The potential for sea-level-rise-induced barrier island loss: Insights from the Chandeleur Islands, Louisiana, USA

    The Potential for Sea-Level-Rise-Induced Barrier Island Loss: Insights from the Chandeleur Islands, Louisiana, USA Laura J. Moore, Kiki Patsch, Jeffrey H. List, S. Jeffress Williams PII: DOI: Reference:

S0025-3227(14)00179-0 doi: 10.1016/j.margeo.2014.05.022 MARGO 5121

To appear in:

Marine Geology

Received date: Revised date: Accepted date:

13 July 2013 10 May 2014 30 May 2014

Please cite this article as: Moore, Laura J., Patsch, Kiki, List, Jeffrey H., Jeffress Williams, S., The Potential for Sea-Level-Rise-Induced Barrier Island Loss: Insights from the Chandeleur Islands, Louisiana, USA, Marine Geology (2014), doi: 10.1016/j.margeo.2014.05.022

This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

ACCEPTED MANUSCRIPT The Potential for Sea-Level-Rise-Induced Barrier Island Loss: Insights from the

PT

Chandeleur Islands, Louisiana, USA

RI

Laura J. Moore (Corresponding Author)

Department of Geological Sciences, University of North Carolina- Chapel Hill, 104 South Road,

NU

SC

Chapel Hill, NC, [email protected], 919-962-5960

Kiki Patsch

MA

Department of Environmental Sciences, University of Virginia, 291 McCormick Road,

Jeffrey H. List

TE

D

Charlottesville, VA 22904, [email protected],

AC CE P

USGS Woods Hole Science Center, 384 Woods Hole Road, Woods Hole, MA 02543, [email protected], 508-457-2343

S. Jeffress Williams

USGS Woods Hole Science Center, 384 Woods Hole Road, Woods Hole, MA 02543, [email protected], 508-243-2383

1

ACCEPTED MANUSCRIPT

Abstract

PT

As sea-level rises and hurricanes become more intense, barrier islands around the world become

RI

increasingly vulnerable to conversion from self-sustaining migrating landforms to submerging or subaqueous sand bodies. To explore the mechanism by which such state changes occur and to

SC

assess the factors leading to island disintegration, we develop a suite of numerical simulations

NU

for the Chandeleur Islands in Louisiana, U.S.A., which appear to be on the verge of this transition. Our results suggest that the Chandeleurs are likely poised to change state, leading to

MA

their demise, within decades depending on future storm history. Contributing factors include high rates of relative sea level rise, limited sediment supply, muddy substrate, current island

TE

D

position relative to former Mississippi River distributary channels, and the effects of changes in island morphology on sediment transport pathways. Although deltaic barrier islands are most

AC CE P

sensitive to disintegration because of their muddy substrate, the importance of relative sea level rise rate in determining the timing of threshold crossing suggests that the conceptual models for deltaic barrier island formation and disintegration may apply more broadly in the future.

1. Introduction

Barrier islands are found on every continent except Antarctica and represent twelve percent of the world’s coastline. As low-lying features, these landforms are dynamic over a range of temporal and spatial scales, and they are vulnerable to changing conditions. As sea level rises, storm activity increases or sediment-supply rates decrease, a barrier island will respond by migrating landward across the underlying substrate to higher elevations or by drowning

2

ACCEPTED MANUSCRIPT transforming into a subaqueous shoal [e.g., Penland et al., 1988] if there is no longer sufficient sand volume and relief above sea level to prevent repeated inundation during storms.

PT

The conversion of landward migrating deltaic barrier islands into inner-shelf shoals is a well-

RI

recognized geomorphic threshold crossing, or state change—Penland et al. [1988] describe three different progressive states for islands that form at the edge of delta lobes:1) barrier attached to

SC

an erosional headland, 2) transgressive barrier island arc and 3) inner-shelf shoal. Trinity Shoal,

NU

Outer Shoal and Ship Shoal, west of the modern Mississippi River delta, are examples of the transition from the second to the third state [Penland, et al., 1988]. Given future predictions of

MA

accelerating sea level rise [e.g., IPCC, 2007] and increases in hurricane intensity [e.g., Knutson et al., 2005], the transition from landward-migrating barrier to inner-shelf shoal may no longer

TE

D

be restricted to sand-starved deltaic barrier islands. To better understand the transition, we use the Chandeleur Islands in southeastern

AC CE P

Louisiana U.S.A.—where changes have been occurring recently—and a modeling approach to suggest a mechanism for the island-shoal transition and to assess the likely timing of, and factors most influencing, island loss. To do this we develop a suite of model simulations to explore a comprehensive range of empirically-based geologic constraints and input parameters. Based on our testing of different combinations of these constraints and parameters we develop a “mostplausible” simulation, which we use to simulate barrier island behavior. From this simulation, we produce threshold crossing estimates that are most likely (of the simulations presented here) to represent future conditions. To account for uncertainties in parameter estimates, we use six additional simulations to consider the broader range of barrier island behavior that may occur. By carrying out additional simulations, we are also able to assess which factors are likely to be most important in determining the timing of threshold crossing.

3

ACCEPTED MANUSCRIPT Previous work on this topic considered the evolution of subaerial morphology using aerial photo and subaerial lidar surveys [e.g., Sallenger et al., 2009]. In a complimentary effort, here

PT

we model recent change in the context of longer-term evolution, and account for erosion of the

RI

entire shoreface profile and removal of sand through alongshore and cross-shore transport. As a result of our analysis, we also constrain the range of possible values for geologic variables

SC

important to the evolution of the Chandeleur Islands, allowing us to present a plausible scenario

NU

for past and potential future evolution of the northern portion of the barrier system.

MA

1.1 Study Area

The Chandeleur Islands (80 km-long, and north-south trending (Figure 1), formed near

TE

D

the distal end of the St. Bernard Delta lobe following abandonment by the Mississippi River approximately 2000 years ago [e.g., Penland et al., 1985; Twichell et al., 2009b]. At present,

AC CE P

North Chandeleur Island comprises the northern two-thirds of the island system while the southern one-third consists primarily of smaller, even more dynamic islands. Estimates of average relative sea level rise rates (RSLRR) in the region vary from a late Holocene (600 – 1600 AD) rate, compiled to include subsidence for a location northwest of the Chandeleur Islands, of 1.5-2 mm/yr (see section 3.1 for details) to mid-20th century estimates (from tide gauges) on the St. Bernard Delta Plain of ~10 mm/yr [Penland and Ramsey, 1990]. It is unclear whether the higher recent rate represents local variations (such as subsidence) or recent accelerations in SLRR. In response to rapidly rising sea level and subsidence, the Chandeleur Islands have evolved via the process of transgressive submergence, moving landward [e.g., Twichell, 2009a]and extending laterally [e.g., Miner et al., 2009] to build north and south flanks via alongshore

4

ACCEPTED MANUSCRIPT sediment transport processes [Penland, et al., 1988]. It is the alongshore, or lateral, transport of sand that drives the cross-shore, landward migration of the Chandeleur Islands [Miner et al.,

PT

2009]. Cross-shore, landward migration, in turn, liberates sediment from the substrate [e.g.,

RI

Penland et al., 1988; Miner et al., 2009; Twichell et al., 2009a; Moore et al., 2010] thereby supplying sand to replenish the loss of sand from the island due to lateral transport and overwash

SC

processes, the former of which has recently been shown to dominate over the latter [Miner et al.,

NU

2009]. Given the large volume of sand that has been deposited at the northern and southern flanks of the Chandeleurs [e.g., Miner et al., 2009 and Twichell et al., 2009], considerable

MA

landward migration and associated shoreface erosion must have occurred over long time scales, at least along the central portion of the island chain, which (given the high rates of lateral

TE

D

transport and the importance of lateral transport in island evolution) must pre-date the flanks which formed later as the coastline straightened and elongated. Because the muddy shoreface

AC CE P

associated with deltaic islands yields a small percentage of sand (~30%) [e.g., List et al., 1997] when eroded, these barriers are more vulnerable to geomorphic threshold crossing than some barriers on the U.S. Atlantic Coast (e.g., North Carolina Outer Banks) [Moore et al., 2010; Moore et al., 2011], needing to migrate farther landward than barriers resting on sandier substrates, to yield sufficient sand to maintain island position above sea level in the face of alongshore losses. North Chandeleur Island has a marsh core (as do many barrier islands) that pre-historically provided a platform for the capture of sand transported landward during storms and a nucleation site for sand delivered to the front of the island during periods of accretion between storms (e.g., Miner et al., 2009; Sallenger et al., 2009]. Historical maps suggest that prior to the 1920s the back-barrier shoreline was migrating landward (i.e., ‘rolling over’) through marsh colonization

5

ACCEPTED MANUSCRIPT of overwash deposits, but that by the 1920s marsh progradation into Chandeleur Sound (and therefore overwash deposition and island rollover) largely ceased [McBride et al., 1992; Miner

PT

et al., 2009]. This results in a narrowing of the island system as the shoreface continues to

RI

transgress. While the processes responsible for this change in barrier island response are not clear, it likely increases vulnerability to threshold crossing by increasing the likelihood of

SC

inundation during storms. Observational evidence suggests that during inundation large amounts

NU

of sand are carried across the island and into the bay behind the barrier instead of being captured

MA

in the back-barrier as occurs during overwash conditions [e.g., Sallenger, 2000].

1.2 Threshold Crossing Definition

TE

D

By 2005, the Chandeleur Islands were in a depleted state due to strong storms occurring in the previous decade. During and following Hurricanes Katrina and Rita in 2005, Sallenger et al.

AC CE P

[2009] document extensive, rapid erosion of the marsh platform that further increases the potential for island loss by requiring additional inputs of sand to maintain a subaerial island [Sallenger et al., 2009]. This idea is supported by the role of mass conservation in barrier island evolution [e.g., Moore et al., 2010] because when the marsh platform disappears both a volume of material that would otherwise eventually underlie the transgressing island, as well as the sandtrapping properties of the marsh, disappear with it. Because observations and mass conservation suggest that the presence of marsh platform is critical to island stability, we consider a geomorphic threshold crossing (i.e., a change in state from subaerial to submarine) to be initiated when the shoreface along the majority of the island has migrated far enough landward to erode the last remnants of marsh core (i.e., when the shoreface intersects the shoreline on the mainland side of the barrier, hereafter referred to as the bay-side shoreline). Central to this definition is the

6

ACCEPTED MANUSCRIPT observation that starting ~1920 AD marshes in the Chandeleur Islands ceased migrating

PT

landward.

RI

2. Methods 2.1 Modeling Framework and Approach

SC

Morphological-behavior models, which are driven by changes in sediment supply, sea-level

NU

rise, and shoreface geometry [e.g., Cowell et al., 1995; Stolper et al., 2005; Moore et al., 2010], without simulating the detailed physical processes of sediment transport, provide a means for

MA

testing whether hypotheses regarding barrier island evolution are geometrically possible. Though these models do not simulate island evolution at the scale of individual storm events and

TE

D

therefore cannot directly simulate changes from one equilibrium state to another, they can be used to assess island migration and location from which we can infer the occurrence and timing

AC CE P

of state change [Moore et al., 2010, Moore et al., 2011]. GEOMBEST (Geomorphic Model of Barrier, Estuarine and Shoreface Translations), [Stolper et al., 2005; Moore et al., 2007; Moore et al., 2010], is a 2-dimensional, cross-shore, numerical morphological-behavior model that simulates the evolution of barrier island morphology and stratigraphy over time scales ranging from decades to millennia, including the important roles of both cross-shore and alongshore sediment transport. Basic model formulation and inputs are described here but the reader is referred to Moore et al., [2010] for more details. Model formulation is based on sediment conservation principles expressed in the continuity equation (all grain sizes can be eroded in the model but only the sand fraction is conserved) and behavior rules originating from the concept that, given appropriate conditions and sufficiently long time scales, a shoreface and barrier profile will tend to remain invariant. Similarly, model

7

ACCEPTED MANUSCRIPT formulation is based on the assumption that storm characteristics, and therefore integrated barrier island storm response (i.e., subaerial and backbarrier morphologies), remains approximately

PT

constant over sufficiently long times scales. Thus, we specify an equilibrium morphology, based

RI

on the modern profile, extending from the base of the shoreface across the subaerial barrier and marsh, and we assume that this profile is the equilibrium profile, i.e., it represents the profile

SC

form that the model will tend to evolve toward during simulations. In support of this

NU

assumption, shoreface profiles from the central portion of the Chandeleur Islands [Miner et al., 2009] appear to have consistent form since the time of the oldest profile in 1870 AD. In

MA

addition, younger abandoned lobes of the Mississippi River delta have been shown to maintain a nearly constant form despite more than 100 years of evolution and up to 1 km of shoreline

TE

D

erosion [List et al., 1997]. However, since it cannot be proven that the modern Chandeleurs shoreface profile represents the equilibrium profile through the entire simulation period, this

AC CE P

represents a source of uncertainty in model simulations. The model domain consists of a cross-shore grid of user-specified cell size in the horizontal (typically on the order of 100 m) and vertical (typically less than 1 m), extending from the base of the shoreface (or deeper) to the mainland and encompassing the shoreface, barrier and bay realms, i.e., the coastal tract (Figure 2). After testing to insure that smaller grid sizes did not change simulation results by more than 5%, we settled on a grid size of 125 m (horizontal) by 0.2 m (vertical) for all simulations reported here. Morphological evolution is driven by differences between the coastal tract surface and the user-specified equilibrium morphology, which maintains its vertical position relative to sea level throughout a simulation. With each time step (typically 10 – 50 years) the equilibrium morphology shifts vertically as sea-level changes and then moves horizontally to the cross-shore position that conserves sand (i.e., sand liberated by

8

ACCEPTED MANUSCRIPT shoreface erosion balances sand deposited on and behind the island). A sand loss from alongshore transport gradients (representing, for the Chandeleur Islands, the transport of sand to

PT

the flanks, where the island is building laterally) tends to make the coastal tract surface move

RI

landward (which produces sediment from shoreface erosion to feed this loss). Achieving sand conservation within the existing geometric and stratigraphic framework may also require

SC

adjustments to the morphology and underlying barrier stratigraphy (e.g., changes in barrier island

NU

height relative to the bay and/or changes in island volume).

GEOMBEST includes input parameters describing how quickly each stratigraphic unit can

MA

erode (i.e., erodibility) and what proportion of the sediment in each stratigraphic layer is sand as opposed to mud (i.e., substrate composition). GEOMBEST does not address heterogeneity of

TE

D

sand distribution within stratigraphic units since the smooth, arcuate ocean-facing shape of most barrier islands (including the Chandeleurs) indicates that any existing alongshore differences in

AC CE P

sand production are ultimately homogenized by alongshore processes. Note that if alongshore transport did not tend to keep shorelines smooth (e.g., Komar, 1976; Ashton and Murray 2006) in the presence of alongshore heterogeneities in sand supply (e.g., Valvo et al., 2006), the shoreline would become increasingly sinuous through time, bulging seaward where local sand supply is greatest (e.g., where distributary channels are present), which is not what is observed in these systems. A depth-dependent response rate (DDRR) allows specification of the maximum rate at which the shoreface can erode or accrete vertically as a function of depth, representing a decrease in near-bed wave energy with increasing depth. This depth-defined parameter works in concert with the stratigraphic-unit-specific erodibility parameter to determine the maximum amount of sand that can be eroded or deposited within each grid cell at each time step, though Moore et al [2010] demonstrate that barrier evolution is relatively insensitive to these parameters

9

ACCEPTED MANUSCRIPT except in the case of an unreasonably low DDRR and a nearly non-erodible substrate. Specification of these parameters allows simulation of complex geological scenarios for which

PT

the assumption of an equilibrium profile is relaxed. For example, the specified morphology will

RI

not be attained if the substrate is nonerodible or if sea-level rises too rapidly for the substrate to achieve the specified shape in the allotted time. In these situations, other geometric adjustments

SC

(e.g., increases in landward migration rate or decreases in barrier island volume) will occur to

NU

create a coastal tract surface, and underlying barrier stratigraphy, that achieves sand

MA

conservation.

2.2. Input Parameters

TE

D

Although the Chandeleur Islands have been relatively well studied [e.g., Lavoie, 2009], uncertainty remains in our understanding of geologic variables such as the late-Holocene relative

AC CE P

sea-level rise rate (RSLRR), the bay sedimentation rate (i.e., the rate of fine-grained sedimentation in Chandeleur Sound), shoreface depth, substrate sand proportions, and the timing and importance to barrier evolution of initial adjustments from a river-dominated delta profile to a wave-dominated shoreface profile after delta abandonment. These variables provide important constraints that affect how an island evolves both in reality and in model simulations. Also unknown is how soon after delta lobe abandonment the islands formed. Given that we seek to simulate island evolution after formation rather than island formation itself, we begin our simulations 1500 ybp based on an estimate that it took ~ 500 years [Penland, personal communication, 2007] following delta lobe abandonment 2000 ybp [e.g., Penland et al., 1985; Twichell et al., 2009b] for the prograding deltaic headland to develop into a transgressive barrier island arc, passing through stage 1—“erosional headland with flanking barriers” [Penland et al.,

10

ACCEPTED MANUSCRIPT 1988]—along the way. Because the northern and southern portions of the Chandeleur Islands have been elongating via lateral transport [e.g., Miner et al., 2009] and must therefore be

PT

younger than the central portion of the island that originally evolved from the erosional

RI

headland, it is the older, center portion of the island that we simulate here. Throughout the remainder of this section we describe how we used the best available geologic data to develop

SC

the most reasonable range of values that can be devised for constraints and input parameters

NU

given the limited empirical bases available. In Section 2.3 we discuss how we then use these empirically-derived reasonable ranges of parameter values to address uncertainties in the

MA

parameter values themselves.

TE

D

In the model realm, input parameters such as substrate sand proportion, sand-loss rate and shoreface depth determine how much sand is liberated at each time step while RSLRR, bay

AC CE P

sedimentation rate and the initial shelf slope are combined spatially before running the model to develop an initial condition. For example, if relative sea level is to rise 2 mm/yr for 1500 years, sea level at the start of the simulation must be 3 m below modern sea level. Similarly, in order to accumulate 2 m of bay sedimentation over 1500 years, accommodation space in the form of a sound must exist behind the barrier at the start of the simulation in the cross-shore position where the sedimentary deposit is to be located at the end of the simulation. To be viable, therefore, a set of geologic constraints and input parameters must both 1) suggest a reasonable initial condition that is geometrically possible and 2) be able to reproduce the observed historical morphology and stratigraphy from that initial condition. 2.2.1 Relative Sea Level Rise Rates Relative sea level rise rate (RSLRR) estimates for the Chandeleur Islands region vary

11

ACCEPTED MANUSCRIPT considerably and exceed eustatic SLRR because of local subsidence. Using basal peat samples that formed directly on Pleistocene basement west of the Mississippi River delta (and which

PT

therefore track the glaco-isostatic component of SLR) Gonzalez and Tornqvist [2006] found

RI

RSLRR for the period 600 - 1600 AD to be 0.55 mm/yr. Compaction-associated subsidence of Holocene strata in the vicinity of the Chandeleur Islands has not been quantified but given a

SC

Holocene thickness of ~40 – 60 m [e.g., Kolb and van Lopik, 1966; Frazier, 1967; Penland et.

NU

al., 1988] and the relationship between thickness and compaction rate from Törnqvist et al. [2008] local subsidence likely adds another 1- 1.5 mm/yr of RSLR bringing the estimated total

MA

RSLRR to approximately 1.5 – 2 mm/yr. Interpolating between RSLRR derived from NOAA (National Oceanic and Atmospheric Administration) National Ocean Service tide gauge data

TE

D

west of the Chandeleur Islands at Grand Isle, Louisiana (LA) of 9.9 mm/yr (50-year record) and to the east (a nearly equal distance) at Dauphin Island, Alabama of 2.9 mm/yr (39-year record)

AC CE P

[Zervas 2004] yields a modern relative sea-level rise rate estimate for the Chandeleur Islands of 6.4 mm/yr. Penland and Ramsey [1990] arrived at a higher mid-20th century RSLRR of ~10 mm/yr based on tide gauge records located on the St. Bernard Delta Plain. In contrast, sediment cores and seismic data reveal that the top of the St. Bernard Delta deposit lies 5 – 6 m below present sea level in the vicinity of the Chandeleur Islands [Twichell et al., 2011]—assuming the delta was abandoned 2000 years ago [Penland et al., 1985] this evidence suggests an average late-Holocene RSLR rate of 2.5-3 mm/yr [Twichell, personal communication]. Although we do not know what the rates of RSLR were immediately following abandonment, it is possible that they exceeded modern rates because rates of subsidence have the potential to be highest immediately following abandonment.

12

ACCEPTED MANUSCRIPT In summary, the compiled range of RSLR rate estimates from all sources described above is ~1.5 – 10 mm/yr, which, extended into the past, results in an estimated total RSLR since 1500

PT

ybp of 3.75 to 15 m. Due to the impossibility of testing all values, because there are known but

RI

unquantified local tectonic and compaction gradients, and assuming that high rates of RSLR post-abandonment (e.g., 10 mm/yr) were unlikely to be sustained, we tested RSLR amounts

SC

(over the last 1500 years) of 3 m, 5 m, and 7.5 m (Table 1), which correspond to average RSLR

NU

rates of 2 mm/yr, 3.3 mm/yr and 5 mm/yr, respectively.

MA

2.2.2 Bay Sedimentation Rate

Based on core records from Brooks et al. [1995] there appears to be 0.6 to 3.5 m of marsh

TE

D

and/or bay sedimentation deposits overlying the distributary and delta front facies on the landward side of North Chandeleur Island. Immediately south of the study area, a more

AC CE P

continuous estuarine/lagoonal layer is identified as having a thickness of up to 3.6 m [Brooks et al., 1995; Kolb and Van Lopik, 1966]. Recent work suggests that the marsh/bay sedimentation layer in our study area may be as thin as 1.5 m [Flocks et al., 2009]. However, from core logs it is unclear what proportion of these sediments were deposited in association with deltaic marsh prior to abandonment of the St. Bernard Delta versus through bay sedimentation after the formation of the Chandeleur Islands ~ 1500 ybp. To address this and the range of estimates for layer thickness in our initial simulation suite, we test bay sedimentation rates of both 0 mm/yr (i.e., assuming the entire 3 m-thick layer is composed of marsh that existed prior to island migration) and 2 mm/yr (i.e., assuming the entire 3-m thick layer is composed of bay sediments accumulated over the last 1500 years) (Table 1). The “most plausible” simulation includes a bay rate of 1 mm/yr (i.e., 1.5 m of the marsh/bay layer is composed of pre-existing marsh and 1.5 m

13

ACCEPTED MANUSCRIPT accumulates in the bay over the last 1500 years). This latter scenario is consistent with the most

PT

recent stratigraphic interpretations of this layer by Flocks et al. [2009].

RI

2.2.3 Modern Morphology and Stratigraphy

To calculate an average 1920 AD barrier and shelf morphology, we extract, along 11 shore-

SC

perpendicular transects with an alongshore spacing of 1.5 km (study area encompasses 30km of

NU

the island, Figure 1), elevations from a 20 m bathymetric grid constructed from 1920 National Geophysical Data Center (NGDC) sounding data [Miner et al., 2008]. Using these profiles we

MA

calculate an average elevation profile from ~22 km behind the barrier shoreline within Chandeleur Sound to approximately 50 km offshore of the barrier within the Gulf of Mexico.

TE

D

We use all 11 profiles to find an average offshore profile out to a depth of ~34 m. For the Chandeleur Sound portion of the profile, we include only the northern 4 profiles to find the

AC CE P

average profile due to the presence of back-barrier islands within Chandeleur sound in the other profiles. Including profiles with these islands would result in an average profile with unrealistic intermediate sound depth. Note that the presence or absence of these back-barrier islands in the average profile does not affect the results presented here due to the limited degree of barrier island migration simulated, although their presence could affect longer-term forecasts. We added island topography to the average sound and offshore profiles using a representative dune profile from Ritchie et al. [1992]. The resulting “hybrid average” cross-shore profile, extending from Chandeleur Sound to 35 m water depth within the Gulf of Mexico, represents the average 1920’s study area morphology. To provide an average cross-shore stratigraphic section for placement below the 1920s morphology, we averaged, across all 11 cross-shore transects described above, the elevation of

14

ACCEPTED MANUSCRIPT the upper surfaces of stratigraphic units, as derived from gridded files representing analyses of sediment cores and geophysical records [Brooks et al.,1995; Flocks, personal communication;

PT

Flocks et al., 2009]. Because distributary deposits are incised into delta front deposits, these two

RI

stratigraphic units were merged into a combined distributary/delta front layer after averaging. The resulting representative average modern morphology and stratigraphy (Figure 3) serves as

SC

both the recent historical condition (~1920) that we seek to simulate through late-Holocene

NU

simulations and as the basis for development of an initial morphology and stratigraphy that

MA

represents a plausible initial condition 1500 years ago.

2.2.4 Initial Morphology and Stratigraphy

TE

D

In developing the initial suite of simulations we create multiple initial morphologies and stratigraphies to cover the range of possible initial conditions resulting from the combination of

AC CE P

3, 5 and 7.5 m of total RSLR, and 0 and 2 mm/yr of bay sedimentation. Twichell et al. [2009b], demonstrate that the edge of the delta is at least 8 km seaward of the current position of the island (8km is the seaward limit of their data). Consistent with this, and given the expectation that the islands formed at the distal edge of the delta [e.g., Penland et al., 1988], as a first step in synthesizing an initial morphology, we estimate that 1500 years ago the barrier was located approximately 10 km seaward of its present location. This location is also consistent with extrapolation into the past of the historical (1855-2002) average shoreline change rate for North Chandeleur Island of 6.6 m/yr [analysis based on data used in Penland et al., 2005; Penland, personal communication]. While there are no data available to confirm that this rate is representative of the last 1500 years, rates of shoreline erosion from much younger lobes of the Mississippi River delta are similarly high [McBride et al., 1992], supporting the assumption that

15

ACCEPTED MANUSCRIPT the Chandeleurs have been eroding at a high rate since abandonment of the St. Bernard lobe. For this reason, in the absence of data to suggest a different long-term rate, we extrapolate the

PT

historical rate. When the bay sedimentation rate is set to 0 mm/yr the marsh/bay layer is present

RI

at the beginning of the simulation and when the bay sedimentation rate is set to 2 mm/yr all bay sedimentation accumulates during the simulation. Therefore, to create an initial morphology and

SC

stratigraphy, we shift the barrier island horizontally to a location 10 km seaward of the current

NU

barrier position and then adjust the profile surface vertically so that it is located relative to sea level at the start of the simulation. (Note: The surface consists of the marsh/bay unit behind the

MA

barrier for the 0 mm/yr bay sedimentation case and the surface of the distributary/delta front unit behind the barrier for the 2 mm/yr bay sedimentation case, plus the surface of the barrier island

TE

D

and the shoreface in both cases, see Figure 4 A and B vs. C.) Below this morphology, we extrapolate surfaces of the underlying stratigraphic units to meet the initial shoreface or delta

AC CE P

profile, a necessary assumption in the absence of information on sediments already eroded). In an additional series of initial conditions, we account for the possible effects of an initial profile that exhibits a river-dominated (i.e., delta), rather than a wave wave-dominated (i.e., shoreface), shape. Immediately following abandonment, the outer edges of the St. Bernard delta lobe likely exhibited a convex upward shape typical of deltas receiving regular influxes of river sediment [e.g., Friedrichs and Wright, 2004; Wright, 1995]. Because a river-dominated delta slope will be out of equilibrium with wave conditions once sediment is no longer being delivered by fluvial processes, large amounts of sand may be liberated rapidly as the profile adjusts to wave conditions. Because this has the potential to alter barrier island evolution, we create additional initial morphologies exhibiting a river-dominated profile for each combination of total RSLR and bay sedimentation rate (e.g., Figure 4 B and C). [Note that the length of time it takes

16

ACCEPTED MANUSCRIPT for the river-dominated profile to convert to a wave-dominated (i.e., shoreface) profile, is not important here because 1) the observed modern profile clearly demonstrates a wave-dominated

PT

shape [e.g., Miner et al., 2009], indicating that conversion is complete, 2) the only contribution of the shift from river- to wave-dominated is an additional contribution of sand, which, based on

RI

mass conservation, will have the same “end effect” regardless of the timing, and 3) our focus is

SC

on recent island evolution and therefore it is only the “end effects” that matter in this case.]

NU

To develop an appropriate estimated delta profile for the study area 1500 years ago we used the relationship, identified by Friedrich and Wright [2004], between riverine sediment discharge

MA

per unit length along-shelf (Qr), wave height (H90), and wave period (T) and the best available estimates for these parameters in our study area. According to buoy data collected between 1980

TE

D

and 2002, average H90 and T values for the Mississippi River are 1.8 meters and 5.7 seconds respectively [Sasaki and Hibiya, 2007; Sasaki, 2009]. The earliest available estimate for

AC CE P

sediment discharge from the Mississippi River is 400 million metric tons/ year at 1700 AD [Meade, 1995]. Converting this to discharge along a shelf distance of 250 km yields Qr = 0.05 kg/m/sec. Friedrichs and Wright [2004] calculate a range of delta slope profiles for different combinations of the parameters just described. The best available parameter estimates for the Mississippi River at the time the St. Bernard Delta was active fall most closely in the Friedrichs and Wright [2004] category of H90 = 2 m, T = 10 sec, and Qr = 1 kg/m/s. We digitized, from Friedrichs and Wright [2004], the delta profile associated with this category and inserted it (in place of the previously defined shoreface profile) to create a series of initial morphologies having a river-dominated delta profile that reasonably approximates the shelf profile following abandonment.

17

ACCEPTED MANUSCRIPT 2.2.5 Shoreface Depth To account for decreases in near-bed wave energy with depth, we define a depth, i.e., the

PT

shoreface depth, beyond which sediment transport is deemed insignificant. This depth increases

RI

with longer time scales [Nicholls et al., 1998] because stronger, less-frequent storms are more likely to occur as time scales lengthen. Shoreface depth is important in actual and modeled barrier

SC

island behavior because it determines the extent of the shoreface that may supply sand to the barrier

NU

island system via erosional processes. Changes in shoreface depth alone will only affect barrier island behavior if a change in shoreface depth results in a change in average barrier island slope

MA

(because average barrier island slope determines barrier island volume, and barrier island volume, in turn, determines the amount of sand that needs to be removed or added to the barrier to approach

D

equilibrium with the substrate slope) [Moore et al., 2010]. Although some barrier island systems are

TE

not sensitive to changes in shoreface depth due to system geometry [e.g., Moore et al., 2010], the

AC CE P

geometry of the Chandeleur Islands shoreface and barrier profile is such that average barrier island slope does change with shoreface depth (Figure 5), which suggests that shoreface depth will somewhat affect simulation outcome. Only one study addresses shoreface depth within the Chandeleur Islands—Miner et al., [2009] analyzed historical (1870-2007) bathymetric change for the Chandeleur Islands out to a depth of 15 m and found shoreface erosion across our study area during this time period providing evidence for shoreface depth greater than 15 m. For comparison, List et al. [1997] found historical shoreface depths to be on the order of 3-15 m between Isle Dernier and Grand Isle, which are located to the west of the modern Mississippi River delta. Examination of the modern average profile for our study area reveals a relatively smooth shoreface with a gentle break in slope at ~18 m and a more pronounced step at -26 m. Everts [1978] suggests that such changes in offshore morphology are indicative of long-term limits to significant cross-shore transport. Given this range of information

18

ACCEPTED MANUSCRIPT and the importance of shoreface depth in our study area, our initial suite of simulations explores the

PT

viability of shoreface depths ranging from 15 m to 30 m, in 5 m increments (Table 1).

RI

2.2.6 Substrate Composition

We derived values for the percentage of sand-sized sediment in each of the depositional faces

SC

based on vibracore descriptions in Brooks et al. [1995], facies descriptions in Kolb & Van Lopik

NU

[1966], and sample analyses from cores collected recently by the U.S. Geological Survey [Flocks et al., 2009] (Table 1). Given the uncertainty with which core samples accurately

MA

represent the percentage of sand-sized sediment across existing stratigraphic unit, this is one of the least-well constrained parameters. Additionally (and obviously), the true sand content of the

D

stratigraphic layers that were eroded during transgression cannot be known. Thus, sand

TE

percentages were held constant throughout an initial suite of simulations but then increased by up

AC CE P

to 40% to refine initial model simulations to more closely reproduce 1920 AD conditions (see Section 4) (Table 2).

2.2.7 Sand-Loss Rates

To estimate the removal of sand from the study area due to longshore transport, we digitized Figure 3B (a plot of longshore transport volume vs. alongshore distance) from Ellis and Stone [2006] and fit the data in the vicinity of our study area with a linear regression resulting in an estimated sand- loss rate of 2.3 m3/m/yr. Ellis and Stone [2006] conducted their analysis using the wave refraction model, WAVENRG and did not include longshore transport during hurricanes. Using a similar approach, but including hurricane effects, Georgiou and Schindler [2009] used a long-term data set of offshore winds to forecast waves in combination with the Coastal Engineering Research Center (CERC) equation to generate longshore sediment transport 19

ACCEPTED MANUSCRIPT gradient estimates for the Chandeleurs. Estimates for the study area from Georgio and Schindler [2009] are slightly higher than those from Ellis and Stone [2006] ranging up to 6

PT

m3/m/yr for the northern section. To provide yet another estimate of sand-loss rate from the

RI

study area we divided the estimated volume of Hewes Point of 3.79 x 108 m3 [Twichell et al., 2009] by the length of coast it was likely derived from (~20 km) and the time period over which

SC

it accumulated (~2000 years since abandonment) to yield 9.5 m3/m/yr. Given a range of

NU

estimates from 2.3 to 9.5 m3/m/yr, we adopted 5 m3/m/yr as a reasonable long-term estimate for

MA

sand-loss rate in our study area.

2.3. Simulation Development and Testing

TE

D

2.3.1 Late-Holocene Simulations (~1500 ybp to 1920 AD) and Extension to 2005 To address uncertainties in geologic constraints and input parameters for the Chandeleur

AC CE P

Islands, we explored a range of values for four parameters centered on best estimates from the literature as described above in section 3 and summarized as follows: total RSLR (3 m, 5 m, 7.5 m, based on SLR estimates from basal peat samples, cores and seismic data), bay sedimentation rate (0 mm/yr and 2mm/yr, based on cores), shoreface depth (15 m - 30 m in 5 m increments, based on bathymetric change and shoreface morphology), and initial shoreface slope (delta vs. shoreface, based on geologic history) (Table 1). Because data do not exist to do so, we did not vary the initial position of the barrier at 1500 ybp (see Section 2.2.4)—the difference between the actual (unknown) and estimated initial barrier position thus remains an unquantified source of uncertainty in our simulations. To test the viability of combinations of input values derived from the literature we held the sand-loss rate and the sand percentages of stratigraphic layers constant while varying total

20

ACCEPTED MANUSCRIPT RSLR, bay sedimentation rate, outer-shelf profile shape and shoreface depth to include all possible combinations of the input values listed above and shown in Table 1. Consideration of

PT

all possible combinations of these parameters results in 48 possible late-Holocene simulations

RI

(Figure 6), each having a different initial condition. Upon testing, we found most of these combinations of input parameters to be geometrically infeasible either because they dictated a

SC

geometrically infeasible initial condition (i.e., all possible simulations having only 3 m of total

NU

RSLR, Figure 7A; all possible simulations having 7.5 m total RSLR and no bay sedimentation, Figure 7B; dark gray boxes, Figure 6) or because they were not able to reproduce the historical

MA

morphology and stratigraphy (e.g., Figure 8; light gray boxes, Figure 6). Of all 48 combinations of values, only 6 resulted in simulations that closely reproduced 1920 morphology and

TE

D

stratigraphy (Figure 6, boxes without shading, and Figure 9; white boxes, Figure 6). Because initial system geometry (and therefore initial conditions) is determined by RSLR,

AC CE P

bay sedimentation rate and outer-shelf slope, the 6 viable scenarios represent three different initial conditions (Figure 4). In all six viable scenario simulations, the barrier island migrates somewhat farther than it should, suggesting that one or more of the parameters in the model is not sufficiently representative of actual conditions. As shown in Figure 6, we have already explored model sensitivity to the likely (or reasonable) range of sea-level rise, bay sedimentation rate, shoreface slope, and outer-shelf slope. To calibrate the simulations so that the cumulative transgression distance better matches the 1920’s observed shoreline position, we modified each of the 6 viable scenarios by increasing stratigraphic layer sand percentages by 0 - 40% of the original input value (to a maximum value of 95% sand) for each layer as needed to improve agreement between simulated and actual 1920 AD conditions (Table 2, scenarios i-vi). This process results in the following stratigraphic layer sand percentages for simulations i-vi: marsh

21

ACCEPTED MANUSCRIPT 20% to 28%, distributary/delta front 30% to 42%, prodelta 20% to 28%, and sand sheet 70% to 95% sand. Though, these percentages are somewhat higher than the average values indicated by

PT

sediment cores (section 2.2.6), they are still within reason given the sedimentary facies

RI

represented by each stratigraphic layer. Another parameter that could be adjusted to reduce simulated barrier migration distance is the sand-loss rate. However, the decrease in sand-loss rate

SC

required to improve the match between observed and modeled barrier island migration distance

NU

so far exceeds the sand-loss rate estimate from Ellis and Stone [2006] that a sand gain would have to be assumed. Bathymetric change observations [Miner et al., 2009], as well as

MA

geomorphic evolution models [e.g., Penland et al, 1988] support the concept that sand has been continually removed from study area and deposited in flanking spits and shoals, indicating that

TE

D

adjusting the sand-loss rate in this manner is not reasonable. Given the narrower range of plausible input parameters suggested by the suite of 6 viable

AC CE P

simulations, we develop a “most-plausible” 1400-year, late-Holocene simulation (time step = 50 years) having initial conditions and parameter values in the middle of the plausible range. The most-plausible simulation also successfully reproduces the morphology and stratigraphy circa 1920, when the last available bathymetric survey prior to Hurricane Katrina was collected (Figure 10A-C, Table 2). In preparation for assessing island vulnerability to threshold crossing, we extend simulations i-vi and the most-plausible late-Holocene simulation to 2005 AD by adding 85 years in two 42.5-year time steps. As an additional measure of the success of these model simulations in reproducing recent and modern conditions, we compared simulated shoreline change for two model run time periods (1870 - 1920 and 1920 - 2005), with shoreline change for the same time periods measured from historical data. We obtained historical shorelines derived from National Ocean Service (NOS)

22

ACCEPTED MANUSCRIPT T-sheets (1855 and 1922) and a recent shoreline measured from Quickbird Imagery (2005) [Martinez, personal communication and Fearnley et al., 2009] and calculated average shoreline

PT

change rates across the study area using the Digital Shoreline Change Analysis System [Thieler

RI

et al., 2009]. We then used the average shoreline change rates for each time period of observation to estimate average observed shoreline change for 1870 -1920 and 1920 -2005

NU

SC

(Table 3) to correspond with shoreline positions generated in GEOMBEST simulations.

2.3.2 Assessing the Timing of Threshold Crossing

MA

We extend simulations i-vi and the most-plausible simulation beyond 2005 AD (in 15-year time steps) to estimate the time window within which threshold crossing is likely to occur.

TE

D

When extending into the future, we increase the sand-loss rate to account for additional stormrelated losses that are likely given the changes in island configuration that began ~1920 and

AC CE P

continue today. As a result of 20th century narrowing (due to cessation of marsh progradation) as well as lowering and narrowing due to recent storms, [Sallenger et al., 2009] storm inundation is likely both more frequent and associated with higher landward flow velocities thereby resulting in the dispersion of larger amounts of sand from the barrier island system [e.g., Sallenger, 2000] into Chandeleur Sound. In addition, because the modern barrier shoreface is now landward of former distributary channels (underlain instead by the former main trunk channel), which have been a primary source of sand for the islands [Twichell et al., 2009b], shoreface erosion will liberate less sand now than it has in the past. The effects of inundation and sediment dispersion, as well as decreases in sand supply from the shoreface, are best captured in the model by increasing the sand-loss parameter for future time steps.

23

ACCEPTED MANUSCRIPT Because it is not possible at this time to quantify actual increases in sand loss and the timing of these increases, we simulate losses 2 - 4 times greater than the initial estimate (e.g., 10 and 20

PT

m3/m/yr, beginning conservatively in 2005) to capture the effect of uncertainties in this

RI

parameter on estimates for the likely timing of threshold crossing. These sand losses, which in reality would be deposited in Chandeleur Sound, are conserved in GEOMBEST (e.g., yellow-

SC

hatched area in Figure 10D) through backbarrier deposition but do not affect our assessment of

NU

threshold crossing timing, which is defined to occur when the shoreface intersects the position of the 1920 bay-side shoreline (i.e., landward marsh edge), under the assumption that back-barrier

MA

transgression has ceased. While GEOMBEST, as designed, continues to simulate transgression of the entire barrier island form, we assume that this mode of translation is no longer active, as

TE

D

suggested by both long-term shoreline change [McBride et al., 1992] and recent post-hurricane observations [Sallenger et al., 2009]. The increased sand-loss rates accelerate shoreface

AC CE P

migration rates despite the fictional appearance of backbarrier sand capture (yellow hatched region, Figure 10D) in the model graphical output.

4. Results

The average offset between the 1870 and 1920 shorelines and the 1920 and 2005 shorelines is greater for the simulated shorelines by a factor of 2-3 and a factor of ~1 respectively (Table 3). Some of the difference between the modeled and observed shoreline change for 1870 – 1920 may be accounted for by errors associated with measuring historical shoreline positions, which can approach +/- 20 m for historical shoreline change measurements from NOS T-Sheets [e.g., Moore et al., 2000], however this represents only a fraction of the offset. Though we would certainly prefer to see better agreement between model simulations and historical observations

24

ACCEPTED MANUSCRIPT for this early time period, a difference of ~ 300 m for a 50-year time span is not surprising given the model assumption that storm characteristics remain constant throughout the entire simulation

PT

period, i.e., the 1870 or 1920 observed shorelines could have been recently storm-impacted, but

RI

the model cannot account for this. The close level of agreement between the modeled and simulated shorelines for the longer, more recent time period is encouraging and, along with the

SC

close reproduction of 1920 morphology and stratigraphy, suggests that model simulations are

NU

reproducing modern conditions well enough to provide insight on the range of potential future island behavior.

MA

Consistent with recent observations, the modeled 2005 AD shoreface does not intersect the 1920 bay-side shoreline in the most-plausible scenario (Figure 10D) (or in any of the 6 additional

D

simulations). Model results from extension of the most-plausible scenario into the future predict

TE

threshold crossing in 120 and 90 years post-2005 (i.e., in 2125 and 2095) for the 10 m3/m/yr and

AC CE P

20 m3/m/yr sand-loss rate scenarios, respectively (Figure 10D, red cross-hair). The corresponding 6 additional simulations bracket this estimate and predict threshold crossing within 30 to 135 years post-2005 AD (i.e., in 2035 to 2140 AD) (Table 2). For comparison, although we consider a future sand-loss rate of 5 m3/m/yr to be unreasonable (because it does not account for losses due to island narrowing and decreasing island height), the range of threshold crossing timing suggested by simulations that do not account for an increase in sand-loss rate is 60 to 150 years post-2005 AD (i.e., 2065-2155 AD). Given predictions that SLRRs may increase in the future, potentially reaching 1 m above 2000 levels by 2100 AD (e.g., IPCC, 2007), we also consider the impact of accelerations in SLRR of this magnitude on the moderate sand-loss scenarios. Exponentially increasing the SLRR at each 15-year future time step to reach the 1 m prediction after 100 years results in threshold crossing 15 – 75 years earlier (i.e., 30 – 75

25

ACCEPTED MANUSCRIPT years post-2005 or 2035-2080 AD). The range of estimates reported for all scenarios does not include variability due to decadal variations in storms. Deviations from time-averaged storm

PT

activity in the past may either shorten or lengthen the model-derived range.

RI

Because initial system geometry—which dictates where intertidal conditions exist and thus where marsh can accumulate—is determined by RSLR, bay sedimentation rate, and outer-shelf

SC

slope, synthesizing initial conditions from geologic data (in preparation for running the model)

NU

allows us to further constrain (compared to ranges provided in the literature) estimates of RSLR and bay sedimentation rates for North Chandeleur Island. For example, RSLR over the last 1500

MA

years had to be greater than 3 m because 3 m (or less) of RSLR does not allow for creation of a geometrically feasible initial condition when combined with other parameter estimates (the bay

TE

D

is too deep to represent a delta lobe just after abandonment, e.g., Figure 7A). If, on the other hand, RSL rose 5 m (or less) over the last 1500 years, the entire marsh/bay sediment layer must

AC CE P

have existed prior to barrier island migration (i.e., it must have been entirely composed of marsh associated with the delta, and therefore sedimentation rate over last 1500 years must have been 0 mm/yr) because removing any of the marsh layer from the initial condition results in a sound that is too deep (subtidal) to represent a marsh-covered delta lobe immediately following abandonment (to visualize this see Figure 4B and consider geometry if some or all of the marsh layer were to be replaced by open water). Similarly, if RSL rose 7.5 m over the last 1500 years, all of the marsh/bay layer must have accumulated throughout the last 1500 years (i.e., bay sedimentation rate = 2 mm/yr as in Figure 4C) because adding the marsh/bay layer to the initial stratigraphy places the surface of the marsh too far above sea level (supratidal) to represent a viable initial condition (Figure 7B). Given that some portion of the marsh/bay layer was likely pre-existing (because delta lobes host marshes) and that some bay sedimentation is likely to have

26

ACCEPTED MANUSCRIPT occurred over the last 1500 years, these combinations of RSLRR and bay sedimentation rate bracket the range of possible values suggesting that RSLR must have been greater than 5 m but

PT

less than 7.5 m and that bay sedimentation rate must have been greater than 0 mm/yr and less

RI

than 2 mm/yr.

SC

5. Discussion

NU

5.1 Limitations

The morphological‐behavior modeling approach, which is useful for providing insights

MA

into barrier island evolution [e.g., Cowell et al., 1995; Stolper et al., 2005; Moore et al., 2010], has some limitations that should be considered prior to a discussion of implications. These are

relevant points here.

TE

D

discussed at length in Moore et al., 2010, but we summarize the most important and most

AC CE P

GEOMBEST is a cross‐shore model and therefore we have necessarily collapsed 30 km of coast having similar characteristics to one cross-shore profile that provides a reasonable approximation of average conditions across the study area. As a result, our simulations do not suggest quantitatively what may occur at any one location along North Chandeleur Island but rather provide a general assessment of likely average behavior of the barrier system in the study area. Additionally, though the effects of inlets and alongshore sediment transport gradients on sediment availability can be incorporated into GEOMBEST experiments using the sand-loss rate parameter, the model does not directly address inlet processes or barrier island formation strictly by alongshore processes. While successful reproduction of 1920 morphology for the Chandeleur Islands does not conclusively determine that GEOMBEST simulations correctly represent the

27

ACCEPTED MANUSCRIPT relevant input parameters and processes, it is consistent with this idea and currently represents the best means for assessing and calibrating the model.

PT

Simplifying assumptions regarding barrier island behavior are made in implementing the

RI

morphological‐behavior modeling approach in general. For example, as a

SC

morphological‐behavior model, GEOMBEST does not simulate changes in island behavior due to variations in storm activity. Future deviations from time-averaged storm activity in the past

NU

may alter future island response beyond the range of potential behavior simulated here, extending or shortening the time until actual threshold crossing occurs. Additionally, model

MA

formulation assumes that, over long time scales, the shoreface tends to evolve toward a profile shape that is in equilibrium with the local wave climate. Numerous coastal models, along with all

TE

D

one‐line coastal models, make this assumption [e.g., Cowell et al., 1995; Ashton et al., 2001; Storms et al., 2002] and there is evidence to suggest that in some locations this is a valid

AC CE P

assumption, especially along the open coast and away from inlets and the effects of coastal engineering [e.g., List et al., 1997]. To allow for the likelihood that some coastal locations do not evolve toward an equilibrium profile even over long time scales, GEOMBEST simulations drive toward an equilibrium morphology but do not require equilibrium to be attained. Another challenge that arises when modeling barrier islands over long time scales is that of defining initial conditions. In the case of the Chandeleur Islands, despite the availability of a considerable amount of geophysical and geological data, several important initial conditions are poorly constrained. We have addressed this issue by developing a comprehensive suite of model simulations that considers all possible combinations of the plausible range of initial conditions and parameter values derived from the available literature. In this way our reporting of the range

28

ACCEPTED MANUSCRIPT of threshold timing estimates across all of the viable simulations incorporates the uncertainty in

PT

geologic constraints.

RI

5.2 Implications

By capturing the range of parameters presented in the literature that can be combined to

SC

create a viable initial condition as well as a range of values for substrate sand content, the

NU

combined results from simulations i-vi and the most-plausible simulation provide a range of estimates for threshold crossing timing that accounts for the effects of uncertainties on model

MA

output. Though we cannot pinpoint the exact timing of threshold crossing given uncertainties in geologic variables and future conditions, and assumptions made in the modeling approach,

TE

D

simulations suggest that the Chandeleur Islands, which formed approximately 1500 years ago following delta lobe abandonment, likely have only ~10% (or less) of their lifetime remaining.

AC CE P

Our analysis suggests that although the islands may persist for a while longer, there is insufficient sand in the system to counter the effects of geometric constraints such as rapid RSLR, sand losses and marsh erosion. These findings are consistent with continued landward recession of the majority of North Chandeleur Island since Hurricane Katrina despite the recent growth of incipient beaches and laterally propagating spits connecting exposed, eroding, marsh segments [Sallenger et al., 2009]. Results are also consistent with predictions by Fearnley et al., [2009], based on trends in island area since 1855, that the marsh core will erode completely and that North Chandeleur Island will convert to an ephemeral sand body between 2013 and 2037. Additionally, lidar observations made by Sallenger et al., [2009] reveal the presence of failure modes (e.g., marsh erosion feedbacks, and unprecedented shoreline retreat rates and patterns) that may ultimately contribute to this process. Interestingly, adding sand (to simulate the effects

29

ACCEPTED MANUSCRIPT of beach nourishment) to the ‘most plausible’ simulation in sufficient quantity to maintain the 2005 shoreline position along the 18-km study area for 50 years requires an average sand volume

PT

of 3.6 million m3/m/yr, or an average total sand volume of 180 million m3. For comparison, in

RI

2010, approximately, 3.5 million m3 of sand was emplaced along a 14-km stretch of North Chandeleur Island to build a berm in response to the Deepwater Horizon Spill of April 20, 2010

SC

in the Gulf of Mexico. A volume of sand similar to the 2010 emplacement would need to be

NU

emplaced annually to stabilize the shoreline. Thus, to the extent that the 2010 sand berm project was intended to restore and stabilize the barrier island, it is not likely to be successful.

MA

Given uncertainties in estimates of geologic constraints for the Chandeleur islands, we cannot rule out the possibility that a threshold crossing has already been initiated and that sand

TE

D

accumulations following Hurricane Katrina represent only a short-term recovery trend in a system poised for imminent disintegration. For example, threshold crossing year estimates at the

AC CE P

high end of the predicted range result from simulations in which the substrate sand proportion is increased by 40%—an increase that results in sand proportions at or beyond the high end of the observed range. If these estimates are too high, or if estimates of recent RSLRR and sand-loss rates are too low, threshold crossing will likely occur sooner than the simulations suggest. Threshold crossing will also occur sooner than simulations suggest if future conditions are considerably stormier than average conditions have been in the past, and vice versa. Interestingly, just as initial conditions are constrained by the feasibility of geometric relationships and therefore provide constraints on the range of possible long-term sea level rise rates (see Results), this modeling exercise—which is based on geometry and conservation of mass—calls attention to the importance of considering geometric constraints holistically. For example, reproducing barrier morphology and straigraphy, as well as shoreline change rates, that

30

ACCEPTED MANUSCRIPT closely resemble modern conditions, requires stratigraphic-layer-scale sand percentages that are higher than those previously reported. Either the current estimates for stratigraphic-layer-scale

PT

sand percentage are incorrect (perhaps heterogeneities have led to undersampling of sand-rich

RI

units, e.g., point bars in distributary channels), or estimates for other parameters are incorrect. For example, instead of increasing sand percentage to calibrate model runs, sand-loss rates

SC

(representing alongshore sediment transport gradients) could have been reduced below reported

NU

values. Such discrepancies between estimates derived from previous field studies and what is geometrically feasible and consistent with principles of mass conservation should spark a debate

MA

and could motivate future work designed to provide better constraints on the range of parameters that are important in determining system behavior.

TE

D

To explore the sensitivity of simulation outcome to RSLRR and sediment-loss rates, we conducted additional experiments to determine how much these two parameters would have to

AC CE P

change–within the most-plausible scenario—between 1920 and 2005, to initiate a threshold crossing in 2005. Only small increases (relative to the range of parameter uncertainties) in RSLRR and/or sand-loss rates beyond those specified in the most-plausible simulation were required to cause intersection of the 2005 shoreface with the 1920 marsh edge (i.e., RSLRR = 6 mm/yr (9 m over 1500 years), or sand-loss rate = 25 m3/m/yr, or RSLRR = 5 mm/yr (7.5 m over 1500 years) and sand-loss rate = 15 m3/m/yr). The relatively smaller increase in RSLRR (50%) than sand-loss rate (500%) required to force a threshold crossing in 2005—along with the effect of increases in future sea level rise rates on predictions of threshold crossing timing—suggests that RSLRR is of greater importance than sand-loss rates in determining when an island system will cross a threshold.

31

ACCEPTED MANUSCRIPT This modeling effort, based on observations of the Chandeleur Islands where change is occurring rapidly, highlights the importance of RSLRR in determining when barrier islands are

PT

likely to change state. The influence of increases in sea level rise rate on the timing of threshold

RI

crossing suggest that although sand-poor barrier island systems are most vulnerable, other types

NU

effects of climate change become more pronounced.

SC

of barrier island systems throughout the world may also begin to disintegrate in the future as the

6. Conclusions

MA

We address climate-change induced deltaic barrier island loss via conversion from subaerial landform to submarine shoal using a morphological-behavior model, which is based on

TE

D

geometric relationships and mass conservation. A suite of numerical model simulations—which encompass the range of plausible values for RSLR, bay sedimentation rate, initial outer-shelf

AC CE P

profile shape, shoreface depth, substrate composition and sand-loss rates as derived from the best-available geologic data—suggest that the Chandeleur Islands have less than 10% of their lifetime remaining. Our results also constrain values for RSLRR in the vicinity of the Chandeleur Islands to between 5 and 7.5 m of total rise over the last 1500 years, and demonstrate that RSLRR is more important than the sand-loss rate in determining when an island will cross a threshold and change state to become a submarine shoal. Given increasing sea level rise rates in many areas worldwide, our results suggest that disintegration of barrier island systems may occur more frequently in the future and that increases in global sea level may cause deltaic-style island loss to become widespread.

7. Acknowledgements

32

ACCEPTED MANUSCRIPT We thank Mike Miner, Ioannis Georgiou, Jim Flocks, Mark Kulp, Dave Twichell, Shea Penland, and Abby Sallenger for assistance with the development of initial conditions and model

PT

input parameters. We especially thank Abby Sallenger for support of this work and insightful

RI

comments on an early draft of this manuscript. Cheryl Hapke, Dave Twichell and Torbjörn Törnqvist also provided helpful comments on an earlier version. We gratefully acknowledge

SC

funding from the U.S. Fish and Wildlife Service via a USGS Cooperative Agreement with

NU

Oberlin College and the University of Virginia.

MA

8. References

Ashton, A., et al. (2001), Formation of coastline features by large-scale instabilities induces by

TE

D

high-angle waves, Nature, 414, 296-300.

AC CE P

Ashton and Murray (2006), High-angle wave instability and emergent shoreline shapes: 1. Modeling of sand waves. Flying spits and capes. Journal of Geophysical Research- Earth Surface, 111 (F04011): doi:10.1029/2005JF0000423.

Balsillie, J. H., and J. F. Donoghue (2004), State of Florida Department of Environmental Protection, Division of Resource Assessment and Management, 66 pp, Florida Geological Survey Report of Investigation no. 103.

Brooks, G. R., et al. (1995), East Louisiana continental shelf sediments: A product of delta reworking, Journal of Coastal Research, 11(4), 1026-1036.

33

ACCEPTED MANUSCRIPT Cowell, P. J., et al. (1995), Simulation of large-scale coastal change using a morphological-

PT

behaviour model, Marine Geology, 126, 45-61.

RI

Ellis, J., and G. W. Stone (2006), Numerical simulation of net longshore sediment transport and granulometry of surficial sediments along Chandeleur Island, Louisiana, USA, Marine Geology,

NU

SC

232(1), 115-129.

Everts, C. H. (1978), Geometry of profiles across inner continental shelves of the Atlantic and

MA

Gulf Coasts of the United States, Technical Report, 92 pp, U.S. Army Corps of Engineers,

TE

D

Coastal Engineering Research Center, Fort Belvoir, Virginia.

Fairbanks, R. G. (1989), A 17000-year glacio-eustatic sea level record: influence of glacial

AC CE P

melting rates on the Younger Dryas event and deep-ocean circulation, Nature, 342, 637-642.

Fearnley, S., et al. (2009), Hurricane impact and recovery shoreline change analysis of the Chandeleur Islands, Louisiana, USA: 1855-2005, Geo-Marine Letters, 29, 455-466.

Flocks. et al., (2009) Sediment sampling analysis to define quality of sand resources, in Sand resources, regional geology, and coastal processes of the Chandeleur Islands coastal system— an evaluation of the Breton National Wildlife Refuge, edited by D. Lavoie, U.S. Geological Survey Scientific Investigations Report 2009–5252, pp 99-121.

34

ACCEPTED MANUSCRIPT Friedrichs, C. T., and L. D. Wright (2004), Gravity-driven sediment transport on the continental

PT

shelf: implications for equilibrium profiles near river mouths, Coastal Engineering, 51, 795-811.

RI

Georgiou, I. and Schindler., (2009) Numerical simulation of waves and sediment transport along a transgressive barrier island, in Sand resources, regional geology, and coastal processes of the

SC

Chandeleur Islands coastal system—an evaluation of the Breton National Wildlife Refuge, edited

NU

by D. Lavoie, U.S. Geological Survey Scientific Investigations Report 2009–5252, pp 143-165.

MA

IPCC (2007), Climate Change 2007: The Physical Science Basis. Contribution of Working Group I to the Fourth Assessment Report of the Intergovernmental Panel on Climate Change

TE

D

[Solomon, S., D. Qin, M. Manning, Z. Chen, M. Marquis, K.B. Averyt, M. Tignor, and H.L.

AC CE P

Miller (eds.)]. Cambridge University Press, Cambridge, UK and New York 996, pp.

Kolb, C. R., and J. R. van Lopik (1966), Depositional environments of the Mississippi River delta plain-- southwestern Louisiana, in Deltas in their Geologic Framework, edited by M. L. Shirley and J. A. Ragsdale, Houston Geological Society, Houston, TX.

Komar, P.D. (1976), Beach Processes and Sedimentation, Prentice Hall, Englewood Cliffs, N.J., 429p.

Knutson et al. (2010), Tropical cyclones and climate change. Nature, 3.3, 157-163.

35

ACCEPTED MANUSCRIPT Lavoie, D., ed. (2009). Sand Resources, Regional Geology, and Coastal Processes of the Chandeleur Island Coastal System: an Evaluation of the Breton National Wildlife Refuge, U.S.

RI

PT

Geological Survey SIR 2009-5252, 174 pp.

List, J. H., et al. (1997), Accelerated relative sea-level rise and rapid coastal erosion testing a

NU

SC

causal relationship for the Louisiana barrier islands. Marine Geology, 140, 347-365.

McBride, R. A., et al. (1992), Analysis of barrier shoreline change in Louisiana from 1853 to

MA

1939, in Atlas of Shoreline Changes in Louisiana, edited by S. J. Williams, et al., p. 103, U.S.

TE

D

Geological Survey Miscellaneous 1-2150-A, Reston, Virginia.

Meade, R. (1995), Contaminants in the Mississippi River, 1987-92, U.S. Geological Survey

AC CE P

Circular 1133, Reston, VA.

Miner, M.D., et al., (2008), Louisiana Barrier Island Comprehensive Monitoring Program,Tasks 5 and 6: South-Central Louisiana and Northern Chandeleur Islands, Bathymetry and Historical Seafloor Change 1873- 2006, vol. 1: Methods and Error Analysis for Bathymetry, Pontchartrain Institute for Environmental Sciences, Technical Report 002-1-2008, Submitted to Louisiana Department of Natural Resources, 44 p.

Miner, M.D. et al., (2009) Historical (1869-2007) sea floor evolution and sediment dynamics along the Chandeleur Islands, in Sand resources, regional geology, and coastal processes of the

36

ACCEPTED MANUSCRIPT Chandeleur Islands coastal system—an evaluation of the Breton National Wildlife Refuge, edited

PT

by D. Lavoie, U.S. Geological Survey Scientific Investigations Report 2009–5252, pp 75-95.

RI

Moore, L.J., (2000). Shoreline mapping techniques. Journal of Coastal Research, v. 16 n. 1, pp. 111-124.

SC

Moore, L. J., et. al., (2007), Modeling barrier island response to sea-level rise. in Coastal

NU

Sediments ’07, edited by N. Kraus and J. Rosati, pp. 1153- 1164, American Society of Civil

MA

Engineers.

Moore, L. J., et. al. (2010), Complexities in barrier island response to sea-level rise: Insights

TE

D

from model experiments, North Carolina Outer Banks, U.S.A., Journal of Geophysical

AC CE P

Research- Earth Surface, doi:10.1029/2009JF001299.

Moore, L. J., et al. (2011), Barriers on the brink: the complex intertwined roles of geologic framework, sediment availability and sea-level rise in island evolution. In: Coastal Sediments 2011, edited by P. Wang, J. D. Rosati and T. M. Roberts, pp. 272-285, World Scientific, Miami, FL, USA.

Nicholls, R. J., et al. (1998), Evaluation of depth of closure using data from Duck, NC, USA, Marine Geology, 148, 179-201.

Penland, S., et al. (1985), Barrier island arcs along abandoned Mississippi River deltas, Marine Geology, 63, 197-233.

37

ACCEPTED MANUSCRIPT

Penland, S., et al. (1988), Transgressive depositional systems of the Mississippi Delta Plain: A

PT

model for barrier shoreline and shelf sand development, Journal of Sedimentary Petrology, 58,

RI

932-949.

SC

Penland, S., and K. E. Ramsey (1990), Relative sea-level rise in Louisiana and the Gulf of

NU

Mexico: 1908 -1988, Journal of Coastal Research, 6, 323-342.

MA

Penland, S., et al. (2005), Changes in Louisiana's Shoreline: 1855-2002, Journal of Coastal

TE

D

Research, SI(44), 7-39.

Ritchie, W., et al. (1992), Coastal Sand Dunes of Louisiana: The Chandeleur Islands Barrier

AC CE P

System -Coastal Geology Technical Report No. 7 174 pp, Louisiana Geological Survey and University of Aberdeen.

Sallenger, A. (2000), Storm-impact scale for barrier islands, Journal of Coastal Research, 16, 890-895.

Sallenger, A., et al. (2009), Extreme coastal changes on the Chandeleur Islands following Hurricane Katrina, in Sand resources, regional geology, and coastal processes of the Chandeleur Islands coastal system—an evaluation of the Breton National Wildlife Refuge, edited by D. Lavoie, U.S. Geological Survey Scientific Investigations Report 2009–5252, Chapter B, pp 2735.

38

ACCEPTED MANUSCRIPT

Sasaki, W., and T. Hibiya (2007), Interannual variability and predictability of summertime

RI

PT

significant wave heights in the western north pacific, Journal of Oceanography, 63, 203-213.

Sasaki, W. (2009), e-mail correspondence on May 13, 2009, edited by O. Brenner,

NU

SC

Charlottesville, VA.

Stolper, D., et al. (2005), Simulating the evolution of coastal morphology and stratigraphy with a

MA

new morphological-behavior model (GEOMBEST), Marine Geology, 218, 17-36.

TE

D

Storms, J. E. A., et al. (2002), Process-response modeling of wave-dominated coastal systems: Simulation evolution and stratigraphy on geological timescales, Journal of Sedimentary

AC CE P

Research, 72(2), 226-239.

Thieler, E. R., E. A. Himmelstoss, J. L. Zichichi, and A. Ergul (2009), Digital Shoreline Analysis System (DSAS) version 4.0- An ArcGIS extension for calculating shoreline change Rep., U.S. Geological Survey Open-File Report.

Törnqvist, T.E., et al. (2008), Mississippi Delta subsidence primarily caused by compaction of Holocene strata. Nature Geoscience, 1, 173-176.

39

ACCEPTED MANUSCRIPT Valvo, L.M., Murray, A.G., Ashton, A.D. (2006), How does underlying geology affect coastline change? An initial modeling investigation. Journal of Geophysical Research-Earth Suface 111,

RI

PT

doi: 10/1029/2005JF0000340.

Twichell, D., et al. (2009a), Subsurface control on seafloor erosional processes offshore of the

NU

SC

Chandeleur Islands, Geo-Marine Letters, 29, 349-358.

Twichell, D. et al., (2009b), Geologic mapping of distribution and volume of potential resources,

MA

in Sand resources, regional geology, and coastal processes of the Chandeleur Islands coastal system—an evaluation of the Breton National Wildlife Refuge, edited by D. Lavoie, U.S.

TE

D

Geological Survey Scientific Investigations Report 2009-5252, Chapter E, pp 75-95.

AC CE P

Twichell, D. et al., (2011) Geologic controls on sediment distribution and transport pathways around the Chandeleur Islands, LA., USA, In: Coastal Sediments 2011, edited by P. Wang, J. D.

Rosati and T. M. Roberts, pp. 2184-2197, World Scientific, Miami, FL, USA.

Wright, L. D. (1995), Morphodynamics of inner continental shelves, 241 pp., CRC Press, Boca Raton, FL.

Zervas, C. (2004), North Carolina Bathymetry/Topography Sea Level Rise Project: Determination of sea level trends, NOAA Technical Report NOS CO-OPS 041, 1-31 pp, National Oceanic Service, NOAA, Silver Spring, MD.

40

ACCEPTED MANUSCRIPT Figure Captions Figure 1. The Louisiana-Mississippi Coastline (upper panel), Chandeleur Islands outlined in red.

PT

Close up of the Chandeleur Islands (lower panel) with the North Chandeleur Island study area

SC

medium resolution shoreline (chart 11363. rev. Nov. 1991).

RI

outlined in red (bathymetry is from Miner et al. (2008) and topography from NOAA 1:80K

NU

Figure 2. Cross-shore schematization of coastal morphology for a low‐gradient barrier island coast. The three functional realms in GEOMBEST (shoreface, barrier, and bay) are distinct

MA

stratigraphic units that comprise the coastal tract. After Stolper et al. [2005] and Moore et al.

TE

D

[2010].

Figure 3. Average modern morphology and stratigraphy for the Northern Chandeleur Island

AC CE P

study area derived from bathymetric data, topographic surveys, sediment cores and geophysical records. The six viable late-Holocene simulations and the most-plausible simulations closely reproduce these conditions.

Figure 4. Initial geometry and stratigraphy for simulations (A) i - iii, (B) iv and v, and (C) vi, which effectively reproduce 1920 conditions (final barrier island configuration would appear as it does in Figure 7B). See Table 3 for simulation input parameters. Figure 5. Average barrier island slope of the study area (represented by the slope of a line between the landward base of the barrier and the base of the shoreface) changes as shoreface depth changes. Elevations are relative to an initial sea level, which is 7.5 m below modern sea level. 41

ACCEPTED MANUSCRIPT

Figure 6. The range of estimated values for geologic inputs is explored through 48 simulations

PT

(outlined in bold). Twenty-four combinations of input parameters do not represent feasible initial

RI

conditions and are eliminated prior to running the model (dark gray boxes). Ten of the remaining combinations (light gray boxes) do not allow reproduction of 1920 conditions. Six

SC

simulations (white boxes, simulations i-vi) result in viable combinations of initial conditions and

NU

input parameters that closely reproduce 1920 conditions. References to Figure 4 indicate the corresponding figure panel that depicts each of the 3 different initial geometries that result from

MA

the 6 viable simulations. References to Figures 7-9 indicate the combination of parameter values

TE

D

depicted in each of the cited figure panels.

Figure 7. Examples of initial conditions (i.e., initial geometry and stratigraphy) that are not

AC CE P

viable. In both cases the entire marsh layer (analogous to marsh/bay layer in other simulations) must have accumulated in association with the river prior to the start of the model run for the bay sedimentation rate to be 0 mm/yr throughout the run. Since marshes grow in intertidal areas on delta lobes the combination of conditions shown is not viable because it dictates (A) a backbarrier bay region that is entirely subtidal and (B) a back-barrier region that is primarily supratidal (up to 4.5 m above sea level).

Figure 8. Representative examples of the two types of mismatch between simulations and observations for simulations having feasible initial conditions, but lacking a combination of parameter values that allows reproduction of the historical (1920) morphology and stratigraphy (light gray boxes, Figure 6). (A) The final time step in a simulation starting 1400 ybp and ending in 1920 AD having the specified input parameters (the modern barrier island appears in yellow). 42

ACCEPTED MANUSCRIPT Each trace represents a 100 year time increment (i.e., every other time step is plotted). The initial surface is shown as a thin black line above the bold black line, which represents the shelf surface

PT

in 1920. (B) Comparisons between the initial, model-generated, and observed morphology and

RI

stratigraphy showing that the 15 m shoreface depth was too shallow to reproduce the historical configuration. The 5 m SLR simulation having the same parameters but with a shoreface depth

SC

of 20 m instead of 15 m, as well as the 5 m SLR simulation having the same parameters but with

NU

an initial shoreface slope instead of an initial delta slope, failed for the same reason. (C) and (D) The final time step and comparisons between the initial, model‐generated, and modern

MA

morphology and stratigraphy for a simulation starting 1400 ybp and ending in 1920 having the specified input parameters. The 7.5 m SLR simulations having shoreface depths of 15-30 m in

TE

D

combination with an initial shoreface slope as well as those having shoreface depths of 20-30 m (20 m shown) in combination with an initial delta slope, were not able to reproduce the historical

AC CE P

configuration because the lower initial position of the barrier (required by the greater total sea level rise) resulted in too much shoreface erosion throughout the simulation.

Figure 9. The final timestep for simulations ii (A; similar to i and iii), iv (C; similar to v) and iv (E) starting 1400 ybp and ending in 1920 (see Table 2 for input parameters). Ghost traces represent 100 year increments (i.e., every other time step) and the modern barrier appears in yellow. Comparisons between the initial, model-generated (1920) and actual (1920) morphology and stratigraphy for simulations ii (B: similar to i and iii), iv (D; similar to v) and iv (F), showing close agreement between the shape of the model-generated morphology and the shape of the actual morphology.

43

ACCEPTED MANUSCRIPT Figure 10. (A) Initial conditions and (B) 1400-year simulation of barrier evolution from ~1500 ybp to 1920 AD for most-plausible scenario with 10 m3/m/yr sand loss. Final time-step shown in

PT

color with ghost traces every 100 years (i.e., every other time step). (C) Comparison between

RI

model-generated and actual shoreface morphology in 1920 AD for most-plausible scenario, and (D) close-up of barrier for forward simulation. In D, every other ghost trace is plotted. Thus, the

SC

2005 ghost trace appears immediately landward of the 1920 trace and ghost traces appearing

NU

beyond 2005 appear in 30-year increments. Threshold crossing occurs in 2125 when the shoreface intersects the 1920 bay-side shoreline (red cross-hair). Yellow hatched area represents

MA

sand captured by the island system in the model post-1920 that in reality appears to have been lost to Chandeleur Sound since cessation of landward back-barrier marsh progradation. Remnant

AC CE P

Tables

TE

D

barrier core in 2125, excluding this lost sand, is shown in green.

Table 1. Estimates for range of geologic parameter values tested in the initial suite of 48 simulations.

Table 2. Summary of parameter values for inputs that vary across the simulations and estimates for threshold crossing timing associated with each scenario.

Table 3: Modeled migration rates, shoreline positions (1870, 1920, and 2005), and shoreline change (1970-1920 and 1920-2005). Modeled shoreline change is compared with observed shoreline change for the same time periods based on actual average shoreline change rates for 1855 - 1922, and 1922 – 2005 from Martinez [personal communication] and Fearnley et al. [2009]. 44

Figure 1

AC CE P

TE

D

MA

NU

SC

RI

PT

ACCEPTED MANUSCRIPT

45

MA

NU

SC

RI

PT

ACCEPTED MANUSCRIPT

AC CE P

TE

D

Figure 2

46

AC CE P

Figure 3

TE

D

MA

NU

SC

RI

PT

ACCEPTED MANUSCRIPT

47

AC CE P

TE

D

MA

NU

SC

RI

PT

ACCEPTED MANUSCRIPT

Figure 4

48

AC CE P

Figure 5

TE

D

MA

NU

SC

RI

PT

ACCEPTED MANUSCRIPT

49

MA

NU

SC

RI

PT

ACCEPTED MANUSCRIPT

AC CE P

TE

D

Figure 6

50

Figure 7

AC CE P

TE

D

MA

NU

SC

RI

PT

ACCEPTED MANUSCRIPT

51

Figure 8

AC CE P

TE

D

MA

NU

SC

RI

PT

ACCEPTED MANUSCRIPT

52

AC CE P

TE

D

MA

NU

SC

RI

PT

ACCEPTED MANUSCRIPT

Figure 9

53

AC CE P

Figure 10

TE

D

MA

NU

SC

RI

PT

ACCEPTED MANUSCRIPT

54

ACCEPTED MANUSCRIPT

3, 5, 7.5

2.2.1

Bay Sedimentation (mm/yr)

0, 2

2.2.2

Initial Outer Shelf Profile Shape

delta, shoreface

2.2.4

Shoreface Depth (m)

15, 20, 25, 30

2.2.5

Substrate Composition (percentage of sand in a stratigraphic layer)

Barrier, 95%* Marsh/Bay, 20%* Distrib./Delta Front,30%* Prodelta, 20%* Sand Sheet, 70%*

2.2.6

*5

2.2.7

MA

PT ED

3

Sand-Loss Rate (m /m/yr)

NU

Total Relative Sea Level Rise (m)

RI

Manuscript section

Tested values

SC

Input parameters

AC

CE

*values held constant in the initial suite of 48 late-Holocene simulations

55

PT

Table 1

ACCEPTED MANUSCRIPT

PT

Table 2

SandLoss Rate Pre-2005 3 (m /m/y)

i

5

0

ii

5

iii

5

shore

20

0.95

0

5

shore

25

0.95

5

0

5

shore

30

iv

5

0

5

delta

25

v

5

0

5

delta

30

vi mostplausible Range of Values

7.5

2

5

delta

6

1

5

5 - 7.5

0-2

5

delta delta & shore

sand sheet

0.28

0.42

0.28

0.95

0.27

0.41

0.27

0.95

0.95

0.28

0.42

0.28

0.95

0.95

0.20

0.30

0.20

0.70

0.95

0.21

0.32

0.21

0.74

15

0.95

0.27

0.41

0.27

0.95

25

0.95

0.26

0.40

0.26

0.91

15 - 30

0.95

0.200.28

0.300.42

0.200.28

0.700.95

AC

prodelta

PT ED

barrier

dist/ delta front

CE

Initial Slope

Shoreface Depth (m)

NU

Bay Sed. Rate (mm/y)

marsh/ bay

MA

Sim.

Total Relative Sea Level Rise (m)

SC

RI

Substrate Composition (stratigraphic layer sand percentage)

56

Threshold Crossing Timing Year A.D. (yrs post-2005) SLRR Future SLRR same as past increase Sand-Loss Rate Post-2005 5 10 20 10 3 3 3 3 m /m/y m /m/y m /m/y m /m/y 2095 2080 2065 2050 (90) (75) (60) (45) 2125 2110 2080 2065 (120) (105) (75) (60) 2155 2140 2110 2080 (150) (135) (105) (75) 2095 2080 2065 2050 (90) (75) (60) (45) 2125 2110 2080 2065 (120) (105) (75) (60) 2065 2065 2035 2035 (60) (60) (30) (30) 2155 2125 2095 2080 (150) (120) (90) (75) 20652055203520352155 2140 2110 2080 (60-150) (60-135) (30-105) (30-75)

ACCEPTED MANUSCRIPT Table 3

Bay Sed. Rate (mm/ yr)

mostplausi ble

6

1

i

5

0

ii

5

0

iii

5

iv

Initial OuterShelf Slope

Shoref ace Depth (m)

Up to 1920

Up to 2005

-25

6.6

6.7

-25

6.3

-20

6.4

0

Delta Shoref ace Shoref ace Shoref ace

NU

Se a Lev el Ris e (m)

Shoreline Positions (km)

5

0

Delta

-25

v

5

0

Delta

-30

vi

7.5

2

Delta

TE

187 0

192 0

Shoreline Change (m)

2005

18701920

19202005

72.202

406

688

72.377

375

687

72.518

406

719

72.213

359

625

72.424

469

781

72.307

406

719

72.294 Avg. Modeled Observe d

500

937

417

737

110

630

6.2

6.3

6.7

6.8

6.6

6.7

7.2

7.4

57

SC

RI MA

6.5

D

-30

-15

6.4

AC CE P

Sim.

PT

Migration Rate (m/yr)

Input Parameters

71.1 09 71.3 15 71.3 93 71.2 29 71.1 74 71.1 82 70.8 57

71.5 15 71.6 9 71.7 99 71.5 88 71.6 43 71.5 88 71.3 57

ACCEPTED MANUSCRIPT

AC CE P

TE

D

MA

NU

SC

RI

PT

Highlights RSLRR is most important factor determining island vulnerability to state change. The deltaic Chandeleur Islands likely have <10% of their lifetime remaining. Increases in global SL may cause deltaic-style island loss to become widespread.

58