatcraft (1993), Groundwater flow, in Handbook ofHydrology, edited by D. R. Maidment, pp. 6.1–6.58, McGraw-Hill,New York.

Snyder, K. A., J. H. Richards, and L. A. Donovan (2003), Night-time con-ductance in C3 and C4 species: Do plants lose water at night?, J. Exp.Bot., 54, 861–865, doi:10.1093/jxb/erg082.

Steinke, W., D. J. von Willert, and F. A. Austenfeld (1996), Root dynamicsin a salt marsh over three consecutive years, Plant Soil, 185, 265–269.

Teal, J. M., and J. W. Kanwisher (1970), Total energy balance in salt marshgrasses, Ecology, 51, 690–695.


21 of 22

Temmerman, S., T. J. Bouma, G. Govers, Z. B. Wang, M. B. De Vries, andP. M. J. Herman (2005), Impact of vegetation on flow routing and sedi-mentation patterns: Three-dimensional modeling for a tidal marsh,J. Geophys. Res., 110, F04019, doi:10.1029/2005JF000301.

TerraPointUSA (2005), Project Report, USGS South Bay Restoration Pro-ject, Contract #2206-H, in South San Francisco Bay 2004 TopographicLidar Survey: data overview and preliminary quality assessment, byA. C. Foxgrover and B. E. Jaffe, Open-File Report 2005-1284, pp. 26,U.S. Geol. Surv., Santa Cruz, Calif.

Tessier, M., J.-C. Gloaguen, and V. Bouchard (2002), The role of spatio-temporal heterogeneity in the establishment and maintenance of Suaedamaritima in salt marshes, J. Veg. Sci., 13, 115–122.

Therrien, R., and E. Sudicky (1996), Three-dimensional analysis of varia-bly-saturated flow and solute transport in discretely-fractured porousmedia, J. Contam. Hydrol., 23, 1–44.

Therrien, R., R. G. McLaren, E. A. Sudicky, and S. Panday (2008), User’sguide, HydroGeoSphere: A three-dimensional numerical model describ-ing fully-integrated subsurface and overland flow and solute transport,Groundwater Simul. Group, Waterloo, Ontario, Canada.

Tobias, C. R., J. W. Harvey, and I. C. Anderson (2001), Quantifyinggroundwater discharge through fringing wetlands to estuaries: Seasonalvariability, methods comparison, and implications for wetland-estuaryexchange, Limnol. Oceanogr., 46(3), 604–615.

Tosatto, O., E. Belluco, S. Silvestri, N. Ursino, A. Comerlati, M. Putti, andM. Marani (2009), Reply to comment by L. R. Gardner on ‘‘Spatial orga-nization and ecohydrological interactions in oxygen-limited vegetationecosystems,’’ Water Resour. Res., 45, W05604, doi:10.1029/2007WR006345.

Tsihrintzis, V., and E. E. Madiedo (2000), Hydraulic resistance determina-tion in marsh wetlands, Water Resour. Manage., 14, 285–309.

Ursino, N., S. Silvestri, and M. Marani (2004), Subsurface flow and vegeta-tion patterns in tidal environments, Water Resour. Res., 40, W05115,doi:10.1029/2003WR002702.

van Genuchten, M. T. (1980), A closed-form equation for predicting thehydraulic conductivity of unsaturated soils, Soil Sci. Soc. Am. J., 44,892–898.

van Proosdij, D., R. G. D. Davidson-Arnott, and J. Ollerhead (2006), Con-trols on spatial patterns of sediment deposition across a macro-tidal saltmarsh surface over single tidal cycles, Estuarine Coastal Shelf Sci., 69,64–86, doi:10.1016/j.ecss.2006.04.022.

Varty, A. K., and J. B. Zedler (2008), How waterlogged microsites help anannual plant persist among salt marsh perennials, Estuaries Coasts, 31,300–312, doi:10.1007/s12237-007-9019-2.

Wilson, A. M., and L. R. Gardner (2005), Comment on ‘‘Subsurfaceflow and vegetation patterns in tidal environments’’ by Nadia Ursino,Sonia Silvestri, and Marco Marani, Water Resour. Res., 41, W07021,doi:10.1029/2004WR003554.

Wilson, A. M., and L. R. Gardner (2006), Tidally driven groundwater flowand solute exchange in a marsh: Numerical simulations, Water Resour.Res., 42, W01405, doi:10.1029/2005WR004302.

Winter, T. C., J. W. Harvey, O. L. Franke, and W. M. Alley (1998),Groundwater and surface water: A single resource, U.S. Geol. Surv.Circ., 1139.

Xin, P., G. Jin, L. Li, and D. A. Barry (2009), Effects of crab burrows onporewater flows in salt marshes, Adv. Water Res., 32, 439–449,doi:10.1016/j.advwatres.2008.12.008.

Xin, P., L.-R. Yuan, L. Li, and D. A. Barry (2011), Tidally driven multi-scale pore water flow in a creek-marsh system, Water Resour. Res., 47,W07534, doi:10.1029/2010WR010110.

S. M. Gorelick and K. Moffett, Department of Environmental EarthSystem Science, Stanford University, MC 4216, 473 Via Ortega, Rm. 140,Stanford, CA 94305, USA. ([email protected])

R. G. McLaren and E. A. Sudicky, Department of Earth and Environ-mental Sciences, University of Waterloo, Waterloo, ON N2L 3G1, Canada.


22 of 22

Documents

Salt marsh ecohydrological zonation due to heterogeneous ...accounting for intertidal inﬁltration, reaction, transport, discharge, and mixing of salt marsh groundwater and tidal