ArticlePDF Available

Sea-Level Rise Impact on Salt Marsh Sustainability and Migration for a Subtropical Estuary: GTMNERR (Guana Tolomato Matanzas National Estuarine Research Reserve)

Authors:

Abstract and Figures

Sea-level rise impacts on salt marsh for Guana Tolomato Matanzas National Estuarine Research Reserve are investigated using field measurements (six sites within the marsh) and a tide-marsh equilibrium model (Hydro-MEM). The hydrodynamic component of the model enables for prediction of spatially variable tidal data (mean low water and mean high water), which are coupled with a marsh equilibrium model (MEM) for prediction of spatially based biomass productivity of Spartina alterniflora. The field measurements corroborate the model results by way of prediction of relatively productive marsh at four of the six sites (percent coverage of Spartina alterniflora of 30–41%, canopy height of 0.27–0.67 m and simulated biomass density of greater than 750 g m⁻² over at least half of the local area within 500-m radii of the measurement sites) and relatively limited marsh at the two other sites (percent coverage of Spartina alterniflora of 3–5%, canopy height of 0.11–0.29 m, and simulated biomass density of greater than 750 g m⁻² over less than one-tenth of the local area within 500-m radii of the measurement sites). The model is applied in a coupled fashion for 50 years of time into the future using ten 5-year increments, where each increment of Hydro-MEM accounts for the natural accretion of the marsh, an update of the digital elevation model and bottom-friction parameterization, and the subsequent feedback to the hydroperiod and marsh productivity. Hydro-MEM is shown to exhibit rate-sensitivity with respect to sea-level rise exceeding the marsh accretion rate, whereby a sudden loss of marsh elevation occurs in such instances of marsh destabilization. Demonstrating rate-critical transition, the model proves flexible to account for the non-homogeneous and transient nature of the fast-slow variables, whereby the marsh migrates away from the tidal creeks and further into the upland zones. Practical implication of the model results is illustrated by identifying zones of lands into which marsh will be able to migrate and where existing marsh will not survive under increasing sea level. Post-analysis compares the final model output against land use/cover zonation to correct the 50-year simulation results of marsh productivity for elevation-appropriate regions but that are developed, freshwater, or otherwise inappropriate land type for salt-marsh habitat.
This content is subject to copyright. Terms and conditions apply.
Sea-Level Rise Impact on Salt Marsh Sustainability and Migration
for a Subtropical Estuary: GTMNERR (Guana Tolomato Matanzas
National Estuarine Research Reserve)
Peter Bacopoulos
1
&Amanda S. Tritinger
2
&Nicole G. Dix
3
Received: 11 January 2017 /Accepted: 22 June 2018
#Springer International Publishing AG, part of Springer Nature 2018
Abstract
Sea-level rise impacts on salt marsh for Guana Tolomato Matanzas National Estuarine Research Reserve are investigated using
field measurements (six sites within the marsh) and a tide-marsh equilibrium model (Hydro-MEM). The hydrodynamic com-
ponent of the model enables for prediction of spatially variable tidal data (mean low water and mean high water), which are
coupled with a marsh equilibrium model (MEM) for prediction of spatially based biomass productivity of Spartina alterniflora.
The field measurements corroborate the model results by way of prediction of relatively productive marsh at four of the six sites
(percent coverage of Spartina alterniflora of 3041%, canopy height of 0.270.67 m and simulated biomass density of greater
than 750 g m
2
over at least half of the local area within 500-m radii of the measurement sites) and relatively limited marsh at the
two other sites (percent coverage of Spartina alterniflora of 35%, canopy height of 0.110.29 m, and simulated biomass density
of greater than 750 g m
2
over less than one-tenth of the local area within 500-m radii of the measurement sites). The model is
applied in a coupled fashion for 50 years of time into the future using ten 5-year increments, where each increment of Hydro-
MEM accounts for the natural accretion of the marsh, an update of the digital elevation model and bottom-friction parameter-
ization, and the subsequent feedback to the hydroperiod and marsh productivity. Hydro-MEM is shown to exhibit rate-sensitivity
with respect to sea-level rise exceeding the marsh accretion rate, whereby a sudden loss of marsh elevation occurs in such
instances of marsh destabilization. Demonstrating rate-critical transition, the model proves flexible to account for the non-
homogeneous and transient nature of the fast-slow variables, whereby the marsh migrates away from the tidal creeks and further
into the upland zones. Practical implication of the model results is illustrated by identifying zones of lands into which marsh will
be able to migrate and where existing marsh will not survive under increasing sea level. Post-analysis compares the final model
output against land use/cover zonation to correct the 50-year simulation results of marsh productivity for elevation-appropriate
regions but that are developed, freshwater, or otherwise inappropriate land type for salt-marsh habitat.
Keywords Ecological models .Rate-induced critical transitions .Spartina alterniflora .Tides .Sea-level rise .Climate systems
1 Introduction
Salt marshes are coastal ecosystems in the intertidal zone that
consist of freshwater, brackish water, and seawater, the distri-
bution of which is regularly modified by tidal fluxes [1]. Salt
marshes provide for some of the most diverse and productive
ecosystems in the world [2]. Terrestrial production of salt
marshes involves above-ground plant growth and below-
ground growth and decay which stabilizes the salt marsh and
is a feedback for additional stabilization by sediment trapping
and accumulation [3]. Salt marsh grasses naturally filter nutri-
ents from estuarine waters which act as a source to the aquatic
biota of the ecosystem, including migratory waterfowl, wading
birds, fish, shellfish, phytoplankton, and bacteria. Additional
Electronic supplementary material The online version of this article
(https://doi.org/10.1007/s10666-018-9622-6) contains supplementary
material, which is available to authorized users.
*Peter Bacopoulos
1
Jacksonville, USA
2
Department of Civil and Coastal Engineering, University of Florida,
Gainesville, FL, USA
3
Guana Tolomato Matanzas National Estuarine Research Reserve,
505 Guana River Road, Ponte Vedra Beach, FL, USA
Environmental Modeling & Assessment
https://doi.org/10.1007/s10666-018-9622-6
ecosystem services provided by salt marshes include buffering
of intruding storm surge and support of economically important
commercial fisheries, recreation, and tourism, including water
sports, beachcombing, birdwatching, and fishing [4].
The greatest natural threat to salt marsh has been identified
as sea-level rise, particularly in that this may not allow marshes
to produce enough organic matter and trap enough sediment to
survive under accelerated sea-level rise [5]. Further, human
impacts can alter the natural feedbacks that promote the marsh
inkeepingupwithincreasingsealevelbychangingtheclimate,
land-use environment, nutrient input, sediment supply, and land
subsidence [3]. Landward salt marsh migration is possible
where natural buffer zones of suitable elevation are present;
however, development and hardened shorelines hinder migra-
tion. Local topography may also restrict the expanse of future
salt marsh if coastal wetlands cannot migrate up steep slopes of
coastal terraces [6].
The rate of sea-level rise has been 23 mm/year over the
past half century for Floridas east coast [7]. There is expected
acceleration of sea-level rise [8] which has been shown to al-
ready be occurring along Floridas east coast [9]. Predictive
ecological modeling tools are needed by land managers and
coastal decision makers to plan for the future of coastal envi-
ronments as populations continue to increase. One tool gaining
momentum is the tide-marsh equilibrium model, Hydro-MEM
(and hereafter referred to as such), useful for prediction of salt
marsh productivity based on tidal hydrodynamic conditions, as
affected by sea-level rise [10].
The overarching concept of the Hydro-MEM model is
the integrated simulation of tidal hydrodynamics and
marsh productivity using a loose coupling of a hydrody-
namic model with a marsh equilibrium model (MEM).
MEM was originally developed as a zero-dimensional
(i.e., site-based) formula of biomass density (Spartina
alterniflora) as a function of the marsh surface elevation,
mean low water (MLW), and mean high water (MHW) [11,
12]. The mechanisms involved in MEM include the accre-
tion of the marsh due to primary production and sediment
trapping, where MEM accounts for the feedback between
the marsh surface elevation and the productivity and accre-
tion of the marsh. MEM predicts the stable (or unstable)
equilibrium of a marsh system based on the rate of marsh
accretion relative to the rate of sea-level rise. Morris et al.
[12] showed that sea-level rise can lead to increased marsh
productivity and elevation. However, if the rate of sea-
level rise exceeds the rate of increase in marsh elevation,
the system destabilizes and a sudden loss of productivity
and elevation occurs. To that end, MEM lies within the
theoretical framework of instability-induced regime shift
and rate-induced critical transitions [13]. Specific to
MEM, this rate-sensitivity is investigated in the context
of the exceedance of a critical rate of change in sea level
relative to the response rate of the salt marsh. As such,
MEM can be classified as a rate-sensitive model [14],
where there are tipping points of marsh stability (or insta-
bility) as subject to the rate of sea-level rise.
The site-based nature of MEM is due to the in-situ appli-
cation of the physio-biological experiment that develops the
biomass curve and physical parameters of MEM [15]. MEM
was first extended for two-dimensional application by incor-
porating a hydrodynamic component that calculates tidal da-
tums (MLW and MHW) based on sea-level rise [11]. MEM
with tidal hydrodynamics (i.e., Hydro-MEM) was then refined
to include a coupling time step that discretizes the long-term
(50-year) simulation into successively applied increments (5-
or 10-year interval) [16]. The coupling time step of Hydro-
MEM captures the nonlinear effects between the mechanisms
of astronomic tides and the salt marsh (productivity, accretion
and surface elevation), which allows for a non-homogeneous
and transient calculation of equilibria. In this study, Hydro-
MEM is applied to investigate the rate-sensitivity (sea-level
rise) of MEM (biomass productivity and marsh accretion)
with an extended capability to account for upland migration
of salt marshes, as demonstrated for a real-world system
(Guana Tolomato Matanzas National Estuarine Research
ReserveGTMNERR). The Bfast^environmental variables
are the system topography, hydroperiod (MLW and MHW),
and bottom-roughness character, and the Bslow^environmen-
tal variables are the marsh response (productivity and accre-
tion) and sea-level rise. The approach accounts for these fast-
slow variables [17] and their spatial-temporal variability over
a two-dimensional landscape that ranges in time from present-
day conditions to a future dynamic steady state.
A crucial ingredient in the application of the Hydro-MEM
model is the digital elevation model (DEM) (see [18] for re-
cent study on the topic). The hydrodynamic model (shallow-
water equations) employs the DEM for simulation of tidal
hydrodynamics, including calculation of mean low water
(MLW) and mean high water (MHW). The simulated tidal
hydrodynamics (MLW and MHW) are loosely coupled with
MEM (a zero-dimensional biomass production formula of
biomass density of Spartina alterniflora versus marsh eleva-
tion and tidal hydrodynamics, a.k.a. the biomass curve; refer
to [12]) to quantify the biomass production of Spartina
alterniflora over the two-dimensional marsh landscape, as
defined by the DEM. Hydro-MEM is applicable for
projection-based simulation of sea-level rise impacts on salt
marsh productivity. Hagen et al. [11] applied the Hydro-MEM
model for sea-level rise scenarios of 0.15 and 0.30 m; howev-
er, those simulations were applied as a single time step to jump
from present-day conditions to the given scenario of sea-level
rise, as well as the fact that they considered an engineered
accretion, as opposed to the natural accretion formula
employed by Hydro-MEM. Recently, Alizad et al. [16]have
enhanced the Hydro-MEM model to include a coupling time
step such that multiple simulations can be applied in
Bacopoulos P. et al.
succession, with natural accretion accounted for, to increment
from present-day conditions to the given scenario of sea-level
rise. The coupling time step provides a more mechanistic sim-
ulation of the tidal hydrodynamics and marsh productivity by
evolution of the marsh via accretion (i.e., updating the DEM
with each coupling time step), friction parameter, and rate of
sea-level rise, which becomes important for nonlinear scenar-
ios of sea-level rise [19].
In this study, tidal hydrodynamics (MLW and MHW) and
the biomass production of Spartina alterniflora are examined
for the Guana Tolomato Matanzas National Estuarine
Research Reserve (GTMNERR), Florida, USA. The physical
and ecological processes were integrally simulated using a
loosely coupled Hydro-MEM model of the GTMNERR to
provide a data-modeling tool for predictions of tidal hydrody-
namics and marsh productivity, accretion, and migration.
Works like this study include Hagen et al. [11] and Alizad et
al. [16] for the St. Johns River, Tritinger [20]forthe
GTMNERR, and Alizad et al. [19] for the Apalachicola marsh
system, which together present the latest methodologies for
tidal hydrodynamic-marsh productivity modeling of Spartina
alterniflora. The purpose of this study is to translate the
Hydro-MEM model and related approaches to the
GTMNERR, and to introduce a new post-analysis technique
that corrects the simulation results based on land use/cover
zonation for more practical end use. The paper proceeds to
present the study site and field methods, the development of
the high-resolution unstructured mesh for the GTMNERR, a
validation of tidal hydrodynamics and marsh productivity, and
a practical demonstration of Hydro-MEM for prediction of
marsh migration throughout the reserve due to sea-level rise
(50 years into the future: 0.13, 0.22, and 0.51 m, as represen-
tative of low, intermediate, and high scenarioscf. [21]) and
the accompanying increase in inundation extent and tidal da-
tum elevations.
2 Study Site
The GTMNERR is located in St. Johns and Flagler counties
on the northeast coast of Florida, USA, and is geographically
divided into a northern component (Guana and Tolomato
River estuaries) and a southern component (Matanzas River
estuary), as separated by St. Augustine (Fig. 1a, b). The estu-
aries are connected to the Atlantic Ocean via St. Augustine
Inlet and Matanzas Inlet. Another feature of the GTMNERR is
the Atlantic Intracoastal Waterway [22], which is dredged at
depths of 3 m to facilitate navigation between the estuaries and
access to the Atlantic Ocean [23].
The GTMNERR encompasses approximately 300 km
2
of
submerged lands and uplands, including salt marsh and man-
grove tidal wetlands, oyster bars, estuarine lagoons, upland
habitat, and open waters. Coastal salt marsh and open-water
habitat constitute approximately 20% of the GTMNERR land
and watershed area. Dominant plants in the salt marsh include
saltmarsh cordgrass (Spartina alterniflora), black needlerush
(Juncus roemerianus), and saltwort (Batis maritima). Storm
events at the GTMNERR include thunderstorms, noreasters,
tropical storms, and hurricanes [24], where the frequency of
noreasters is about five times per year and the frequency of
tropical storms/hurricanes is about once every 3 years [25].
The drainage basin of the GTMNERR covers approximately
1900 km
2
of surface area.
A defined research gap of the GTMNERR is an under-
standing of coastal circulation in the nearshore and offshore
waters and the connections of these waters to the estuarine
outflows from St. Augustine and Matanzas Inlets. Another
research priority of the GTMNERR is the development of
predictive tools to guide management decisions affecting nat-
ural biodiversity in the face of climate change. Prior modeling
studies on the GTMNERR were conducted by CDM/DHI [26]
using a two-dimension, finite difference model and by Sheng
et al. [27] using a three-dimensional hydrodynamic model.
The importance of the wetting and drying of intertidal zones
was the major finding from these prior modeling studies. A
recent modeling study on the GTMNERR was conducted by
Linhoss et al. [28] using sea level affecting marshes model
(SLAMM). SLAMM is a useful tool for developing a general
understanding of the trends that a region may face under sea-
level rise [29]; though, the model has its limitations, e.g., the
fact that the broad spatial scale of SLAMM does not accom-
modate for simulation of mechanistic processes at specific
locations.
3 Field Methods
In 2012, the GTMNERR established a long-term intertidal
vegetation monitoring program with the goal of determining
impacts of large-scale environmental changes (e.g., climate
change, sea-level rise) on coastal habitat structure and func-
tion. Data were collected from the field in the fall seasons of
2012 and 2013 and spring and summer seasons of 2014, in-
cluding percent coverage based on visual site inspection and
canopy height of Spartina alterniflora. Sampling protocols
are a combination of the NERRS Biological Monitoring pro-
tocols [30] and the National Park Service Southeast Coastal
Network (SECN) protocols [31,32]. Sites were selected using
a spatially balanced random sample design developed by the
SECN [33] to facilitate salt marsh-wide inference. Six perma-
nent low-marsh sampling sites were chosen that met the se-
lection criteria (Fig. 1bf).
Within each of the six sites, three replicate stations were
established within a 70-m buffer of the original site location.
Each station consisted of five permanent 1-m
2
vegetation
plots. In each vegetation plot, maximum canopy height was
Sea-Level Rise Impact on Salt Marsh Sustainability and Migration for a Subtropical Estuary: GTMNERR (Guana...
determined by averaging the height of the five tallest individ-
uals of the dominant species.Percent cover was determined by
visual estimates (in 5% increments) in the field as well as by
the acquisition of close-to-earth nadir (downward facing) im-
ages processed with SamplePoint software [34]. A lightweight
collapsible camera stand (2 m high with a 1 × 1 m base) fab-
ricated to the specifications outlined by Booth et al. [35]and
Curtis et al. [32] was employed to acquire images in the field
using a digital single-lens reflex camera with a remote trigger.
The nadir images were cropped with Adobe Photoshop
Element 9 [36] to the inside corners of the camera stand base
for a 1-m
2
photo plot. Images were imported into SamplePoint
[34] where 100 randomly generated pixels were classified to
the species level. For species with cover less than 5%, a cover
Fig. 1 aThe large-scale finite element mesh used for model simulation
telescoping from the western North Atlantic Ocean into the GTMNERR,
Florida (FL), USA. bMap of the GTMNERR showing boundaries of
county, reserve (north and south), and watershed. The six sites where
field data were collected are numbered by id. Zoom-in maps show
aerial photography for sites c00, d01, e06, f22, g40, and h46
Bacopoulos P. et al.
of 1% was assigned in the field. If a species was missed in the
field but identified by SamplePoint, or vice versa, a cover of
1% was assigned in the laboratory.
4 Tide-Marsh Equilibrium Model (Hydro-MEM)
4.1 Overview
In this study, the Hydro-MEM model was applied that accom-
modates simulation of mechanistic processes for the entire
domain, and the modeling approach was accomplished with
dense yet flexible resolution. The Hydro-MEM model has the
added capability to describe accretion due to primary produc-
tivity and sediment trapping, where accretion is fed back into
the hydrodynamic model for successive simulation of tides
and marsh productivity to increment over a long time duration
of sea-level rise. Ultimately, the Hydro-MEM model has the
capability to generate projections of marsh productivity for
practical scenarios of sea-level rise (i.e., 50 years into the
future) to be used for strategic land-use planning of the
GTMNERR. Sea-level rise scenarios of 0.13, 0.22, and
0.51 m were applied in the model simulations, which are rep-
resentative of low, intermediate, and high global sea-level rise
scenarios [21].
4.2 Marsh Equilibrium Model
The biomass curve is a product of physio-biological experi-
ments as implemented at a specific site [12]:
B¼aD þbD2þcð1Þ
where Bis biomass density (g m
2
); a,b,andcare data-driven
coefficients of the biomass curve (g m
2
); and Dis relative
depth, which is a dimensionless value:
D¼MHWY
MHWMLW ð2Þ
where Dis relative depth (dimensionless), MLW is mean low
water (m), MHW is mean high water (m), and Yis topographic
elevation (m). In the absence of biomass data for the
GTMNERR, the coefficients of a= 1000 g m
2
,b=
3718 g m
2
,andc= 1021 g m
2
were utilized in Hydro-
MEM (i.e., the biomass curveMEM version 3.4), as repre-
sentative of marsh productivity of Spartina alterniflora for
North Inlet, South Carolina [37]. Although species distribu-
tion is spatially variable throughout the GTMNERR marshes,
including Spartina alterniflora,Batis maritima,andJuncus
roemerianus, the Hydro-MEM modeling performed in this
study yields qualitative predictions of Spartina alterniflora
productivity for the GTMNERR, as regionalized into low,
medium, and high productivity [16,18,19].
The D-squared term in the biomass curve (see Eq. 1)with
negative coefficient value of byields a downwardly concaved
parabola. The peak of the biomass curve is an optimum point
that corresponds with surface elevation in relation to tidal
hydrodynamics (MLW and MHW) yielding the greatest pro-
ductivity of Spartina alterniflora [37]. The biomass curve
predicts that the marsh will be stable against rising sea level
when the surface elevation is greater than what is optimal for
primary production, while the marsh will beunstable when the
surface elevation is less than optimal [12]. Consider a scenario
with no sea-level rise (r= 0), where the marsh is in equilibri-
um (i.e., net accretion equals net erosiondY/dt =0; Eq. 3)
when the marsh surface elevation is at mean high water (i.e.,
Y=MHWD=0;Eq. 2). If sea-level rise is gradual (r>0),
the marsh elevation starts to lag behind the equilibrium marsh
surface elevation. If sea-level rise is too rapid (r>> 0), the
marsh elevation cannot keep up and a sudden loss of vegeta-
tion and elevation occurs.
Accretion rate of the marsh is a complex function of organ-
ic material and inorganic sediment accumulation [38], both of
which are related to biomass density in the marsh [37].
Organic material accumulation results from primary produc-
tion in the marsh. Sedimentation is a function of the ability of
the marsh grasses to trap sediments. Morris et al. [12]devel-
oped an equation for the rate of total (organic and inorganic)
accretion of the marsh:
dY
dt ¼qþkBðÞDð3Þ
where dY/dt is the total accretion rate (m year
1
), qis the
sediment loading rate (m year
1
), kaccounts for the organic
and inorganic contributions resulting from primary production
and sediment trapping (m g
1
year
1
), Band Dare biomass
density (g m
2
) and relative depth (dimensionless), respective-
ly. The values of qand kare data-driven and estuary-specific
[12]. Considering the definition of relative depth D(see Eq.
2), accretion is positive when the surface elevation of the
marsh Yis below MHW, and when the surface elevation of
the marsh Yis above MHW, there is no accumulation of inor-
ganic sediments [15]. The nonlinearity of MEM is clearly
evident between Eq. 3(containing dY/dt and B) and Eqs. 1
and 2(containing Yand B).
4.3 Hydrodynamic Model
The shallow-water equation model (ADCIRC, which stands
forADvancedCIRCulation)ofLuettichetal.[39] was used
for the numerical simulation of tidal hydrodynamics.
Simulated hydrodynamic variables included water surface el-
evations (units = m) and depth-integrated velocities (units = m
s
1
). The numerical procedures employed by ADCIRC in-
clude a continuous Galerkin-based discretization for solution
Sea-Level Rise Impact on Salt Marsh Sustainability and Migration for a Subtropical Estuary: GTMNERR (Guana...
of the shallow-water equations [40] in the generalized wave
continuity formulation [41].
Tab le 1details the data sources used in mesh generation
and model building for the GTMNERR. Aerial imagery [44]
provided a basis for delineation of shorelines, islands, water-
ways, and tidal creeks for the GTMNERR. Model bathymetry
for the inshore domain, including St. Augustine Inlet,
Matanzas Inlet, and the Atlantic Intracoastal Waterway, was
sourced from John et al. [42]. Model bathymetry for the off-
shore and deep-water domain was provided by the western
North Atlantic Ocean-based model of Hagen et al. [43].
Model topography for the GTMNERR land and watershed
area was sourced from NOAA Office for Coastal
Management [45].
Meshing of the inshore domain was guided by a 50-m
element-size criterion; however, shorelines of high curvature
and narrow channels were defined using elements sized as low
as 25 m. The localized mesh developed for the GTMNERR
was incorporated into the large-scale domain of Hagen et al.
[43] to generate the large-scale finite element mesh used for
model simulation telescoping from the western North Atlantic
Ocean into the GTMNERR (Fig. 1a). The finalized mesh for
the GTMNERR including the large-scale domain contains
209,688 nodes and 408,798 elements. Figure 2ac shows
the finite element mesh for the GTMNERR, the mesh size
distribution, and the model representation of bathymetry and
topography, respectively. Figure 2di shows insets of the
model mesh and bathymetry/topography for the six sites
where field data were collected. Refer to Fig. 3for highly
zoomed-in inlets of the model mesh and elevation for the six
field-observation sites. Note the fine resolution of the mesh
along the entire length of the Atlantic Intracoastal Waterway
and the offshoot tidal creeks that penetrate into the marsh. The
mesh size in channels is 2550 m and in the marsh is 50
100 m. St. Augustine and Matanzas Inlets are resolved with
an approximate 30-m mesh size. The bathymetry of the chan-
nels generally ranges in depths of 25 m. The channel invert
depths at St. Augustine and Matanzas Inlets are 14 and 6 m,
respectively. The channel banks are generally at the 0-m ele-
vation contour. The topography of the marsh ranges in eleva-
tions of 02m.
The hydrodynamic model settings and parameters were set
as follows. Bottom friction was treated nonlinearly in the sim-
ulations with a spatially varying Manningsnvalue based on
National Land Cover Data 2011 [46]. Manningsnwas pa-
rameterized in the model as follows [19,47]: 0.025 for open
water, 0.050 for wetlands and marsh regions with biomass
density less than 750 g m
2
, 0.075 for marsh regions with
biomass density greater than 750 g m
2
, and 0.100 for upland
zones. The parameterization of Manningsnwas specified at
the beginning of the 50-year simulation and updated each 5-
year increment for pre-existing marsh that becomes sub-
merged by sea-level rise and converts to open water (i.e.,
0.050 0.025) or if the biomass density of pre-existing
marsh changes due to sea-level rise (i.e., 0.050 0.075).
Nonlinear advection and nonlinear finite amplitude effects
were enabled in the simulations. Wetting and drying of ele-
ments was enabled in the simulations [39]. Horizontal eddy
viscosity was set to 5 m
2
s
1
. Coriolis effects were enabled in
the simulations [39]. A time step of 1 s was employed, where
the simulations were run for 30 days with the first 5 days of the
simulations being a ramp of the model from a cold start (still-
water condition) to full dynamics. Boundary conditions were
applied with eight tidal constituents (M2, S2, N2, K1, O1, K2,
Q1, and P1) along the deep-water, open-ocean boundary of
the model along the 60°W meridian, as derived from the large-
scale domain of Hagen et al. [43]. Additionally, boundary
conditions were specified for the two ends of the Atlantic
Intracoastal Waterway that extend approximately 20 km be-
yond the northern and southern boundaries of the GTMNERR
(Fig. 2a). These boundary conditions included the same eight
tidal constituents listed above and were derived from a highly
validated tidal model of the South Atlantic Bight estuaries and
intertidal zones [48]. Tidal potentials were applied over the
interior of the model domain. Harmonic analysis was applied
to extract 23 tidal constituents for all mesh nodes.
Table 1 The data sources used in mesh generation and model building for the GTMNERR
Type Source Detail Date access Web access
Aerial imagery GeoCover Landsat UTM17N-25 circa 2000 Aug 2, 2014 http://glcf.umd.edu/data/mosaic/
Bathymetry/
topography
John et al. [42] Survey data (ADD, MAT,
NOC, NOR, SOC, SOU)
Aug 1, 2014 http://floridaswater.com/technicalreports/
USACE-JAX Survey data (ICW00) Aug 1, 2014 http://www.saj.usace.army.
mil/Missions/CivilWorks/Navigation/
HydroSurveys.aspx
NOAA-CSC 5-m DEM (last-return ground lidar) Aug 1, 2014 http://coast.noaa.gov/digitalcoast/
Boundary GTM
NERR
1.524-m
(5-ft) contour
Aug 20, 2014 http://www.dep.state.fl.us/gtm/
Reserve and watersheds Aug 1, 2014 http://cdmo.baruch.sc.edu/
Coastline NOAA NGDC World vector shoreline Aug 2, 2014 http://www.ngdc.noaa.gov/mgg/coast/
Large-scale domain Hagen et al. [43] Western North Atlantic Aug 17, 2014 https://doi.org/10.1080/10618560601046846
Bacopoulos P. et al.
4.4 Conceptual Model
The loose coupling of MEM with the hydrodynamic mod-
el was applied in this study using the procedure shown in
Tab le 2, which comprises the latest developments of the
conceptual model for integrated simulation of tidal
hydrodynamics and marsh productivity [11,16]. Steps
13 of the procedure apply the hydrodynamic model, step
4 can be done with any GIS-based software [16](herein
done with SMSsurface water modeling system; refer to
[49]), step 5 applies MEM (i.e., biomass curve; refer to
[12]), and step 6 categorizes simulated marsh productivity
Fig. 2 Maps of the GTMNERR showing athe finite element mesh used
for model simulations with boundary conditions (B.C.) applied on north
and south terminals of the Atlantic Intracoastal Waterway, bthe mesh size
distribution, and cthe model representation of bathymetry and
topography. The six sites where field data were collected are numbered
by id. Zoom-in maps show the model mesh and bathymetry/topography
for sites d00, e01, f06, g22, h40, and i46
Sea-Level Rise Impact on Salt Marsh Sustainability and Migration for a Subtropical Estuary: GTMNERR (Guana...
into low, medium, and high categories for qualitative as-
sessment [19]. Steps 15 constitute a single coupling time
step of the incremental time stepping approach with DEM
updating within the Hydro-MEM model based on sedi-
mentation and primary production [37], as defined by
Alizad et al. [16].
4.5 Hydro-MEM Coupling Time Step
The coupling time step used in the Hydro-MEM model is
applied with Eq. 3based on a linear time-stepping scheme:
YtþΔtðÞ¼YtðÞþ ΔtðÞ
dY
dt ð4Þ
where Δtis the coupling time step, Yis the surface elevation
of the marsh, and dY/dt is the accretion rate. In this study, a
coupling time step of 5 years [16] was used for the sea-level
rise scenarios considered in this study: 0.13 m (low), 0.22 m
(intermediate), and 0.51 m (high) [21]. To account for nonlin-
earity of the interactive hydrodynamic-marsh processes,
which can occur for high scenarios of sea-level rise [19], each
50-year simulation (per sea-level rise scenario) with Hydro-
MEM was carried out over a succession of ten 5-year-
increment coupling time steps.
5 Model Results and Validation
5.1 Marsh Productivity
Biomass density was computed for all ADCIRC mesh nodes
using the biomass curve (see step 5 in Table 2). Given the
uncertainties and assumptions in the Hydro-MEM model,
the model-computed biomass density is not treated as an exact
representation of the actual biomass density. Instead, the
model-computed biomass density is considered qualitatively
[16,19] by identifying regions based on low, medium, and
high productivity (see step 6 in Table 2). Figure 4shows
model-computed biomass density categorized into low (0
370gm
2
), medium (370750 g m
2
),andhigh(>
750 g m
2
)productivity[18]. The producing marsh landscape
was computed as covering 8.2% of the overall GTMNERR
area, with 1.2% low, 2.2% medium, and 4.8% high produc-
tivity (Table 3).
As a qualitative validation, model-computed biomass den-
sity of Spartina alterniflora was compared with National
Land Cover Data 2011 [46]. National Land Cover Data
2011 utilizes a range of 20 different land cover classification
of which the following are relevant to the GTMNERR: 11
open water, 20sdeveloped, 40sforests, and 90s
wetlands. The wetland classification includes emergent herba-
ceous wetlands (95) and woody wetlands (90). Figure 5shows
land cover data for the GTMNERR with zoom-in maps for the
six sites where data were collected. There is noticeable simi-
larity between the land cover data and the model-computed
biomass density of Spartina alterniflora (Fig. 4). For example,
there is local expanse of simulated biomass productivity at site
00, where the data show there to be local expanse of wetlands
(land cover classes 95 and 90), while there is little area of
simulated biomass productivity at site 46, where the data show
there to be local expanse of forests (land cover classes 41, 42,
and 43).
To more closely examine land cover at the six sites,
the National Land Cover Data 2011 were clipped to
within 500-m radius of the point location for each of
the six sites (refer to Fig. 3for reference with respect to
elevations and tidal-creek inverts) and categorized into
land cover classifications of 11 (open water); 95 and 90
(wetlands); 41, 42, and 43 (forests); 21, 22, 23, and 24
(developed); and other (other) (Table 4). Of the wetland
classification, most of the areal coverage is emergent
herbaceous wetlands (95) while there is limited areal
coverage of woody wetlands (90). The data show site
06 comprises 61% forests and site 22 comprises 88%
wetlands, which correlates with model-based upland ar-
ea of 79% at site 06 and marsh area of 69% at site 22,
respectively. There is also correlation between data and
model for site 46, where the data show areal coverage
of 28% open water, 37% wetlands, and 36% forests,
and the model shows areas areal coverage of 50% wa-
ter, 34% marsh, and 16% upland. At site 00, the model
computed areal coverage of marsh to be 66% and the
data show areal coverage of wetlands to be 60%. At
sites 01 and 40, the data show widely varied landscape
coverage, where this was reproduced by the model.
Detailed focus was placedonthesixsitesinthe
GTMNERR where field data were collected (Table 4).
Percent coverage of Spartina alterniflora was 3041% at sites
00, 06, 22, and 40 and 35% at sites 01 and 46. The dominant
species at site 01 was Batis maritima, and site 46 was domi-
nated by both Batis maritima and Juncus roemerianus.
Canopy height was 0.270.67 m at sites 00, 06, 22, and 40
and 0.110.29 m at sites 01 and 46. The model-represented
DEM and model-computed MLW, MHW, and biomass den-
sity were plotted based on the distribution of model informa-
tion (sorted by increasing topographic elevation) within 500-
m radius of the point location for each of the six sites (Fig. 6).
The plots show that there is biomass density of Spartina
alterniflora when surface elevation in relation to the tidal hy-
drodynamics (MLW and MHW) is optimal for primary pro-
duction (see Eqs. 1and 2). Sites 00, 06, 22, and 40 are in local
Bacopoulos P. et al.
areas that have surface elevation at or near optimal for primary
production of Spartina alterniflora (i.e., productive), as
shownbythebroadareaofproducingmarshlandscape.
Sites 01 and 46 are in local areas that have surface elevation
that is too low and/or too high for primary production of
Spartina alterniflora (i.e., limited), as shown by the restricted
area of producing marsh landscape. The model predicted lo-
cally broad zones of Spartina alterniflora productivity at sites
00, 06, 22, and 40 (where the field data show the predominant
coverage to be Spartina alterniflora). The model predicted
Fig. 3 Highly zoomed-in maps of
the model mesh showing
bathymetry/topography
(elevationz) and normalized
nonlinearity index (NNL) for six
field-observation sites within the
GTMNERR: sea-level rise (SLR)
scenarios of 0.13, 0.22, and
0.51 m. The bold black lines
delineate the channel inverts of
the tidal creeks. The bold brown
lines represent the 500-m-radius
circles centered about the
individual site locations
Sea-Level Rise Impact on Salt Marsh Sustainability and Migration for a Subtropical Estuary: GTMNERR (Guana...
locally limited zones of Spartina alterniflora productivity at
and around sites 01 and 46 (where the field data show the
predominant coverage to be other than Spartina alterniflora).
The field data for the six sites were correlated with the
model results in the following ways: (1) percent coverage of
Spartina alterniflora (data) versus percent coverage of marsh
(B>0; Eq. 1) (model) and (2) canopy height of Spartina
alterniflora (data) versus biomass density (B) (model). These
correlations are Bloose^in the sense that they are not direct
comparisons of the same variable type; however, they are
useful in providing some insight into the relationship between
the collected field data related to Spartina alterniflora and the
model calculations of marsh productivity. For the first corre-
lation, there is a linear relationship (y=0.84x+24.06; R
2
=
0.28) between the percent coverage of marsh (modely)and
the percent coverage of Spartina alterniflora (datax)
(Fig. 7a). The model over-predicts the percent coverage of
marsh relative to the data for all sites except site 06.
Excluding site 06 from the correlation results in a model-
data linear fit of y=1.53x+16.63 (R
2
= 0.90). For the second
correlation, there is a logarithmic relationship (y=25.90ln(x)-
+ 879.84; R
2
= 0.11) between the biomass density (modely)
and the canopy height of Spartina alterniflora (datax)(Fig.
7b). The logarithmic curve was selected for it being more
appropriate than a linear curve in this casefor example,
the upper bound of biomass density should level off at a max-
imum (i.e., the peak B-value of the biomass curve) regardless
of how large a value of in-situ canopy height. Considering the
measures for site 00 as an outlier with regard to the range of
the overall site measures, the correlation results in a model-
data logarithmic fit of y= 34.48ln(x) + 873.64 (R
2
= 0.40).
Although it is a Bloose^relationship between the field data
and the Hydro-MEM results, there is some value to the above
correlations, more so for percent coverage of marsh (R
2
=
0.280.90) than for canopy height / biomass density (R
2
=
0.110.40).
5.2 Tidal Datums
Using the model-produced tidal constituents, tidal resynthesis
and analysis of resynthesized water levels for tidal datums
(MLW and MHW; refer to [50] for details) were performed
for all ADCIRC mesh nodes. Then, the MLW and MHW
values were negated for the ADCIRC mesh nodes that dried
out at any point over the duration of the simulation. This step
of negating MLW and MHW values for non-fully wetted
zones is taken because the calculation of tidal datums (espe-
cially MLW) is impractical where the signal dries out. These
methods are explained in detail by Alizad et al. [16] and con-
stitute the first three steps of the procedure for the loose cou-
pling of the marsh equilibrium model with the hydrodynamic
model (Table 2).
Figure 8shows the model results of MLW and MHW for
the fully wetted zones throughout the GTMNERR, includ-
ing a comparison between model results and observed
MLW and MHW at eight NOAA tide gauging stations lo-
cated within the domain. The model results suggest that
MLW and MHW are non-uniform in space. In fact, there
is noticeable spatial variability of MLW within the ranges
between 0.8 and 0.2 m and of MHW within the ranges
Table 2 Procedure for the loose
coupling of the marsh equilibrium
model with the tidal
hydrodynamic model. Modified
from an original process
flowchart ([16]cf. Fig. 2)
Step Directions
1
a
Run the ADCIRC simulation to obtain tidal constituents
for all ADCIRC mesh nodes.
2
a
Using the tidal constituents, perform a tidal resynthesis and
calculate tidal datums for all ADCIRC mesh nodes: mean
low water (MLW) and mean high water (MHW).
3
a
Load MLW, MHW and the mask
b
into GIS-based tool and clip
out and remove the MLWand MHW values for the intermittently
wetted and fully dry zones, to generate MLW and MHW information
for the fully wetted zones only.
4
a
Extrapolate (IDW) the masked/clipped MLW and MHW information
to all ADCIRC mesh nodes.
5
a
Compute D from E, then MLWand MHW for all ADCIRC mesh
nodes to compute B from D and the biomass
curve [12] for all ADCIRC mesh nodes.
6 Categorize the biomass density such that marshes with productivity
less than 370 g m
2
are categorized as low, between 370 and 750 g m
2
are categorized as medium, and greater than 750 g m
2
are
categorized as high productivity [19].
a
Steps 15 constitute a single coupling time step. To account for marsh migration in the model, the DEM is
updated each coupling time step based on simulated biomass productivity
b
The mask is a map file that separates by polygon the fully wetted zones versus the intermittentlywetted and fully
dry zones
Bacopoulos P. et al.
of 0.4 and 0.8 m. Tidal datums are reported in this paper as
being relative to NAVD88. For the eight stations, the data
show that MLW ranges from 0.71 to 0.54 m and MHW
ranges from 0.56 to 0.71 m. The model predictions show
that MLW ranges between 0.74 and 0.34 m and MHW
ranges between 0.54 and 0.75 m (Table 5). Among the
Fig. 4 a Areal coverage of landscape classification was mapped out for
present-day conditions. The six sites where field data were collected are
numbered by id. Zoom-in maps show landscape classification for sites b
00, c01, d06, e22, f40, and g46. Categories of biomass density include
low B (B<370gm
2
), medium B (370 g m
2
<B<750gm
2
), and high
B(B>750gm
2
)
Table 3 Average (AVG) and standard deviation (STD) of tidal datums
(MLW and MHW) over the producing marsh landscape for present-day
conditions and for 50-year simulations of sea-level rise of 0.13, 0.22, and
0.51 m. The NNL values are values of normalized nonlinear index based
on the AVG values of MLW and MHW. Areal coverage by landscape
classification for present-day conditions and for 50-year simulations of
sea-level rise of 0.13, 0.22, and 0.51 m
Scenario MLW MHW Areal coverage of landscape coverage (%)
AVG (m) STD (m) NNL ()AVG(m)STD(m)NNL() Water Upland Marsh Low B Med B High B
Present-day 0.56 0.17 0.72 0.09 35.8 56.0 8.2 1.2 2.2 4.8
Sensitivity v1 0.56 0.17 0.72 0.09 35.8 56.1 8.1 1.3 2.1 4.7
Sensitivity v2 0.56 0.17 0.72 0.09 35.8 56.1 8.1 1.3 2.1 4.7
SLR = 0.13 m 0.45 0.14 0.15 0.85 0.08 0.00 36.0 53.9 10.1 2.3 3.0 4.8
SLR = 0.22 m 0.38 0.13 0.18 0.95 0.08 0.05 36.3 53.2 10.5 1.7 3.5 5.3
SLR = 0.51 m 0.14 0.10 0.18 1.25 0.07 0.04 37.9 51.5 10.5 1.6 2.2 6.7
Sea-Level Rise Impact on Salt Marsh Sustainability and Migration for a Subtropical Estuary: GTMNERR (Guana...
eight stations, root mean square errors between observa-
tions and model results were calculated to be 0.06 and
0.08 m for MLW and MHW, respectively.
The simulated fields of MLW and MHW (for the fully
wetted zones) were extrapolated using inverse-distance-
weighting method [16] to the intermittently wetted zones
and fully dry zones (see step 4 in Table 2). Statistics were
computed to quantify the spatial variability of MLW and
MHW over the producing marsh landscape (Table 3). The
average MLW was computed to be 0.56 m with standard
deviation of 0.17 m (approximately 30% of the average val-
ue). The average MHW was computed to be 0.72 m with
standard deviation of 0.09 m (approximately 13% of the av-
erage value). The standard deviation values for MLW are
greater than the standard deviation values for MHW, suggest-
ing more spatial variability of MLW than MHW. These results
of spatial variability of tidal datums (especially with MLW)
correspond with the data (Table 5) and other studies of tides in
salt marsh landscapes [11,16,19,20].
6 Model Application for Analysis of Sea-Level
Rise Impacts
The analysis of sea-level rise impacts is presented in the fol-
lowing two subsections: the marsh migration due to sea-level
rise and the hydrodynamic response due to sea-level rise.
However, before applying Hydro-MEM for the analysis of
sea-level rise impacts, the model was tested for sensitivity to
DEM error. The sensitivity analysis was conducted not neces-
sarily to arrive at a Bperfect^DEM for the model domain but
to generate an Benvelope^of model sensitivity with respect to
Fig. 5 aMap of National Land Cover Data 2011 showing water, wetland,
and other land cover classifications. The six sites where field data were
collected are numbered by id. Zoom-in maps show land cover data for
sites b00, c01, d06, e22, f40, and g46. Wetlands 90 correspond to
woody wetlands. Wetlands 95 correspond to emergent herbaceous
wetlands
Bacopoulos P. et al.
DEM error, in general. In this regard, the sensitivity analysis
provides context for the evaluation of marsh productivity and
tidal datums for the sea-level rise scenarios. Refer to
Appendix for details of the sensitivity analysis.
6.1 Marsh Migration Due to Sea-Level Rise
Figure 9shows model-computed biomass density categorized
into low (0370 g m
2
), medium (370750gm
2
), and high
(> 750 g m
2
)productivity[18] for sea-level rise scenarios of
0.13, 0.22, and 0.51 m. The evolution of the marsh is spatially
based such that the marsh migrates with increasing sea level,
namely to migrate further away from the existing tidal creeks.
However, there are some lands at higher elevations into which
the marsh cannot migrate, thus limiting the expanse of the
marsh due to a combined Bsqueeze^effect of increasing sea
level (from the coastal side) and locally steep topographic
gradients (from the landward side). There are also regions of
existing marsh that will drown under sea-level rise and con-
vert to open water. Areal coverages were calculated by land-
scape classification for the sea-level rise scenarios of 0.13,
0.22, and 0.51 m (Table 3). There is a decrease in areal cov-
erage of upland for greater sea-level rise, an increase in areal
coverage of water, and a fairly consistent (slightly increasing)
areal coverage of marsh (low B + medium B + high B), which
corroborates with the migration of marsh into the upland as
more of the overall area becomes affected by water (tides)
because of sea-level rise. The areal coverages of low B, me-
dium B, and high B are dissimilar for the different sea-level
rise scenarios because of the relationship between biomass
density and sea state (i.e., MLW, MHW, and sea-level rise).
To rectify the projections of marsh migration, the model
predictions were compared with the National Land Cover
Data 2011 (Fig. 5) which identified developed/hardened zones
where marsh cannot migrate, undeveloped/available zones
where marsh can migrate, and zones where existing marsh
can be potentially restored (e.g., spreading of dredge spoils)
to prevent conversion to open water. For this, the model re-
sults for the sea-level rise scenario of 0.51 m were analyzed
according to the following criteria: criterion (1) if there was
predicted marsh for the sea-level rise scenario when there was
no predicted marsh for present-day conditions and where de-
veloped land cover exists, criterion (2) same as criterion 1
except where undeveloped land cover exists, and criterion
(3) if there was predicted open water for the sea-level rise
scenario when there was predicted marsh for present-day con-
ditions. Figure 10 shows maps of the rectified projections of
marsh migration for the sea-level rise scenario of 0.51 m.
Table 4 Six sites in the GTM-NERR where field data were collected.
Field data include species identification, percent coverage of Spartina
alterniflora, and canopy height of Spartina alterniflora. The field data
are averaged per site for the three stations each with five vegetation plots.
Model results include areal coverage of upland, water and marsh zones
for the producing marsh, and are based on the spatial values within 500-m
radii of the six point locations. The areal coverage of National Land
Cover Data (NLCD) 2011 are based on the spatial values within 500-m
radii of the six point locations
Site Location Field data Model results National Land Cover Data (NLCD) 2011
Percent Canopy Areal coverage (%) Land cover classification by areal coverage (%)
coverage height Water Wetland
a
Wetland
b
Forest Developed Other
(id, name) °N °W (%) (m) Upland Water Marsh 11 95 90 40s 20s Other
00
Hat Island
29.9781 81.3196 36 0.27 34.4 0.0 65.6 24.9 60.0 0.0 3.4 6.5 5.1
01
Washington Oaks
c
29.6289 81.2190 5 0.11 16.4 50.0 33.6 28.8 63.6 2.4 5.2 0.0 0.0
06
Moses Creek
29.7715 81.2874 41 0.40 79.0 0.0 21.0 0.0 35.0 1.1 60.7 2.4 0.7
22
JasonsCreek
29.8215 81.2891 30 0.47 0.0 31.0 69.0 12.2 87.8 0.0 0.0 0.0 0.0
40
Pine Island
30.0851 81.3667 34 0.67 13.6 17.3 69.2 58.3 37.6 0.0 3.4 0.0 0.6
46
Pellicer Creek
d
29.6619 81.2466 3 0.29 50.5 38.2 11.3 27.7 34.5 2.3 35.5 0.0 0.0
a
Emergent herbaceous wetlands
b
Woody wetlands
c
Dominant species are Batis maritima and Juncus roemerianus
d
Dominant species are Batis maritima and Juncus roemerianus
Sea-Level Rise Impact on Salt Marsh Sustainability and Migration for a Subtropical Estuary: GTMNERR (Guana...
Fig. 6 Model-represented DEM and model-computed MLW, MHW, and
biomass density were plotted for each of the six sites where field data
were collected. The plot data range for each site is based on the
distribution of model information (sorted by increasing topographic
elevation) within 500-m radius of the point location. The gray shading
in the pie plots represents the percent coverage of Spartina alterniflora for
each site. The cartooned marsh grasses are scaled according to the canopy
height of Spartina alterniflora measured in the field
Bacopoulos P. et al.
There are identifiable zones where existing marsh will not
survive and will convert to open water due to sea-level rise
(i.e., criterion 3), and these zones tend to be located near and
along the tidal creeks. Zones of marsh migration tend to be
located far away from the tidal creeks showing where the
marsh will migrate into undeveloped lands (i.e., criterion 2).
However, even if lands are at appropriate elevations for future
marsh productivity, the marsh cannot migrate into developed
lands (i.e., criterion 1), thus nullifying those predictions.
Rectifying the marsh projections for land cover demonstrates
a practical use of the model results for identifying wetlands
that can be preserved and marsh that can be restored. In this
manner, the model is a guidance tool for coastal planners in
making decisions regarding marsh sustainability and migra-
tion in the continually developing reserve.
6.2 Hydrodynamic Response Due to Sea-Level Rise
Hydrodynamic response due to sea-level rise can be described
as static or dynamic [10,51], where the hydrodynamic re-
sponse can be one-to-one with sea-level rise (i.e., static) or
greater than or less than one-to-one with sea-level rise (i.e.,
dynamic). This analysis examines the resulting outcome (im-
pacts on hydrodynamics) in the context of linear (static) versus
nonlinear (dynamic) response with regard to the change in
model input (sea-level rise). The normalized nonlinearity in-
dex (NNL) is used in the analysis to measure the nonlinearity
of the hydrodynamic response (MLW and MHW) due to a
given amount of sea-level rise [52]:
NNL ¼ζSLR
ζCONTROL
λ
1ð5Þ
where NNL is the normalized nonlinearity index, ζis
the hydrodynamic variable (MLW and MHW), λis the
applied sea-level rise, and the subscripts SLR and
CONTROL indicate the hydrodynamic variables for the
given sea-level rise scenario and present-day conditions,
respectively. NNL value of zero indicates static re-
sponse, where a non-zero NNL value indicates dynamic
response (either positive or negative).
MLW and MHW for present-day conditions and sea-level
rise scenarios (0.13, 0.22, and 0.51 m) were analyzed for the
eight NOAA tide gauging stations in the GTMNERR (see Fig.
8a for station locations). Figure 11 plots NNL values for MLW
and MHW resulting from the sea-level rise scenarios of 0.13,
0.22, and 0.51 m, where the envelope of NNL is bound by the
maximum and minimum values for the eight NOAA tide
gauging stations in the GTMNERR and the AVG value of
NNL is the average value for the eight NOAA tide gauging
stations in the GTMNERR. The zero line is plotted to indicate
a purely static situation (i.e., one-to-one change of MLW/
MHW with sea-level rise). For both MLW and MHW, the
envelope of NNL closes with respect to increasing value of
sea-level rise, suggesting that tidal datums in the GTMNERR
will change more linearly in the longer term (i.e., greater sea-
level rise), while there will be measurable nonlinearity (both
positive and negative) in the shorter term (i.e., lesser sea-level
rise). Furthermore, the AVG values of NNL for MLW and
MHW tend towards zero with increasing value of sea-level
rise. The AVG values of NNL for MHWare slightly negative
and the AVG values of NNL for MLW are notably negative,
suggesting that MHW will increase near linearly with sea-
Fig. 7 Correlation plots between the field data and the Hydro-MEM
results. aPercent coverage of marsh (model) versus percent coverage of
Spartina alterniflora (data). bBiomass density (model) versus canopy
height of Spartina alterniflora (data)
Sea-Level Rise Impact on Salt Marsh Sustainability and Migration for a Subtropical Estuary: GTMNERR (Guana...
level rise while MLW will increase nonlinearly (less than one-
to-one) with sea-level rise, which has the combined effect of
dynamic increase (greater than one-to-one) of the tide range
(MHWMLW) with sea-level rise. These observations are
reflected in the spatial results of the model when inspected
for the six field-data sites (Fig. 3).
Statistics were computed to quantify the spatial vari-
ability of MLW and MHW over the producing marsh
landscape for the sea-level rise scenarios of 0.13, 0.22,
and 0.51 m (Table 3). The AVG values for MLW and
MHW increase with sea-level rise; however, the increase
is not linear (one-to-one). The STD values for MLW
and MHW decrease with sea-level rise, which suggests
more uniformity in tidal datum elevations with greater
sea-level rise. The NNL values are negative for MLW
(ranging between 0.15 and 0.18) and slightly posi-
tive for MHW (ranging between 0.00 and 0.05), where
the NNL magnitudes for MLW are about three times
those of MHW. The combined effect of nonlinear in-
creases in MLW and MHW results in dynamic increase
(greater than one-to-one) of the tide range (MHW
MLW), as impacted by sea-level rise.
7 Conclusions
Sea-level rise impacts on inundation, tidal hydrodynamics,
and biomass productivity of Spartina alterniflora for the
Guana Tolomato Matanzas National Estuarine Research
Reserve (GTMNERR) were investigated using a loosely
Fig. 8 Maps of model-computed tidal datums for the fully wetted zones of the GTM. aMean low water (MLW). bMean high water (MHW). The eight
NOAA tide gauging stations in the GTMNERR are numbered by id. Regions colored in gray represent the modeled inundation extent
Bacopoulos P. et al.
coupled tide-marsh equilibrium model (Hydro-MEM).
The hydrodynamic component of Hydro-MEM
accounted for the spatially variable tidal hydrodynamics
over the two-dimensional landscape as subject to sea-
level rise. MEM accounted for the marsh productivity
and accretion as driven by the hydrodynamic model-
computed tidal datums (mean low waterMLW and
mean high waterMHW). Hydro-MEM was shown to
exhibit rate-sensitivity with respect to sea-level rise ex-
ceeding the marsh accretion rate, whereby a sudden loss
of marsh elevation occurred in such instances of marsh
destabilization. Demonstrating rate-critical transition, the
model proved flexible to account for the non-
homogeneous and transient nature of the fast-slow var-
iables, whereby the marsh migrated away from the tidal
creeks and further into the upland zones.
The Hydro-MEM model was defined for the
GTMNERR with dense yet flexible mesh resolution
for simulation of the mechanistic processes of tidal hy-
drodynamics (shallow-water equations) and biomass pro-
ductivity of Spartina alterniflora (biomass curve). The
Hydro-MEM modeling is novel with respect to its ap-
plication to the GTMNERR and providing local man-
agers with a utility to identify lands available for future
marsh migration (e.g., zonation and preservation) and
maintenance of existing marsh by engineered accretion
(e.g., spreading of dredge spoils to facilitate elevation
change of the marsh table). The modeling tool is exten-
sible to other estuaries, where the logical next step is to
fully automate the model coupling procedure for tech-
nology transfer to end users.
The modeling considered the spatial variability of tidal da-
tums in the prediction of biomass density, which was
accomplished with the hydrodynamic component of the mod-
el. Coupled with the model-computed MLW and MHW, the
marsh equilibrium model predicted regions of low, medium,
and high biomass productivity that compared qualitatively
with wetland coverage defined by land cover data. The model-
ing generated a detailed map of biomass productivity that
compared locally with field data collected at six vegetation
sites in the GTMNERR, which included four Bproductive^
sites and two Blimited^sites. The Bproductive^sites contained
percent coverage of Spartina alterniflora of 3041%, canopy
height of 0.270.67 m, and simulated biomass density of
greater than 750 g m
2
over at least half of the local area
within 500-m radii of the measurement sites. The limited sites
contained percent coverage of Spartina alterniflora of 35%,
canopy height of 0.110.29 m, and simulated biomass density
of greater than 750 g m
2
over less than one-tenth of the local
area within 500-m radii of the measurement sites. The model-
ing demonstrated the sensitivity of simulated biomass produc-
tivity due to error in lidar data used in the digital elevation
model, which affects the prediction of where the marsh is
more or less productive with regard to distance from the tidal
creeks. The hydrodynamic model reproduced tidal datums,
including mean low water (MLW) and mean high water
(MHW), with root mean square error of (at greatest) 0.08 m,
when compared with data at eight NOAA tide gauging sta-
tions in the GTMNERR.
The modeling showed the impacts of sea-level rise on the
GTMNERR in the following ways: productive Spartina
alterniflora marsh will migrate further away from the existing
tidal creeks, given available landscape at suitable elevations,
due to sea-level rise and the subsequent increase in tidal datum
elevations; tidal datums (MLW and MHW) will change dy-
namically (both greater than and less than one-to-one) in the
Table 5 Eight NOAA tide gauging stations in the GTMNERR where tidal datums (MLW and MHW) were compared: observation (OBS) vs. model
(MOD)
Station Location Locale description MLW (m)
a
MHW (m)
a
(id-name) (°N) (°W) OBS MOD OBS MOD
1Pablo Creek 30.3230 81.4380 Intracoastal (near boundary) 0.57 0.50 0.58 0.54
2Tolomato River 29.9900 81.3200 Intracoastal 0.69 0.67 0.62 0.75
3Vilano Bridge 29.9100 81.3000 Back of St. Augustine Inlet 0.66 0.74 0.64 0.74
4St. Augustine 29.8920 81.3100 Back of St. Augustine Inlet 0.69 0.72 0.68 0.75
5Anastasia Island 29.7930 81.2720 Intracoastal 0.71 0.74 0.67 0.73
6Fort Matanzas 29.7150 81.2380 Back of Matanzas Inlet 0.61 0.70 0.56 0.70
7Matanzas River Headwaters 29.6300 81.2100 Intracoastal 0.54 0.49 0.57 0.59
8St. Augustine Beach 29.8567 81.2633 Open coast 0.69 0.72 0.71 0.75
Root mean square error (m) 0.06 0.08
Sea-Level Rise Impact on Salt Marsh Sustainability and Migration for a Subtropical Estuary: GTMNERR (Guana...
Bacopoulos P. et al.
shorter term (i.e., lesser sea-level rise) but will tend towards
static response (one-to-one) in the longer term (i.e., greater
sea-level rise); and the tide range (MHWMLW) will increase
dynamically (greater than one-to-one) with sea-level rise. The
marshes in the GTMNERR were shown to be resilient in their
ability to sustain productivity in the event of risingsea level by
migrating into the uplands with topographic elevation favor-
able for productivity. So long as those uplands are preserved,
the marshes will be able to migrate with sea-level rise, while
some existing low-topography marsh will be converted into
water-type classification (e.g., mud flat or open water).
As for future work, this study expresses the need for field
data collection of sedimentation and primary production to
derive values of q(sediment loading rate) and k(organic/in-
organic accumulation by primary production/sediment trap-
ping) specific to the GTMNERR, including the installation
of marsh organs of Spartina alterniflora for development of
biomass curves for the GTMNERR. In addition, the Hydro-
MEM application for the GTMNERR, and others like tidal
hydrodynamic-marsh modeling efforts of the region, supports
the need of a dedicated field campaign for the collection of
RTK data over the marsh platform. Incorporating field-based
coefficients for the biomass curve (a,b,andc;Eq.1) and data-
derived MEM parameters (qand k;Eq.3), with RTK data and
remotely sensed data for Btruthing^the DEM, constitute the
next steps in the Hydro-MEM modeling for the GTMNERR
Fig. 9 Areal coverage of landscape classification was mapped out for 50-
year simulations of sea-level rise (SLR) of a0.13, h0.22, and o0.51 m.
Categories of biomass density include low B (B<370gm
2
), medium B
(370gm
2
<B<750 g m
2
),andhighB(B>750 g m
2
). The six sites
where field data were collected are numbered by id. Zoom-in maps show
landscape classification for sites b,i,p00; c,j,q01; d,k,r06; e,l,s22; f,
m,t40; and g,n,u46
Fig. 10 aMap of rectified projections of marsh migration for the sea-
level rise scenario of 0.51 m showing zones meeting three criteria:
criterion (1) developed lands into which marsh cannot migrate; criterion
(2) undeveloped lands into which marsh can migrate; and criterion (3)
existing marsh that cannot survive and will convert to open water. The six
sites where field data were collected are numbered by id. Zoom-in maps
show rectified projections for sites: b00, c01, d06, e22, f40, and g46
Sea-Level Rise Impact on Salt Marsh Sustainability and Migration for a Subtropical Estuary: GTMNERR (Guana...
towards gaining furthered confidencewith the results ofmarsh
productivity and migration. Nevertheless, the present results
demonstrate Hydro-MEMs utility for qualitative assessment
of the marsh response in the GTMNERR due to sea-level rise
impacts on the hydroperiod (MLW and MHW), which serves
useful for coastal management and biological monitoring of
the reserve.
Acknowledgements This research was funded in part by the Taylor
Engineering Research Institute. Gratitude is given to Dr. Donald Resio
for his contribution to the study. The vegetation monitoring was conduct-
ed under an award from the Estuarine Research Division, Office for
Coastal Management, National Oceanic and Atmospheric
Administration. Special thanks are given to the staff of the Guana
Tolomato Matanzas National Estuarine Research Reserve for the field
data. We thank two anonymous reviewers for their insightful comments
on the manuscript.
Sensitivity to DEM error
Lidar error arises in coastal regions with tall-grass salt marsh
where the data can be polluted by vegetation hits
misrepresenting true bare-earth elevation (see [53] for review
of the subject). Since the DEM is an essential ingredient in the
Hydro-MEM model by way of underlying the unstructured
mesh [54,55], this section investigates the sensitivity of the
model predictions regarding lidar error in the DEM, and the
correction thereof [18,19].
As an initial case (version 1v1), the DEM was corrected
for lidar error based on lowering the marsh surface elevation
by 0.32 m for all ADCIRC mesh nodes with high biomass
density, by 0.23 m for medium biomass density and by 0.16 m
for low biomass density [18]. Supplement 1shows a map of
elevation differences between the lidar error-corrected DEM
and the original DEM, where elevation differences range from
0to0.32 m. Naturally, the spatial distribution of the DEM
differences is reflective of the spatial distribution of biomass
density predicted by the Hydro-MEM model (Fig. 4), since
the lidar error corrections to the DEM were based on the
model-computed biomass density. The lidar error-corrected
DEM was applied in the Hydro-MEM model to compute tidal
datums (MLW and MHW) and biomass density of Spartina
alterniflora for present-day conditions.
Differences of model-computed biomass density were cal-
culated as Blidar error-corrected DEM^minus Boriginal
DEM^and categorized into small (magnitude 0370 g m
2
),
medium (magnitude 370750gm
2
), and large (magnitude >
750 g m
2
) sensitivity, as shown in Supplement 2. The results
show that lowering the DEM based on correction for lidar
error can lead to prediction of decreased biomass density for
locations nearer the tidal creeks and increased biomass density
for locations farther from the tidal creeks. The sensitivity of
modeled biomass density due to DEM error is correlated with
the prediction of where the marsh is more or less productive,
such that the predicted locations of productive marsh can be
off-computed with regard to distance from the tidal creeks.
Tidal datums were shown to be uninfluenced by the lidar-
error correction of the DEM, such that MLW and MHW were
computed the same for the lidar error-corrected DEM and the
Fig. 11 Values of normalized
nonlinearity index (NNL) were
calculated for sea-level rise (0.13,
0.22, and 0.51 m) impacts on a
mean low water (MLW) and b
mean high water (MHW). The
envelope of NNL is bound by the
maximum and minimum values
for the eight NOAA tide gauging
stations in the GTMNERR. The
AVG value of NNL is the average
value for the eight NOAA tide
gauging stations in the
GTMNERR. The zero line is
shown for reference as a purely
linear response
Bacopoulos P. et al.
original DEM, and the landscape classification was computed
to be almost the same (i.e., lidar error-corrected DEM versus
original DEM) when considering the areal coverages of water,
upland, and marsh, including the producing marsh categorized
into low, medium, and high productivity (Table 3).
In the absence of real-time kinematic (RTK) topographic
survey data or remotely sensed data for the region to use as
basis for correction of the inherent lidar error [18,19], the only
resort for guiding ground-truth data was to employ National
Land Cover Data 2011 [46]. As an added case, the DEM was
corrected for lidar error based on lowering the marsh surface
elevation by 0.23 m for all ADCIRC mesh nodes located
within wetland land types (land cover classes 95 and 90),
and this second version (v2) of a lidar error-corrected DEM
was applied in the Hydro-MEM model to compute tidal da-
tums (MLW and MHW) and biomass density of Spartina
alterniflora for present-day conditions. For this extra case,
the landscape classification was computed to be virtually the
same as the initial case of lidar-error correction in the DEM
(Table 3).
References
1. Fagherazzi, S., Wiberg, P. L., Temmerman, S., Struyf, E., Zhao, Y.,
& Raymond, P. A. (2013). Fluxes of water, sediments and biogeo-
chemical compounds in salt marshes. Ecological Processes, 2,3.
2. Kennish, M. J. (2001). Coastal salt marsh systems in the US: A
review of anthropogenic impacts. Journal of Coastal Research,
17,731748.
3. Kirwan, M. L., & Megonigal, J. P. (2013). Tidal wetland stability in
the face of human impacts and sea-level rise. Nature, 504,5360.
4. Barbier, E. B., Hacker, S. D., Kennedy, C., Koch, E. W., Stier, A.
C., & Silliman, B. R. (2011). The value of estuarine and coastal
ecosystem services. Ecological Monographs, 81,169193.
5. Kirwan, M. L., Langley, J. A., Guntenspergen, G. R., & Megonigal,
J. P. (2013). The impact of sea-level rise on organic matter decay
rates in Chesapeake Bay brackish tidal marshes. Biogeosciences,
10,18691876.
6. Montague, C. J., & Wiegert, R. G. (1990). Salt marshes. In R. L.
Myers & J. J. Ewel (Eds.), Ecosystems of Florida (pp. 481516).
Orlando: University of Central Florida Press.
7. Zervas, C. 2001. Sea level variations of the United States 1854
1999. NOAA Technical Report NOS CO-OPS 36, Washington,
DC, 80 p.
8. Jevrejeva, S., Moore, J. C., & Grinsted, A. (2010). How will sea
level respond to changes in natural and anthropogenic forcings by
2100? Geophysical Research Letters, 37, L07703.
9. Walton, T. L. (2007). Projected sea level rise in Florida. Ocean
Engineering, 34,18321840.
10. Passeri, D. L., Hagen, S. C., Medeiros, S. C., Bilskie, M. V., Alizad,
K., & Wang, D. (2015). The dynamic effects of sea level rise on
low-gradient coastal landscapes: a review. EarthsFuture,3,159
181.
11. Hagen, S. C., Morris, J. T., Bacopoulos, P., & Weishampel, J. F.
(2013). Sea-level rise impact on a salt marsh system in lower St.
Johns River. Journal of Waterway, Port, Coastal and Ocean
Engineering, 139,118125.
12. Morris, J. T., Sundarreshwar, P. V., Nietch, C. T., Kjerfve, B., &
Cahoon, D. R. (2002). Responses of coastal wetlands to rising sea
level. Ecology, 83,28692877.
13. Siteur, K., Eppinga, M., Doelman, A., Siero, E., & Rietkerk, M.
(2016). Ecosystems off track: rate-induced critical transitions in
ecological models. Oikos, 125, 16891699.
14. Scheffer, M., van Nes, E. H., Holmgren, M., & Hughes, T. (2008).
Pulse-driven loss of top-down control: the critical-rate hypothesis.
Ecosystems, 11,226237.
15. Morris, J. T. (2007). Ecological engineering in intertidal
saltmarshes. Hydrobiologia, 577,161168.
16. Alizad, K., Hagen, S. C., Morris, J. T., Bacopoulos, P., Bilskie, M.
V., & Weishampel, J. F. (2016a). A coupled, two-dimensional hy-
drodynamic-marsh model with biological feedback. Ecological
Modelling, 327,2943.
17. Ashwin, P., Wieczorek, S., Vitolo, R., & Cox, P. (2012). Tipping
points in open systems: bifurcation, noise-induced and rate-
dependent examples in the climate system. Philosophical
Transactions of the Royal Society, Series A, 370, 11661184.
18. Medeiros, S. C., Hagen, S. C., Weishampel, J. F., & Angelo, J.
(2015). Adjusting lidar-derived digital terrain models in coastal
marshes on estimated aboveground biomass density. Remote
Sensing, 7,35073525.
19. Alizad, K., Hagen, S. C., Morris, J. T., Medeiros, S. C., Bilskie, M.
V., & Weishampel, J. F. (2016b). Coastal wetland response to sea-
level rise in a fluvial estuarine system. Earths Future, 4,483497.
20. Tritinger, A.S., 2015. Numerical modeling and analysis of tidal
variance and marsh productivity in the Guana Tolomato Matanzas
National Estuarine Research Reserve. MS thesis, University of
North Florida, 136 p.
21. Parris, A., Bromirski, P., Burkett, V., Cayan, D., Culver, M., Hall, J.,
Horton, R., Knuuti, K., Moss, R., Obeysekera, J., Sallenger, A.,
Weiss, J., 2012. Global sea level rise scenarios for the US
National Climate Assessment. NOAA Technical Memorandum
OAR CPO, Washington, DC, 37 p.
22. Parkman, A. 1983. History of the waterways of the Atlantic coast of
the United States. Navigation History NWS-83-10, National
Waterways Study, U.S. Army Engineer Water Resources Support
Center, Institute for Water Resources, Virginia, 147 p.
23. Crawford, W.G. 2007. Floridas big dig: the Atlantic Intracoastal
Waterway. Florida Historical Society Press (1
st
edition), 371p.
24. Dix, N. G., Phlips, E. J., & Gleeson, R. A. (2008). Water quality
changes in the Guana Tolomato Matanzas National Estuarine
Research Reserve, Florida, associated with four tropical storms.
Journal of Coastal Research, 55,2637.
25. Scavia, D., Field, J. C., Boesch, D. F., Buddemeier, R. W., Burkett,
V., Cayan, D. R., Fogarty, M., Harwell, M. A., Howarth, R. W.,
Mason, C., Reed, D. J., Royer, R. C., Sallenger, A. H., & Titus, J. G.
(2002). Climate change impacts on U.S. coastal and marine sys-
tems. Estuaries, 25,149164.
26. CDM/DHI, 2002. Generation of a curvilinear-orthogonal grid and
tidal model for the northern Coastal Basin. Final report to the St.
Johns River Water Management District, Palatka, Florida, 88 p.
27. Sheng, Y. P., Tutak, B., Davis, J. R., & Paramygin, V. (2008).
Circulation and flushing in the lagoonal system of the Guana-
Tolomato-Matanzas National Estuarine Research Reserve
(GTMNERR), Florida. Journal of Coastal Research, 55,925.
28. Linhoss, A. C., Kiker, G., Shirley, M., & Frank, K. (2015). Sea-
level rise, inundation and marsh migration: simulating impacts on
developed lands and environmental systems. Journal of Coastal
Research, 31,3646.
29. Wu, W., Yeager, K. M., Peterson, M. S., & Fulford, R. S. (2015).
Neutral models as a way to evaluate the sea level affecting marshes
model (SLAMM). Ecological Modeling, 303,5569.
30. Moore, K. 2013. NERRS SWMP vegetation monitoring protocol:
long-term monitoring of estuarine vegetation communities.
Sea-Level Rise Impact on Salt Marsh Sustainability and Migration for a Subtropical Estuary: GTMNERR (Guana...
Technical report, National Estuarine Research Reserve System,
Gloucester Point, VA. [Accessed online July 10, 2015: http://
gtmnerr.org/documents/Research_Publications/NERRS_
Vegetation_Monitoring_2013-09-06.pdf.]
31. Asper, J., Curtis, A.C. 2013. Salt marsh elevation monitoring.
Southeast Coast Network Standard Operating Procedure NPS/
SECN/SOP-1.3.6, National Park Service, Athens, GA. [Accessed
online July 10, 2015: http://irmafiles.nps.gov/reference/holding/
501180.]
32. Curtis, A.C., Asper, J., Eastman, S., Baron, L.C. 2013. Photoplot-
based monitoring of salt marsh vegetation. Southeast Coast
Network Standard Operating Procedure NPS/SECN/SOP-1.3.9,
National Park Service, Athens, GA. [Accessed online July 10,
2015: http://irmafiles.nps.gov/reference/holding/510298.]
33. Byrne, M.W. 2012. Generating a spatially-balanced random sample
with the RRQRR tool in ArcGIS 9.1 and ArcGIS 9.3.1 and apply-
ing selection criteria to select sampling locations. Southeast Coast
Network Standard Operating Procedure NPS/SECN/SOP-1.1.1,
National Park Service, Athens, GA. [Accessed online July 10,
2015: http://irmafiles.nps.gov/reference/holding/501173.]
34. Booth, D. T., Cox, S. E., & Berryman, R. D. (2006). Point sampling
digital imagery with SamplePoint. Environmental Monitoring and
Assessment, 123,79108.
35. Booth, D. T., Cox, S. E., Louhaichi, M., & Johnson, D. E. (2004).
Technical note: Lightweight camera stand for close-to-earth remote
sensing. Journal of Range Management, 57,675678.
36. Adobe. 2015. Adobe Photoshop Elements [Accessed online July
10, 2015: http://www.adobe.com/products/photoshop-elements.
html?promoid=DJDWS.]
37. Morris, J. T., Sundberg, K., & Hopkinson, C. S. (2013). Salt marsh
productivity and its responses to relative sea level and nutrients in
estuaries at Plum Island, Massachusetts, and north inlet, South
Carolina, USA. Oceanography, 26,7884.
38. Stevenson, J. C., Ward, L. G., & Kearney, M. S. (1986). Vertical
accretion in marshes with varying rates of sea level rise. In D. A.
Wol f e ( Ed. ) , Estuarine variability (pp. 249259). New York:
Academic.
39. Luettich, R.A., Westerink, J.J., Scheffner, N.W. 1992. ADCIRC: an
advanced three-dimensional circulation model for shelves, coasts,
and estuaries, I: theory and methodology of ADCIRC-2DDI and
ADCIRC-3DL. Technical Report DRP-92-6, U.S. Army Corps of
Engineers, Waterways Experiment Stations, Vicksburg, MS.
40. Kolar, R. L., Westerink, J. J., Gray, W. G., & Luettich, R. A.
(1994b). Shallow water modeling in spherical coordinates: equation
formulation, numerical implementation, and application. Journal of
Hydraulic Research, 32,324.
41. Kolar, R. L., Westerink, J. J., Cantekin, M. E., & Blain, C. A.
(1994a). Aspects of nonlinear simulations using shallow-water
models based on the wave continuity equation. Computers and
Fluids, 23,523538.
42. John, C., Morris, F., Soeder, E. 2003. Northern Coastal Basin bathy-
metric survey: Oak Landing to Ponce de Leon Inlet. Technical
Memorandum No. 49, St. Johns River Water Management
District, 150 p.
43. Hagen, S. C., Zundel, A. K., & Kojima, S. (2006). Automatic,
unstructured mesh generation for tidal calculations in a large do-
main. International Journal of Computational Fluid Dynamics, 20,
593608.
44. GeoCover Landsat. 2014. Earth Science Data Interface. [Accessed
online August 2, 2014: http://glcf.umd.edu/data/mosaic/.]
45. NOAA Office for Coastal Management, 2014. DigitalCoast.
[Accessed online August 1, 2014: http://coast.noaa.gov/
digitalcoast/.]
46. Jin, S., Yang, L., Danielson, P., Homer, C., Fry, J., & Xian, G.
(2013). A comprehensive change detection method for updating
the National Land Cover Database to circa 2011. Remote Sensing
of Environment, 132,159175.
47. Medeiros, S. C., Hagen, S. C., & Weishampel, J. F. (2012).
Comparison of floodplain surface roughness parameters derived
from land cover data and field measurements. Journal of
Hydrology, 452453,139149.
48. Bacopoulos, P., & Hagen, S. C. (2017). The intertidal zones of the
South Atlantic bight and their local and regional influence on astro-
nomic tides. Ocean Modelling, 119,1334.
49. Aquaveo. 2015. SMS 11.2 The complete surface-water solution.
[Accessed online July 10, 2015: http://www.aquaveo.com/
software/sms-surface-water-modeling-system-introduction.]
50. NOAA Tides and Currents. 2014. Tidal datums. [Accessed online
August 2, 2014: http://tidesandcurrents.noaa.gov/.]
51. Bacopoulos, P., & Hagen, S. C. (2014). Dynamic considerations of
sea-level rise with respect to water levels and flooding in
Apalachicola Bay. Journal of Coastal Research, 68,4348.
52. Bilskie, M. V., Hagen, S. C., Medeiros, S. C., & Passeri, D. L.
(2014). Dynamics of sea-level rise and coastal flooding on a chang-
ing landscape. Geophysical Research Letters, 41,927934.
53. Hladik, C., & Alber, M. (2012). Accuracy assessment and correc-
tion of a lidar-derived salt marsh digital elevation model. Remote
Sensing of Environment, 121,224235.
54. Bilskie, M. V., & Hagen, S. C. (2013). Topographic accuracy as-
sessment of bare earth lidar-derived unstructured meshes. Advances
in Water Resources, 52,165177.
55. Zhu, M., Hoctor, T., Volk, M., Frank, K., & Linhoss, A. (2015). The
conservation value of elevation data accuracy and model sophisti-
cation in reserve design under sea-level rise. Ecology and
Evolution, 5,43764388.
Bacopoulos P. et al.
... Regardless of the platform used, or the ecosystem of interest, SampleFreq can be used to measure plant frequency. As an example, vegetation point classification from nadir images using the closely-related program SamplePoint has been completed across many ecosystems with varying plant communities, including grasslands, temperate and tropical forests, alpine tundra, salt marshes and deserts (Guo et al., 2016, Parrish et al., 2017, Skipper et al., 2013, Goonan et al., 2009, Bacopoulos et al., 2018,Tabeni et al., 2014 as well as agricultural settings (Nielsen et al., 2015). We therefore think it reasonable to anticipate successful use of SampleFreq in these ecosystems, with the caveats that image resolution must be adequate for identification of the species of interest, and that the vegetation canopy allows determination of where a plant is rooted. ...
Article
As our understanding of ecological systems grows, natural resource management becomes ever more dependent on timely, accurate, and inexpensively-collected monitoring data to support management decisions. Vegetation cover, density, and frequency are abundance metrics used in resource management; however, frequency data can be collected more quickly than density data and with more repeatability and less sensitivity to inter- and intra-seasonal variation in plant morphology. Moreover, frequency is perhaps the best method for monitoring invasive species across extensive areas. A limitation to the use of frequency data is that plot size affects frequency. The optimal plot size is one that yields measurements suitably removed from 0 or 100% to allow detection of both upward and downward frequency trends, yet the optimum plot size cannot be known before sampling. We addressed this conundrum by developing SampleFreq software that facilitates frequency measurements from digital nadir images of any scale with up to 10 nested plot sizes within the confines of the image dimensions. We conducted accuracy and agreement tests of the software using both artificial populations and field plots. Using artificial population plots, accuracy across all users was 93.4% with a repeatability coefficient of 1.4%, indicating high precision. In a field test, SampleFreq and standard field data averaged a 3.4% difference, and were within approximately 10.5% of each other 95% of the time. From the same field test, SampleFreq repeatability coefficient was 6.7%, while the field method was 4.3%, illustrating that both methods have relatively high precision. Because SampleFreq has high potential accuracy, high agreement with field data, and high precision across a range of users, we recommend using SampleFreq with nadir digital images as a suitable alternative method for monitoring plant frequency.
... Models used for forecasting marsh elevation's dyna mic response to SLR all require accurate elevation as an initial condition. These models include the Marsh Equilibrium Model (MEM; Morris & Bowden 1986, Morris et al. 2002, Morris 2006, MEM variations that include more detailed spatial analyses (Schile et al. 2014), Hydro-MEM (Alizad et al. 2016, Bacopoulos et al. 2019, Wetland Accretion Rate Model for Ecosystem Resilience (WARMER; Swanson et al. 2014, Thorne et al. 2018), models by Kirwan and Mudd (Kirwan & Mudd 2012), and the SLR Affecting Marshes Model (SLAMM; Park et al. 1989). Commonly used pointbased models such as the MEM, WARMER, and Kirwan-Mudd models also require information on plant tolerance to flooding, for which relative position in the tidal range is used as a proxy (Janousek et al. 2019). ...
Article
Full-text available
Elevation is a major driver of plant ecology and sediment dynamics in tidal wetlands, so accurate and precise spatial data are essential for assessing wetland vulnerability to sea-level rise and making forecasts. We performed survey-grade elevation and vegetation surveys of the Global Change Research Wetland, a brackish microtidal wetland in the Chesapeake Bay estuary, Maryland (USA), to both intercompare unbiased digital elevation model (DEM) creation techniques and to describe niche partitioning of several common tidal wetland plant species. We identified a tradeoff between scalability and performance in creating unbiased DEMs, with more data-intensive methods such as kriging performing better than 3 more scalable methods involving post-processing of light detection and ranging (LiDAR)-based DEMs. The LiDAR Elevation Correction with Normalized Difference Vegetation Index (LEAN) method provided a compromise between scalability and performance, although it underpredicted variability in elevation. In areas where native plants dominated, the sedge Schoenoplectus americanus occupied more frequently flooded areas (median: 0.22, 95% range: 0.09 to 0.31 m relative to North America Vertical Datum of 1988 [NAVD88]) and the grass Spartina patens , less frequently flooded (0.27, 0.1 to 0.35 m NAVD88). Non-native Phragmites australis dominated at lower elevations more than the native graminoids, but had a wide flooding tolerance, encompassing both their ranges (0.19, -0.05 to 0.36 m NAVD88). The native shrub Iva frutescens also dominated at lower elevations (0.20, 0.04 to 0.30 m NAVD88), despite being previously described as a high marsh species. These analyses not only provide valuable context for the temporally rich but spatially restricted data collected at a single well-studied site, but also provide broad insight into mapping techniques and species zonation.
Article
Full-text available
The Guana-Tolomato-Matanzas (GTM) system is a well-flushed estuary in Northeastern Florida, USA, and characterized as having an extraordinarily high abundance of oysters that resembles the populations described by Euro-American settlers. Historically, dense populations of oysters, such as those found in GTM, are believed to play an important role in water filtration; however, most research teams seeking to simulate this role have not had access to such robust populations to parametrize their models. To quantify the filtration service (FS) of Eastern oysters (Crassostrea virginica) in GTM at several spatial scales (i.e., reef, watershed, estuary), we implemented a model that solves for the hydrodynamics and depletion of particulate matter passing over model oyster populations, the latter of which were derived from detailed bay-wide surveys. The model results suggested that oyster reefs populating the GTM play an important role in water quality by filtering ~60% of the estuary’s volume within its residence time. Our approach teases apart the role of reef size, residence time, particle concentration, and other physical factors on the generation of FS at different spatial scales. Downstream effects were found to be very important for estuary FS, which depend on the spatial distribution of the reefs in the GTM and local and estuarine-scale hydrodynamics. Therefore, the difference between “realized” FS and the “potential” FS of a given reef may be substantial when considering the complex hydrodynamic and connectivity among populations at several scales. Our model results provide clear and actionable information for management of these oyster populations and conservation of their ecosystem services.
Article
Full-text available
Coastal wetlands are likely to lose productivity under increasing rates of sea-level rise. This study assessed a fluvial estuarine salt marsh system using the Hydro-MEM under four sea-level rise scenarios. The Hydro-MEM was developed to apply the dynamics of sea-level rise as well as capture the effects associated with the rate of sea-level rise in the simulation. Additionally, the model uses constants derived from a two-year bioassay in the Apalachicola marsh system. In order to increase accuracy, the lidar-based marsh platform topography was adjusted using Real Time Kinematic survey data. A river inflow boundary condition was also imposed to simulate freshwater flows from the watershed. The biomass density results produced by the Hydro-MEM were validated with satellite imagery. The results of the Hydro-MEM simulations showed greater variation of water levels in the low (20 cm) and intermediate-low (50 cm) sea-level rise scenarios and lower variation with an extended bay under higher sea level rise scenarios. The low sea-level rise scenario increased biomass density in some regions and created a more uniform marsh platform in others. Under intermediate low sea-level rise scenario, more flooded area and lower marsh productivity were projected. Higher sea-level rise scenarios resulted in complete inundation of marsh areas with fringe migration of wetlands to higher land. This study demonstrated the capability of Hydro-MEM to simulate coupled physical / biological processes across a large estuarine system with the ability to project marsh migration regions and produce results that can aid in coastal resource management, monitoring and restoration efforts.
Article
Full-text available
The balance between organic matter production and decay determines how fast coastal wetlands accumulate soil organic matter. Despite the importance of soil organic matter accumulation rates in influencing marsh elevation and resistance to sea-level rise, relatively little is known about how decomposition rates will respond to sea-level rise. Here, we estimate the sensitivity of decomposition to flooding by measuring rates of decay in 87 bags filled with milled sedge peat, including soil organic matter, roots and rhizomes. Experiments were located in field-based mesocosms along 3 mesohaline tributaries of the Chesapeake Bay. Mesocosm elevations were manipulated to influence the duration of tidal inundation. Although we found no significant influence of inundation on decay rate when bags from all study sites were analyzed together, decay rates at two of the sites increased with greater flooding. These findings suggest that flooding may enhance organic matter decay rates even in water-logged soils, but that the overall influence of flooding is minor. Our experiments suggest that sea-level rise will not accelerate rates of peat accumulation by slowing the rate of soil organic matter decay. Consequently, marshes will require enhanced organic matter productivity or mineral sediment deposition to survive accelerating sea-level rise.
Article
Full-text available
During the past century, human modification of environmental systems has greatly accelerated tidal salt marsh deterioration and shoreline retreat in many coastal regions worldwide. As a result, more than 50% of the original tidal salt marsh habitat in the U.S. has been lost. Numerous human activities have contributed directly or indirectly to wetland loss and alteration at local, regional, and global scales. Human impacts at the local scale include those that directly modify or destroy salt marsh habitat such as dredging, spoil dumping, grid ditching, canal cutting, leveeing, and salt hay farming. Indirect impacts, which can be even more significant, typically are those that interfere with normal tidal flooding of the marsh surface, alter wetlands drainage, and reduce mineral sediment inputs and marsh vertical accretion rates. These impacts usually develop over a greater period of time. At the regional scale, subsidence caused by subsurface withdrawal of groundwater, oil, and gas has submerged and eliminated hundreds of square kilometers of salt marsh habitat in the Chesapeake Bay, San Francisco Bay, and Gulf of Mexico. At the global scale, atmospheric warming due to increased burden of anthropogenic greenhouse gases and tropospheric sulfate aerosols appears to be strongly coupled to glacial melting, thermal expansion of ocean waters, and eustatic sea-level rise. Changes in coastal water levels ascribable to eustatic sea-level rise pose a long-term threat to the stability and viability of these critically important coastal systems.
Article
Full-text available
Many studies have explored the value of using more sophisticated coastal impact models and higher resolution elevation data in sea-level rise (SLR) adaptation planning. However, we know little about to what extent the improved models and data could actually lead to better conservation outcomes under SLR. This is important to know because high-resolution data are likely to not be available in some data-poor coastal areas in the world and running more complicated coastal impact models is relatively time-consuming, expensive, and requires assistance by qualified experts and technicians. We address this research question in the context of identifying conservation priorities in response to SLR. Specifically, we investigated the conservation value of using more accurate light detection and ranging (Lidar)-based digital elevation data and process-based coastal land-cover change models (Sea Level Affecting Marshes Model, SLAMM) to identify conservation priorities versus simple “bathtub” models based on the relatively coarse National Elevation Dataset (NED) in a coastal region of northeast Florida. We compared conservation outcomes identified by reserve design software (Zonation) using three different model dataset combinations (Bathtub–NED, Bathtub–Lidar, and SLAMM–Lidar). The comparisons show that the conservation priorities are significantly different with different combinations of coastal impact models and elevation dataset inputs. The research suggests that it is valuable to invest in more accurate coastal impact models and elevation datasets in SLR adaptive conservation planning because this model–dataset combination could improve conservation outcomes under SLR. Less accurate coastal impact models, including ones created using coarser Digital Elevation Model (DEM) data can still be useful when better data and models are not available or feasible, but results need to be appropriately assessed and communicated. A future research priority is to investigate how conservation priorities may vary among different SLR scenarios when different combinations of model-data inputs are used.
Article
Astronomical tides in the South Atlantic Bight are simulated with a fine-resolution (down to ∼10 m), shallow-water equations, finite element model that fully represents the contiguous geometry of the system, including the: width- and depth-variable continental shelf; inlet-punctuated coastline; and riverine and intertidal character of the estuaries. Tidal levels are analyzed for the entirety of the South Atlantic Bight which produces highly detailed maps (resolution of 10–100 m) of tidal datums (MLW—mean low water, MHW—mean high water) throughout the estuarine rivers and intertidal zones. Model skill (performance) when evaluated over 142 gaging stations is R² = 92% for tidal datums (MLW and MHW) and less than 10% error for the full astronomical tide signal inside the estuaries, and within 0.01 m s–1 error for M2 shelf velocities. Tidal analysis reveals a sensitivity of the M2-resonant shelf circulation in the South Atlantic Bight with respect to the tidal inlets, estuarine rivers and intertidal zones, primarily from the Florida/Georgia border to Winyah Bay (South Carolina). The inlets generate an ‘openness’ the South Atlantic Bight coastline, but more important are the geometric-dynamic influences of the estuarine rivers (the cause-effect being enhanced resonance due to extended effective shelf width) and intertidal zones (the cause-effect being tidal decay due to energy dissipation). The riverine and intertidal features of the coastline subtly change the mode of tidal propagation over the continental shelf. Dynamically, the standing wave behavior (resonance) of astronomical tides in the South Atlantic Bight is a function of the shelf and coastline geometries. Modeling and assessment of coastal and shelf circulation should consider the domain as a continuum, including high-resolution definition of the coast's estuaries and intertidal zones.
Article
Theory suggests that gradual environmental change may erode the resilience of ecosystems and increase their susceptibility to critical transitions. This notion has received a lot of attention in ecology in recent decades. An important question receiving far less attention is whether ecosystems can cope with the rapid environmental changes currently imposed. The importance of this question was recently highlighted by model studies showing that elevated rates of change may trigger critical transitions, whereas slow environmental change would not. This paper aims to provide a mechanistic understanding of these rate-induced critical transitions to facilitate identification of rate sensitive ecosystems. Analysis of rate sensitive ecological models is challenging, but we demonstrate how rate-induced transitions in an elementary model can still be understood. Our analyses reveal that rate-induced transitions 1) occur if the rate of environmental change is high compared to the response rate of ecosystems, 2) are driven by rates, rather than magnitudes, of change and 3) occur once a critical rate of change is exceeded. Disentangling rate-induced transitions from classical transitions in observations would be challenging. However, common features of rate-sensitive models suggest that ecosystems with coupled fast–slow dynamics, exhibiting repetitive catastrophic shifts or displaying periodic spatial patterns are more likely to be rate sensitive. Our findings are supported by experimental studies showing rate-dependent outcomes. Rate sensitivity of models suggests that the common definition of ecological resilience is not suitable for a subset of real ecosystems and that formulating limits to magnitudes of change may not always safeguard against ecosystem degradation.