Marine Geology 427 (2020) 106272

Contents lists available at ScienceDirect

Marine Geology
journal homepage: www.elsevier.com/locate/margo

Late-Holocene shoreline responses to competing shelf, bay, and beach
accommodation spaces under conditions of relative sea level change, and
the potential for future catastrophic beach retreat in the Columbia River
Littoral Cell, Washington and Oregon, USA

T

⁎

Curt D. Petersona, , Tamara C. Lindeb, Sandy Vanderburghc
a
b
c

Geology Department, Portland State University, Portland, OR 97207, USA
Cascade Radon, Inc, Portland, OR 97230, USA
Kwantlen Polytechnic University, Surrey, British Columbia V3W2M8, Canada

A B S T R A C T

The Columbia River Littoral Cell (CRLC) (160 km in length) provides opportunities to compare competing accommodation space relations under diﬀerent conditions
of relative sea level change. The CRLC system includes abundant littoral sand supply from the large Columbia River, late-Holocene prograded beach plains and
barrier spits (0.5–5 km in width), two large marine-dominated estuaries (Willapa Bay and Grays Harbor), and a high-wave-energy inner-shelf. Littoral sand accumulation rates in prograded beach plains and barrier deposits are based on paleo-shoreline positions that are dated by great-earthquake catastrophic beach retreat
scarps (n = 10) from 0.3 to 5.0 ka. The retreat scarp timelines are mapped in across-shore GPR transects (n = 79), yielding timeline-bounded polygons (n = 247).
The polygons are evaluated for littoral sand volume and bounding ages, yielding volume accretion rates (m3 ka−1), which are summed for the four CRLC subcells;
Clatsop Plains (27.4 × 106 m3 ka−1), Long Beach (19.8 × 106 m3 ka−1), Grayland Plains (15.9 × 106 m3 ka−1), North Beaches (11.7 × 106 m3 ka−1). Major
submarine sinks of littoral sand, including the inner-shelf and large marine-dominated estuaries, are evaluated for increased littoral sand accommodation space that
could result from potential future sea level rise of 1, 2 and 3 m during the next century or two. The estimated beach and nearshore sand erosion needed to ﬁll the
increased submarine accommodation space from a 2.0 m rise in relative sea level would result in averaged beach retreat values of ~0.7 km (Clatsop Plains), ~1.2 km
(Long Beach), and ~ 1.3 km (Grayland Plains), about 30–50% larger than previous estimates based on the Brunn method. These catastrophic shoreline retreat
distances (0.7–1.3 km) represent 25–50% of the present widths of prograded barrier spits and beach plains. They serve as warnings about future catastrophic beach
erosion resulting from potential future SLR in other similar barrier spit and beach plain shorelines worldwide.

1. Introduction
The Columbia River Littoral Cell (CRLC) (Fig. 1) includes the extraordinary coincidences of high wave-energies, abundant littoral sand
supply, and reversing relative sea level changes from subduction zone
neotectonics, as superimposed on longer-term eustatic sea level rise.
These conditions have previously been used to constrain accommodation space controls on littoral sand accumulation rates in 1) the uppershoreface under the prograded barrier spits and beach plains (Herb,
2000; Vanderburgh et al., 2010; Peterson et al., 2010a), 2) the innershelf (Twichell and Cross, 2001; Twichell et al., 2010; Peterson et al.,
2016), and 3) the large mesotidal estuaries (Peterson et al., 2013;
Peterson and Phipps, 2016; Peterson and Vanderburgh, 2018a, 2018b).
In this article, the littoral sand volumes that were partitioned in the
CRLC barrier spits and beach plains (0.5–5 km in width) are used to
demonstrate the complex interplays between the shelf, bays, and beaches in competition for available littoral sand supplies, during latest-

⁎

Holocene time (3–0 ka). These relationships are used in this study to
predict responses of high-wave-energy sandy-shorelines to potential
future conditions of sea level rise (SLR) from ongoing global warming
(Kopp et al., 2014; Bamber et al., 2019; Kopp et al., 2019).
Partitioned littoral sand volumes in the four subcells of the CRLC
system (Fig. 1) are based on coseismic beach retreat scarps that provide
timelines (200–800 yr recurrence intervals) throughout the length of
the littoral system (Meyers et al., 1996; Woxell, 1998; Peterson et al.,
2010b). Littoral sand accumulation rates for speciﬁc time intervals and
associated relative sea level changes during late-Holocene time (5–0 ka)
range widely between the four diﬀerent subcells (Linde, 2014). Sand
volumes that were partitioned into each of the four subcells during
corresponding time intervals are compared in this article for 1) across
inner-shelf gradients (slopes), 2) shoreline orientations relative to
dominant wave forcing, and 3) intervening estuary sand sinks. The
prograded shoreline accretion rates are compared to corresponding ﬁll
rates in the inner-shelf and in the large marine-dominated estuaries

Corresponding author.
E-mail addresses: curt.d.peterson@gmail.com (C.D. Peterson), sandy.vanderburgh@kpu.ca (S. Vanderburgh).

https://doi.org/10.1016/j.margeo.2020.106272
Received 17 September 2019; Received in revised form 11 May 2020; Accepted 9 June 2020
Available online 11 June 2020
0025-3227/ © 2020 The Authors. Published by Elsevier B.V. This is an open access article under the CC BY-NC-ND license
(http://creativecommons.org/licenses/BY-NC-ND/4.0/).

Marine Geology 427 (2020) 106272

C.D. Peterson, et al.

Fig. 1. Location of the Columbia River Littoral Cell (CRLC) including 1) bounding headlands (Point Grenville, Tillamook Head), 2) subcell beach segments (Clatsop
Plains, Long Beach, Grayland Plains, North Beaches), 3) inner-shelf bathymetric contours (20 m, 40 m, 60 m water depth), 4) onshore ground penetrating radar (GPR)
transects, 5) oﬀshore seismic lines, 6) selected inner-shelf vibracores, 7) estuary vibracores, and 8) interpreted net-littoral sand transport directions (arrows). The
Columbia River mouth supplied the Holocene littoral sand to the CRLC system. Littoral sand sinks include the inner-shelf, Willapa Bay, Grays Harbor, and the
prograded subcell barriers and beach plains. GPR transects, including WARR and LOOM, are from Peterson et al. (2010b). Oﬀshore seismic lines are from Twichell
et al. (2010). Selected oﬀshore vibracores are from Kaminsky (2006). Beach toe vibracore/drill sites are from Vanderburgh et al. (2010) and Peterson et al. (2016).
Estuary vibracores in Willapa Bay and Grays Harbor, respectively, are from Peterson and Vanderburgh (2018a) and Peterson and Vanderburgh (2018b). Littoral sand
transport directions are from Peterson et al. (2010a) and Peterson et al. (2016).
2

Marine Geology 427 (2020) 106272

C.D. Peterson, et al.

Fig. 2. Sea level curves from the CRLC system. Part
A. Averaged sea level curve (dashed line) in middleto late-Holocene time (8–0 ka). The regional
NAVD88 elevation datum (0 m) is equivalent to
about −1.0 m mean tidal level (MTL). Figure is redrafted from Peterson et al. (2010a). Part B. Highresolution sea level curve for latest-Holocene time
(3–0 ka) including average sea level (dashed line)
and neotectonic ‘cyclic' sea levels (dotted line) from
coseismic subsidence and interseismic uplift. Subsidence events from youngest to oldest correspond
to retreat scarp timelines (Table 2) as follows A
(0.3 ka), B (1.1 ka), C (1.3 ka), D (1.7 ka), E (2.5 ka),
and F (2.8 ka). Figure is redrafted from Peterson and
Vanderburgh (2018a).

2. Background

(Willapa Bay and Grays Harbor), under longer-term time scales (millennial) of net SLR. Reversals of relative sea level occur over shorter
time scales (decades to centuries) and resulted in widespread reversals
of catastrophic beach retreat and then recovery. Such short time scales
are based on previously reported ages of beach erosion and accretion
associated with the last subduction zone neotectonic cycle (1700 CE to
the present) (Woxel, 1998; Phipps et al., 2001; Peterson et al., 2000).
The relative sea level changes (0.5–2.0 m) forced by the neotectonic
cycles in the study area also resulted in episodic accretion of littoral
sand in shallow tidal ﬂats of the large marine-dominated estuaries,
Willapa Bay and Grays Harbor (Peterson and Vanderburgh, 2018a,
2018b). Because the potential shoreline retreat rates for the larger SLR
scenarios (2 and 3 m) are so large in the CRLC system we use the known
past beach accretion to apportion the estimated beach erosion and associated shoreline retreat distances following predicted future SLR.
The sequences of littoral sand accumulation that are documented in
this study are used ‘in reverse’ to yield beach erosion sequences under
conditions of future SLR (1–3 m), as predicted to potentially occur
during the next century or two (Jevrejeva et al., 2012; DeConto and
Pollard, 2016; Mengel et al., 2016; Kopp et al., 2019). Predicted sequences and volumes of beach erosion in the CRLC are based on established littoral sand transfers to more competitive sand sinks or increasing marine accommodation spaces in the inner-shelf and the large
marine-dominated estuaries, under conditions of net SLR. Accommodation spaces for littoral sand in those competing sand sinks will
increase with future SLR at the same time that decreasing accommodation spaces in the onshore (beach) settings will cause shoreline
destabilization and retreat. As shown for the high-energy CRLC system,
such sand transfers to the inner-shelf and marine-dominated estuaries
did occur at the one or two century time scales or at about the durations
of possible near-future SLR scenarios as proposed here. The relations
reported here, for the sand-rich CRLC system, serve as warnings for
high-wave-energy shorelines world-wide that could experience catastrophic erosion of barrier spits and broad beach plains under potential
future conditions of sustained global SLR.

2.1. Littoral sand sources and dispersal in the CRLC system
The Columbia River littoral cell system (160 km in length) contains
four subcells including the Clatsop Plains, Long Beach Spit, Grayland
Plains and North Beaches (Fig. 1). The subcells are separated by large
tidal inlets of the intervening estuaries; Columbia River, Willapa Bay,
and Grays Harbor. The larger CRLC system is bounded by the small
headlands, Tillamook Head (south), and Point Grenville (north).
Through-put of hypersthene-rich Columbia River sand to the littoral
zone (Baker et al., 2010; Peterson et al., 2013) ﬁlled the submerged
inner-shelf and marine-dominated estuaries in mid-Holocene time
(Vanderburgh et al., 2010; Twichell et al., 2010; Peterson and Phipps,
2016; Peterson et al., 2016). The initial ﬁlling of those accommodation
spaces then led to progradation of the barrier spits and beach plains in
late-Holocene time (Woxell, 1998; Vanderburgh et al., 2010; Peterson
et al., 2010a, 2010b). The Holocene Columbia River sand was transported north and south of the Columbia River mouth by nearshore and
inner-shelf longshore transport within, and between, the adjoining
subcells (Fig. 1). There is no evidence of late-Pleistocene shelf sand
(augite-rich) being delivered to the late-Holocene prograded barriers
and beach plains or to the large marine-dominated estuaries by the
Holocene marine transgression. Pre-transgressive shelf sand deposits
were swept north of the CRLC system by strong northward littoral
transport along the shelf during late-Pleistocene and early-Holocene
marine low-stands. A northward shift of storm tracks during Holocene
time (Alder and Hostetler, 2015; Peterson et al., 2016) diminished the
strong net-northward littoral transport, yielding only slightly netnorthward sand transport from the Columbia River mouth in late-Holocene time (5–0 ka).

2.2. Sea level curves for the CRLC system
Reported sea level curves for the CRLC system are shown in Fig. 2.
Long-term net SLR slows from ~7 m ka−1 in middle-Holocene time to
~1.0 m ka−1 in latest-Holocene time. The onset of barrier and beach
plain net-progradation in the CRLC system, at 5–4 ka (Rankin, 1983;
Woxell, 1998; Peterson et al., 2010a, 2010b), corresponds to the
3

Marine Geology 427 (2020) 106272

C.D. Peterson, et al.

(Bruun, 1962; Doyle, 1996; Peterson et al., 2000). Assuming a 1.5 m
rise in relative sea level and 17.5 m interannual depth of closure, the
measured across-shore proﬁles yield calculated retreat values of
~300 m in distance. Doyle (1996) also considered greater values of
coseismic subsidence (2.0 m) and interannual depth of closure (20 m),
as discussed later, in Section 5.6. Most of the beach retreat, following
coseismic subsidence, is assumed to have taken place during the earlystage, or ﬁrst several decades, of the interseismic cycle, as interpreted
below. Where suﬃcient interseismic progradation occurred between
the coseismic retreat events, the preserved sequence of 14C dated retreat
scarps provide timelines of net upper-shoreface progradation (Fig. 4
Part A).
The duration of catastrophic beach retreat, following the last coseismic subsidence event in the Cascadia subduction zone at 1700 CE
(Satake et al., 1996), is constrained by the earliest ages of foredune
reactivation in prograded barriers and beach plains of the CRLC system.
The foredune reactivation and abrupt abandonment by beach progradation, respectively, reﬂect the onset and then acceleration of beach
recovery during the tectonic uplift portion or the middle stage of the
interseismic interval (Cruikshank and Peterson, 2017). Such records of
post-1700 dune reactivation have been reported for 14 sites in the CRLC
subcells (Woxell, 1998). Particularly well-developed dune reactivation
records occur in the WARR and WASH proﬁles, and the adjacent
Washaway Beach cut-bank section, located along the north side of the
Willapa Bay tidal inlet (Fig. 1 and Fig. 4 Part A). Thick sand layers
(1–3 m in thickness) above a weak paleosol (soil A-horizon) in the
WARR and WASH proﬁles greatly exceed thin sand layers (1–5 cm
thick) produced by paleotsunami sand transport in the adjacent interdune valley bogs (Schlichting and Peterson, 2006). Such thick dune
sand layers rule-out foredune dune reactivation by paleotsunami inundation, as initially proposed by Phipps et al. (2001). Numerous trees/
stumps were rooted in the pre-1700 paleosol and/or overlying dune
deposits, as represented by sample sites BD, WASH8 and WASH5 (Fig. 4
Part B). Radiocarbon dating of the dune-buried trees/stumps yield approximate kill dates, calibrated (1σ) median ages (Calib7.10, 2019), of
181 cal BP (site BD) (Phipps et al., 2001), and 184 cal BP (site WASH8)

substantial decrease in the rate of SLR, at 5–4 ka (Fig. 2 Part A). Due to
the neotectonic forcing of relative SLR in the study area (Atwater et al.,
2004), the relative sea level changes follow step-like patterns of cyclic
coseismic subsidence and interseismic uplift (~1.0 ± 0.5 m vertical).
These shorter-term neotectonic cycles are supper-imposed on the
longer-term conditions of eustatic net SLR (Fig. 2 Part B). The shortterm cyclic patterns of relative sea level change are established from
buried tidal marsh or supratidal forest sequences, as regionally correlated by near-ﬁeld paleo-tsunami inundation and 14C dating in Grays
Harbor, Willapa Bay, and the Columbia River estuary (Darienzo et al.,
1994; Jurney, 2001; Atwater et al., 2004). The interseismic intervals of
sea level change are likely more complicated than those shown in Fig. 2
Part B, including the early-and late-stages of little to no uplift and the
intervening middle-stages of sustained tectonic uplift (Cruikshank and
Peterson, 2017).
2.3. Beach and estuary response to cyclic coseismic subsidence and
interseismic uplift
Catastrophic beach-retreat scarps are mapped throughout the CRLC
from ground penetrating radar (GPR) proﬁles and ground-truthing with
auger borehole and/or vibracore sampling (Fig. 1) (Meyers et al., 1996;
Woxell, 1998). The retreat scarps resulted from regional coseismic
subsidence events (0.5–2.0 m) (Fig. 2 Part B), which are correlated to
14
C-dated tidal marsh burial sequences in adjacent estuaries (Atwater
and Hemphill-Haley, 1997; Peterson et al., 2010b). Retreat scarps crosscut or truncate seaward-dipping beach face layers that formed during
preceding interseismic intervals (200–800 yr durations) of beach face
progradation (Jol et al., 1999). Retreat scarp depths (generally 3–10 m
below prograded beach plain surfaces) are imaged by ground penetrating radar (GPR) and ground-truthed by heavy-mineral lag deposits,
as recovered in auger and/or vibracore boreholes (Fig. 3).
The landward distances of the catastrophic beach retreats are not
directly measured, due to erosional losses of the pre-subsidence
shoreline positions. However, beach retreat from the last coseismic
subsidence event at 1700 has been estimated using Bruun's relations

Fig. 3. Ground penetrating radar 3D image of a very-large beach retreat scarp (high-amplitude reﬂections) resulting from a coseismic subsidence event (~2.0 m
subsidence), dated to ~1.7 ka in the LOOM transect in the Long Beach subcell (Fig. 1) (Meyers et al., 1996; Jol et al., 2003; Peterson et al., 2010b). The erosional
scarp (dipping seaward to image left) is enhanced by iron-rich heavy mineral lag deposits, which were ground-truthed for depth and extent by auger drilling and/or
vibracoring. A calibrated two-way signal velocity (0.08 m ns−1) yields a scarp scour depth of 11 m below surface and subsequent prograded beach facies to 6 m depth
below surface, in this image. The site surface elevation (road) is ~7 m above the 0 m NAVD88 vertical datum. The 0 m NAVD88 datum is about 1.0 m below present
mean sea level (MSL) in the study area. The imaged retreat scarp cut about 4 m below the post-subsidence paleo-sea level during the period of abrupt retreat.
4

Marine Geology 427 (2020) 106272

C.D. Peterson, et al.

Fig. 4. Part A. GPR proﬁle (WARR), located north of
the Willapa Bay tidal-inlet cut-bank (Fig. 1)
(Peterson et al., 2010b). The GPR proﬁle includes
the last two catastrophic retreat scarps (bold lines)
at 1700 CE and 1100 yr BP. Boreholes deﬁne a duneburied paleosol (dashed line) between the underlying pre-1700 foredune and the overlying post1700 accreted dune sand deposits. Part B. Map of
Willapa Bay showing the Washaway GPR proﬁle and
adjacent cut-bank exposure (solid line) with sand
buried trees/stumps (sold circles) at sites BD (Phipps
et al., 2001) and WASH8 and WASH5 (Woxell,
1998). Selected vibracore sites (solid squares) are
plotted in the shallow tidal ﬂats of Willapa Bay. Part
C. Representative vibracore logs from Willapa Bay
show basin-wide littoral sand accumulation (stippled patterns) above episodically-subsided wetlands
(shaded patterns). Net SLR (Fig. 2 Part B) preserves
only some subsidence/uplift cycles under conditions
of longer-term tidal ﬂat transgression in Willapa Bay
(Peterson and Vanderburgh, 2018a).

shallow tidal ﬂats of Willapa Bay and Grays Harbor occurred during the
relatively brief periods (< 100 years) of substantial submergence
(≥1.0 m), following the coseismic subsidence events (Peterson and
Vanderburgh, 2018a, 2018b). However, the long-term net SLR
(1.0 m ka−1) during latest-Holocene time (3–0 ka) permitted the inﬂux
and net deposition of littoral sand in the marine dominated estuaries,
Willapa Bay and Grays Harbor (Table 1). Late-Holocene accumulation
rates of littoral sand are estimated to be 110 × 106 m3 ka−1 in the
Willapa Bay tidal ﬂats and 117 × 106 m3 ka−1 in the Grays Harbor
tidal ﬂats and tidal channels. Littoral sand sinks in the subtidal channels

and 184 cal BP (site WASH5) (Woxell, 1998). Adjusting for 0 yr 14C age
at 1950, the average calendar age (CE) for the three prehistoric tree
stumps is 1767, or ~ 70 years after the 1700 subsidence event. Catastrophic beach retreat and associated oﬀshore sand transport likely
occurred within several decades of the 1700 subsidence event in the
high-wave-energy CRLC system. The onset of beach recovery and dune
reactivation, prior to 1770, probably coincided with the beginning of
the middle stage (~1750) of the interseismic interval (Cruikshank and
Peterson, 2017).
Following the initial post-1700 foredune reactivation, wide-spread
beach progradation ensued in the CRLC barriers sand beach plains
(Fig. 4 Part A). The progradation abandoned the reactivated foredunes,
developing the backshore ﬂats between the prehistoric foredune and
modern foredunes. The relatively rapid beach progradation, between
~1750–1885, was likely forced by 1) regional onshore sand transport,
following tectonic uplift and 2) localized gradients in along-shore sand
transport, following river sand throughput to the littoral zone from the
Columbia River mouth. However, the rates of beach accretion greatly
accelerated after 1885 throughout nearly all of the CRLC system
(Woxell, 1998). The natural conditions of beach sand supply in the
CRLC system were greatly altered in the early 1900s due to anthropogenic impacts from 1) grazing, logging, and agriculture in the Columbia River tributary drainage basins and 2) jetty construction and
oﬀshore dredge spoil disposal at the Columbia River and Grays Harbor
estuary mouths (Sherwood et al., 1990; Kaminsky et al., 2010). Sediment retention rates from numerous impoundments of the Columbia
River tributaries (1950s to present) are not established. Although these
anthropogenic impacts are important to regional sediment management
practices, they do not clarify the relations of shoreline responses to
changing marine accommodation spaces resulting from relative SLR, so
they are not further addressed in this article.
Littoral sand supplies to the large marine-dominated estuaries,
Willapa Bay and Grays Harbor, responded to the neotectonic cycles of
coseismic subsidence and interseismic uplift. For example, post-subsidence accumulations of littoral sand layers at 13 tidal ﬂat sites in
Willapa Bay (Fig. 4 Parts B and C) averaged 0.5 m thickness for 30 postsubsidence sequences (Peterson and Vanderburgh, 2018a). It is assumed that most of the littoral sand that was transported into the

Table 1
Late-Holocene rates of littoral sand supply and accumulation (sinks) in the
CRLC system.
Settings

Surface area
(km2)

Littoral sand
supply (×106
m3 ka−1)

Littoral sand sink
(×106 m3 ka−1)

Columbia River
throughput
Prograded shoreface
Willapa Bay
Grays Harbor
Inner-shelf
Total

–

5000 *3200

–

300
347
236
1600
–

–
–
–
–
–

860
110
650
2100
3720

Notes: The estimated rate of Columbia River sand throughput to the littoral
zone (5000 × 106 m3 ka−1) is based on decreasing sedimentation rates in the
Lower Columbia River Valley between middle and late Holocene time (Peterson
et al., 2013). A second method (*) of estimating Columbia River sand
throughput to the marine side in Holocene time is based on the total net accumulation rate, as summed from all of the measured CRLC sand sinks (total is
3720 × 106 m3 ka−1). Littoral sand sinks or net accumulation rates in the
prograded barriers, beach plains, and the underlying upper-shoreface deposits
(7–0 ka) are from Herb (2001) and Peterson et al. (2010a). Littoral sand sinks or
accumulation rates, in Grays Harbor averaged about 650x106m3 ka−1 during
Holocene time (Peterson and Phipps, 2016). Littoral sand accumulation rates in
Willapa Bay are only known for latest-Holocene time (110 × 106 m3 ka−1)
(Peterson and Vanderburgh, 2018a). Littoral sand accumulation rates in the
inner-shelf (0–50 m water depth or ~ 10 km inner-shelf width) are based on
Holocene (12–0 ka) ﬁll rates above the marine transgressive ravinement surface
(Peterson et al., 2016).
5

Marine Geology 427 (2020) 106272

C.D. Peterson, et al.

orientation and continuity of the proposed timelines were veriﬁed by
semi-continuous shore-parallel foredune segments, as shown in Lidar
digital elevation models (Lidar DEM), where the foredune segments did
not locally bifurcate, converge, or oﬀset (Linde, 2014). The mostly-orthogonal intersections of the across-shore GPR proﬁles with shore-parallel timelines generally formed rectangular polygon cells (Fig. 5 Part
A). However, more complex cell boundaries occur at tidal inlets. The
intersecting GPR proﬁles, shore-parallel timelines, and tidal inlet
shorelines were used to map 247 polygon cells in the prograded barriers
and beach plains of the CRLC system (Linde, 2014).
Though retreat scarps are locally inclined or seaward dipping
(Fig. 3), the polygon cell timeline-boundaries are extended vertically
downward (Fig. 5 Part B). The vertically projected timelines form
straight sided polygon cells that simplify cell volume calculations.
Average polygon cell heights (top surface elevations), including abandoned foredunes and interdune valleys, were established by ArcGIS
averaging of bare-ground Lidar digital elevation models (Lidar DEM).
The bare-ground Lidar datasets, including thirty-two 1/9 arc sec NED
ﬁles (Gesch et al., 2002) as surveyed from 2001 to 2005, were stitched
together in ArcGIS to form one seamless base ﬁle (Linde, 2014). The top
surface averaging accounts for all irregularities associated with locally
developed foredune ridge and intervening valley topographies. The
polygon cell bottoms (basal elevations) are based on projected paleobeach toe elevations, starting with the modern beach toe (0 ka) at 0 m
NAVD88. The prehistoric beach toe elevations were estimated for corresponding timelines, based on the latest-Holocene rate of sea level
change (1.0 m ka−1) (Fig. 2). The polygon cell bottom elevation is
taken as an average of the two paleo-beach toe elevations that correspond to the two bounding retreat scarp timelines (Table 3). For purposes of bounding the onset location and age of the most-landward
polygon cells, the timeline preceding the ﬁrst recorded retreat scarp is
mapped at 1) the barrier spit bay shoreline or 2) the beach plains backedge or base of hill slope (Linde, 2014). The most-seaward extents of
the historic polygon cells (1700 to modern) are mapped at the 0 m
NAVD88 elevation datum (Lidar DEM) or the modern beach toe
(Ruggiero et al., 2005). Where sequential retreat scarps or timelines are
missing in a GPR transect, possibly due to subsequent subsidence event
erosion, the two closest preserved retreat scarps are used as bounding
timelines, as shown in Table 3. Polygon cell thicknesses (average top
-to- average bottom elevations) were multiplied by polygon cell surface
areas to yield accreted sand volumes in the polygon cells. Sand grainsize data from numerous borehole samples in the CRLC barrier and
beach plains are provided in Woxell (1998), Herb (2000), and Peterson
et al. (2010a, 2010b).
Each of the 247 mapped polygon cells (Fig. 6) was assigned a centroid (ArcGIS calculated center point) with position coordinates
(NAD83 UTM sector 10 N) in a MSAccess™ relational database. Additional polygon attributes were linked to the centroids in the database
tables (Linde, 2014). The centroid attributes include UTM east and
UTM north (m), mean elevation (m), minimum age (yr BP), maximum
age (yr BP), mean centroid age (yr BP), surface area (m2), average
thickness (m), and volume (m3). The attributes for all 247 centroids in
the four subcells (North Beaches, Grayland Plains, Long Beach, and
Clatsop Plains) are summarized in four spreadsheet ﬁles (CentroidData.xls), as presented in Supplementary Data. Error estimates are
as follows: retreat scarp position, as projected vertically ( ± 25 m in
UTM NAD83), retreat scarp (timeline) age ( ± 75 yr), polygon cell
centroid position ( ± 10 m), polygon cell elevations ( ± 1 m), as reported by Linde (2014).
In re-analysis of shoreline retreat estimates resulting from past
episodes of SLR in the CRLC beaches (Doyle, 1996) it became apparent
that a means was needed by which to apportion future sand volume loss
from the modern shorelines following large (2–3 m) values of SLR. Such
associated retreat distances could exceed several hundred meters in the
existing barrier and beach plain widths. A guide is needed to predict
what sequences of barrier and beach plain sand losses would occur in

of Willapa Bay have not been established, but they could be substantial
as they represent about 45% of the surface area in Willapa Bay. The
neotectonic cycles of coseismic subsidence and interseismic uplift are
also recorded in buried wetland sequences around the margin of the
Columbia River estuary (Peterson and Cruikshank, 2014). Such buried
wetland sequences were not preserved in the interior areas of the Columbia River estuary due to frequent channel lateral migrations and
deposit reworking (Peterson et al., 2014). The roles of short-term episodic submergence and emergence in controlling river sand accumulation or bypassing in the Columbia River estuary are not known.
However, increased accommodation space for river sand in the estuary,
following episodic subsidence (Fig. 2 Part B), could have temporarily
reduced river sand throughput to the littoral zone (see further discussion in Section 5.5).
Frequent storm-wave reworking of oﬀshore deposits in the innershelf by large storm surf (peak wave height 10–12 m) (Ruggiero et al.,
1997) apparently precluded any stratigraphic evidence of neotectonic
forcing of littoral sand accumulation in the inner-shelf (Kaminsky,
2006). Nevertheless, the inner-shelf served as 1) a short-term reservoir
of the littoral sand that was removed from the beaches following coseismic subsidence events and 2) the major long-term sink of littoral
sand (2100 × 106 m3 ka−1) in the CRLC system during Holocene time
(Table 1) (Twichell et al., 2010; Peterson et al., 2016). Although some
littoral sand was lost from the inner-shelf to the middle-shelf (> 50 m
water depth) and from the southern Washington shelf to the northern
Washington shelf (Fig. 1), during early- and middle-Holocene time
(Peterson et al., 2010a) the inner-shelf was likely a more eﬀective trap
of Columbia River sand delivered to the CRLC system during late-Holocene time (Peterson et al., 2016).
3. Methods
Previously reported beach-retreat scarps (160 in number) (Peterson
et al., 2010b) in shore-normal GPR proﬁles (Fig. 1) were connected by
straight or simple curved lines to form shore-parallel timelines in the
CRLC prograded barriers and beach plains (Linde, 2014). Including the
modern shoreline (0 m NAVD88 elevation contour), a total of 11
timelines, ranging from 0 to 5.0 ka, were constructed in the CRCL
subcells (Table 2). The 0 m NAVD88 datum (time 0 ka) is approximately 1.0 m below modern mean sea level (MSL) in the study area. The
Table 2
Retreat scarp time lines in the CRLC system.
Retreat scarp
timeline

Estimated age
(ka)

Estimated subsidence
(m)

Retreat scarp
size (m)

Modern
A
B
C
D
E
F
G
H
I
J*

0
0.3
1.1
1.3
1.7
2.5
2.8
3.2
4.0
4.7
5.0

–
1.5
1.0
0.5
2.0
0.5
1.0
1
–
–
–

–
5–10
3–7
~3
8–12
~3
5–10
5–10
3–7
5–10
3–7

Notes: Along-shore timelines are constructed between corresponding catastrophic beach retreat scarps (letters), which have been previously assigned to
approximate age (ka) and estimated coseismic subsidence (m) (Peterson et al.,
2010b). Retreat scarp size (m) is taken (vertically) from the most landward
(shallowest) truncated beach strata elevation to the most seaward (deepest)
truncated beach strata or heavy-mineral (placer lag) deposit. Retreat scarp
vertical extents (minimums based on GPR reﬂections) are interpreted from 79
GPR lines comprising 35 across-barrier/beach plain proﬁles in the CRLC system
(Peterson et al., 2010b). The J timeline (*) is arbitrarily mapped along the backedges of the prograded barrier-spits and beach plains, thereby possibly overrepresenting volume accretion during the ‘assumed' 4.7–5.0 ka time interval.
6

Marine Geology 427 (2020) 106272

C.D. Peterson, et al.

Fig. 5. Part A. Map view of polygon cell (shaded) deﬁned by retreat
scarps (solid squares) in GPR transects (bold line) and digitized
timelines between corresponding retreat scarps. The timelines are
guided by, or digitized parallel to, continuous abandoned foredune
ridges, which are assumed to represent shoreline orientations prior to
foredune abandonment. The center of the mapped polygon cell is assigned a unique centroid (solid circle), with deﬁned position coordinates, and corresponding polygon cell attributes imported into a
relational database (Microsoft Access™). Part B. Cross-section of a
polygon cell (boxed) deﬁned by vertically-extended retreat scarp
timelines (dashed lines), an averaged surface elevation (bold line)
from ArcGIS analysis of Lidar digital elevation models (LDEM), and an
averaged basal depth (bold line) from the average of timeline intersections with a paleo-sea level curve. The sea level curve (gradient
1.0 m ka−1) is normalized to the modern beach toe at 0 m NAVD88
(Linde, 2014), so it represents paleo-beach toe elevations at dated
retreat scarps or timelines. Vertical exaggeration is 50:1.

presented below in the Results Sections (4.1–4.5). These results are then
used in combinations with estimated sand sinks in the inner-shelf and
large marine-dominated estuaries to predict shoreline erosion sequences, as presented below in the Discussion sections (5.1–5.6).

Table 3
Mean ages and projected basal depths of polygon cell centroids.
Timeline interval

Polygon cell age (years)

Calculated basal depth (m
NAVD88)

Modern/A
A/B
A/C⁎
B/C
B/D⁎
C/D
D/E
E/F
E/G⁎
F/G
G/H
F/I⁎
H/I
I/J

150
700
800
1200
1400
1500
2100
2650
2850
3000
3600
3750
4350
4850

−0.15
−0.70
−0.80
−1.20
−1.40
−1.50
−2.10
−2.65
−2.85
−3.00
−3.60
−3.75
−4.35
−4.85

4. Results
In this section, the CRLC prograded barrier spits and beach plains
are divided into 247 polygon cells, as based on across-shore GPR
transects and age-correlative catastrophic beach retreat scarps. The
polygon cells are analyzed for measured attributes including center
(centroid) position coordinates, mean ages, and accreted sand volumes.
4.1. Mapped timelines in the CRLC barrier spits and beach plains
Modern foredunes that front the CRLC barrier spits and beach plains
develop at the furthest landward distance of winter (storm) wave runup
at the back-edge of the beach backshores. The modern foredune ridge
crests are uniformly shore-parallel, both in straight beach segments
(Fig. 7) and in the more complex or curved shorelines, such as those
that occur at the ends of recurved spits. The modern foredunes developed over multiple years to decades so they represent time- and distance-averaged shoreline orientations, rather than seasonally ephemeral shoreline morphologies associated with beach cusps, rip heads,
and/or attached swash bars. Although the modern and late-historic
foredune ridges locally bifurcate, converge, and/or show small lateral
oﬀsets, their distance-averaged orientations are uniformly shore-parallel. Abandoned foredune ridges in the prograded CRLC barriers and

Notes: In rare cases a retreat scarp was missing within a mapped polygon cell
(*), so mean centroid age was determined by the closest existing retreat scarp
ages. The mean age of the polygon cell was used to calculate the basal depth (m
NAVD88) of the polygon cell, based on paleo-beach toe projections from the
late-Holocene sea level change rate (1.0 m ka−1) as shown in Fig. 2 Part B.

the CRLC system. To that end, recent past sequences of dated beach
accretion in the CRLC subcells (Linde, 2014) are used ‘in reverse’ to
evaluate most likely sequences of future sand losses and associated
shoreline retreats in the CRLC shorelines, following future SLR. The
dated sequences of late-Holocene barrier and beach plain accretion are

7

Marine Geology 427 (2020) 106272

C.D. Peterson, et al.

Fig. 6. Potted locations of 247 polygon cell centroids in prograded barrier spits and beach plains are shown for the North Beaches subcell (Part A), Long Beach subcell
(Part B), Grayland Plains subcell (Part C) and Clatsop Plains subcell (Part D). Representative centroids are identiﬁed for timeline intervals and corresponding midpoint ages, including Mod/A at 0.15 ka or historic time (solid circles), D/E at 2.10 ka (solid squares), and H/I at 4.35 ka (solid hexagons). All other centroid intervals
are identiﬁed by open circles. See Table 3 for polygon centroid ages. Centroid position coordinates, mid-point ages, and other attributes for the four CRLC subcells are
presented in four summary databases (CentroidData.xls) in Supplementary Data. UTM northings are in kilometers. See Fig. 1 for the subcell locations in the study
area.

within diﬀerent neotectonic cycles, the preserved foredune ridge sequences (Reckendorf et al., 2001) do not show a one-to-one correspondence with catastrophic retreat scarp sequences. Whereas the
abandoned foredune ridges do not directly correlate by number to great
earthquake cycles in the CRLC system, they do serve as guides to paleoshoreline orientations during net accretionary phases of the barrier
spits and beach plains. Continuous and distinct foredune ridges are used

beach plains are therefore assumed to represent paleo-shoreline orientations during corresponding periods of shoreline stability or modest
accretion, as presented below.
The abandoned foredunes in the prograded CRLC barriers and beach
plains were associated with episodic periods of shoreline stability between events of catastrophic retreat and rapid progradation (Fig. 4 Part
A). However, due to the variations in progradation and retreat distances

8

Marine Geology 427 (2020) 106272

C.D. Peterson, et al.

Fig. 7. Hill-shaded topographic relief map based on bare-ground Lidar DEM showing the modern or late-Historic foredune ridge (bold line at left side) and abandoned
linear foredune ridges in the Clatsop Plains subcell (DELR transect in Fig. 6). Great earthquake retreat scarps (A-F) are plotted along the GPR transect DELR (Peterson
et al., 2010b) that terminates to the east (landward) at submerged interdune valleys (bogs). Continuous distinct linear dune ridges were used to bridge timelines (G,
H, I) to retreat scarps in longer GPR transect extensions (ESTR-OSTR) located to the south of the DELR transect. See Table 2 for ages of retreat scarps and
corresponding timelines.

4.2. Cross-sections in the CRLC barrier spits and beach plains

infrequently in this study to 1) bridge between longer GPR transects
where shorter GPR transects were terminated in back barrier bogs
(Fig. 7) and 2) to extend timelines beyond GPR transects where submerged bogs and/or tidal sloughs precluded GPR surveys (Fig. 8). Such
extensions of timelines by abandoned dune ridges were only used in
terminal spit areas, as located adjacent to the tidal inlets of the Columbia River, Willapa Bay, and Grays Harbor estuaries (Linde, 2014).
The modern estuary shorelines (0 ka) are used to bound the polygon
cell timelines that terminate at tidal inlets, as shown in Fig. 8.

Selected across-shore proﬁles of the CRLC barrier spit and beach
plain surfaces, as based on Lidar DEMs (Linde, 2014), are plotted together with corresponding paleo-beach toe basal elevations to construct
representative cross-sections of the prograded beach and dune deposits
(Figs. 9 and 10). The approximate positions of the last coseismic catastrophic retreat scarps, representing timeline ‘A' at 1700 (Fig. 4 Part A),
are shown, as projected to the high-resolution Lidar DEM proﬁles from

9

Marine Geology 427 (2020) 106272

C.D. Peterson, et al.

Fig. 8. Hill-shaded topographic relief map based on bare-ground Lidar DEM showing multiple late-Historic foredune ridges (located seaward of the ‘A' timeline) and
abandoned pre-historic foredune ridges (located landward of the ‘A' timeline) in the northern part of the Clatsop Plains subcell (GPR transects CAMP and PERK in
Fig. 6). The modern estuary shoreline (bold line) at the south side of the Columbia River mouth is used as an arbitrary boundary for the timeline terminations.
Reported retreat scarps (A-I) are plotted along the two GPR transects (CAMP and PERK) that crossed the full width of the prograded beach plain (Peterson et al.,
2010b). Submerged bogs and tidal sloughs precluded GPR surveying of the northernmost end of the Clatsop Plains. However, distinct semi-continuous foredune
ridges were used to extend corresponding timelines to the Columbia River tidal inlet (Linde, 2014). See Table 2 for ages of retreat scarps and corresponding timelines.

development in the two North Beach cross-sections (A and B). In contrast, most of progradation in the low-elevation Long Beach barrier spit
(averaging ~6 m elevation) occurred during pre-historic time (pre1700). The net progradation began at about 5 ka and 3 ka, respectively,
at the south and north ends of the Long Beach barrier spit (Vanderburgh
et al., 2010; Peterson and Vanderburgh, 2018a). The Grayland Plains

the nearest GPR transects (Peterson et al., 2010b). In the North Beaches
subcell, the only substantial shoreline progradation (> 500 m) occurred in the north spit of the Grays Harbor estuary (Fig. 6). Most of
that progradation occurred during early historic time (early 1900s),
following jetty construction and shoreline realignment (Kaminsky et al.,
2010). The very-rapid historic progradation largely precluded foredune

10

Marine Geology 427 (2020) 106272

C.D. Peterson, et al.

Fig. 9. Across-shore cross-sections of prograded barrier spits in the North Beaches and Long Beach subcells. Cross-sections include 1) top surface, 2) 0 m NAVD88
datum at the modern beach toe (dashed line), and the elevations of paleo-beach toes (dotted line) as projected landward from a sea level change of 1 m ka−1 in lateHolocene time (Fig. 2). The positions of 1700 retreat scarp or timeline A (solid square) are approximated from the nearest GPR transects (Peterson et al., 2010b).
Vertical exaggeration for all cross-sections is 10:1. See Fig. 6 for the cross-section locations in the North Beaches and Long Beach subcells.

4.4. Summed sand volumes within paired timelines for the four CRLC
subcells

are also relatively low in surface elevation, though two large abandoned foredunes (10–15 m elevation) occur in the northern Grayland
Plains cross-section (A). By comparison to the northern subcells in the
CRLC system, the Clatsop Plains are characterized by abundant large
foredunes, reaching 20–30 m in elevation. A combination of prolonged
backshore sand supply and modest rates of net shoreline progradation
permitted the development of large foredunes in the Clatsop Plains
subcell, located south of the Columbia River mouth.

The summed volumes of sand accretion within bounding, or paired,
timelines for each of the four CRLC subcells are presented in Table 5.
The accreted sand volumes for the youngest prehistoric interval,
bounded by the A/B timelines (0.3–1.1 ka), are 21.5 × 106 m3 (Clatsop
Plains), 22.1 × 106 m3 (Long Beach), 18.8 × 106 m3 (Grayland Plains),
and 9.4 × 106 m3 (North Beaches). These volumes are modest by
comparison to the historic interval (Modern/A), which are greater than
the A/B interval volumes by some 5–10 times (5–10×). The next older
prehistoric interval, B/C at 1.1–1.3 ka, is represented by slightly larger
volumes in the Clatsop Plains (> 31.1 × 106 m3), Long Beach
(760.0 × 106 m3), and Grayland Plains (25.6 × 106 m3), even though
the B/C interval of 0.2 ka) is four times shorter than the A/B interval of
0.8 ka (Table 2). The A/B interval included two large coseismic events,
and associated shoreline retreats, but the ‘C' retreat scarps are amongst
the smallest recorded in the CRLC (Table 2). These relations point to the
importance of both the timing (duration) and scale (volume) of onshore/oﬀshore transport reversals, in controlling net beach accretion
volumes over short time intervals. The variability in Columbia River
sand throughput to the littoral zone over such short time scales (century) is not known, but it could also be inﬂuenced by the neotectonic
cycles of coseismic subsidence and interseismic uplift. Averaging beach
volume accretion data over multiple time intervals reduces the eventresponse resolution, but it clariﬁes the relative eﬀectiveness of the four

4.3. Polygon cell attributes in the CRLC barrier spits and beach plains
Centroid locations, bounding timeline ages, and volume accretion
data, are established for 247 mapped polygon cells in the CRLC prograded barrier spits and beach plains (Fig. 6). The number of digitized
polygon cells in the four subcells range from 43 (Grayland Plains) to 97
(Long Beach) (Table 4). Average cell surface areas range from
707,833 m2 (North Beaches) to 1,078,018 m2 (Clatsop Plains). Average
cell thickness ranges from 6.8 m (North Beaches and Grayland Plains)
to 12.3 m (Clatsop Plains). The calculated averages of cell sand volumes
are as follows; 11,270,668 m3 (Clatsop Plains), 7,251,318 m3 (Long
Beach), 5,266,772 m3 (Grayland Plains), and 3,894,488 m3 (North
Beaches).

11

Marine Geology 427 (2020) 106272

C.D. Peterson, et al.

Fig. 10. Across-shore cross-sections of beach plains in the Grayland Plains and Clatsop Plains subcells. Cross-sections include 1) top surface, 2) 0 m NAVD88 datum at
the modern beach toe (dashed line), and 3) the elevations of paleo-beach toes (dotted line) as projected landward from a sea level change of 1 m ka−1 in lateHolocene time (Fig. 2). The positions of 1700 retreat scarp or timeline A (solid square) are approximated from the nearest GPR transects (Peterson et al., 2010b).
Vertical exaggeration for all cross-sections is 10:1. See Fig. 6 for the cross-section locations in the Grayland Plains and Clatsop Plains subcells.

measured accumulation by interval duration (yr), to yield shorelinenormalized volume accretion rates (m3 m−1 yr−1) (Table 6, Fig. 11).
The oldest timeline interval I/J is only represented in the Long Beach
subcell, where the very-high rate there (23.5 m3 m−1 yr−1), might
reﬂect some uncertainty about the back-edge age of the prograded
barrier spit. A similar uncertainty exists for the timeline interval (E/F)
in North Beaches subcell and in the rapidly-prograded Columbia River
spit in the Clatsop subcell. With those exceptions, and the anomalouslynarrow A/B intervals in all four subcells, the shoreline-normalized accretion rates generally increase through time in the CRLC prograded
barriers and beach plains.
Some striking aspects of the dated shoreline progradations in the
CRLC system are the late onset ages of accretion at timeline intervals F/
G (3.2–2.8 ka) and E/F (2.8–2.5 ka), respectively, in the two northernmost subcells, North Beaches and Grayland Plains (Table 6, Fig. 11).

subcells in competing for littoral sand supplies in the CRLC system. The
longest time interval represented by net barrier and beach plain progradation in each subcell yields total beach-sand accretion volumes of
597.3 × 106 m3 (Clatsop Plains), 703.4 × 106 m3 (Long Beach),
231.7 × 106 m3 (Grayland Plains) and 210.3 × 106 m3 (North Beaches). The total beach/dune sand accretion volume for the summed
subcells during late-Holocene time (5–0 ka) is 1742.8 × 106 m3. See
Tables 2 and 4 for subcell ages.

4.5. Average shoreline-normalized volume accretion rates in the four CRLC
subcells
The relative eﬀectiveness of littoral sand accumulation in the four
CRLC subcells is addressed from normalizing 1) the timeline interval
accreted sand volumes by subcell alongshore distance (m) and 2) the
Table 4
Summary of polygon cell parameters for the four CRLC sub-cells.
Sub-cell Name

Retreat scarps (timelines) present

Number of polygon cells

Average cell surface area (m2)

Average cell thickness (m)

Average cell volume (m3)

Clatsop Plains
Long Beach Peninsula
Grayland Plains
North Beaches

A-J
A-J
A-F
A-G

53
97
43
54

1,078,018
907,870
805,486
707,833

12.3
8.4
6.8
6.8

11,270,668
7,251,318
5,266,772
3,894,488

Notes: Ages of coseismic catastrophic retreat scarps or time lines ‘A–J' are presented in Table 2. Data are summarized from centroid database ﬁles for the Clatsop
Plains, Long Beach, Grayland Plains, and North Beaches subcells (see CentroidData.pdf ﬁles in Supplementary Data).
12

Marine Geology 427 (2020) 106272

C.D. Peterson, et al.

interval (Fig. 11) remain unknown.

Table 5
Volume totals (km3) for each of the timeline intervals within the CRLC subcells.
Timeline
interval

Modern/A
A/B
A/C
B/C
B/D
C/D
D/E
E/F
E/G
F/G
G/H
F/I
H/I
I/J
Total (km3)

Clatsop
Plains
(×106 m3)

Long
Beach
(×106
m3)

Grayland
Plains (×106
m3)

North
Beaches
(×106 m3)

Total
volume
(×106 m3)

109.5
21.5
N.A.a
31.1
62.9
16.2
87.4
119.0
N.A.a
54.8
50.6
8.5
35.8
N.A.a
597.3

144.9
22.1
N.A.a
76.0
N.A.a
67.4
84.0
52.0
0.2
34.5
83.7
N.A.a
36.8
101.8
703.4

77.7
18.8
N.A.a
25.6
N.A.a
35.9
28.5
45.3
N.A.a
N.A.a
N.A.a
N.A.a
N.A.a
N.A.a
231.7

151.4
9.4
6.5
4.2
N.A.a
12.4
15.0
8.2
N.A.a
3.1
N.A.a
N.A.a
N.A.a
N.A.a
210.3

483.5
71.8
6.5
136.9
62.9
131.9
214.8
224.6
0.2
92.3
134.3
8.5
72.6
101.8
1742.8

5. Discussion
In this section, the evolutions of the CRLC barriers and beach plains
are explained in terms of abundant sand supply from the Columbia
River mouth, local shoreline conditions relative to directional wave
forcing, and competing accommodation spaces in the large estuarine
and inner-shelf sand sinks. Modeling studies of across-shelf transport
processes have been completed for other barrier systems (LorenzoTrueba and Ashton, 2014; Ortiz and Ashton, 2016), but they diﬀer from
those developed for the CRLC system (Sternberg, 1986; Kachel and
Smith, 1986), which incorporate both wave dynamics and geostrophic/
downwelling or underﬂow currents to transport littoral sand across the
inner-shelf. In this article, the relative importance of the inshore and
oﬀshore marine sinks for littoral sand accumulation in the high-waveenergy CRLC system are evaluated empirically for predicted accelerations of future SLR. Estimates of future shoreline retreats are based on
the loss of destabilized beach sand to the more competitive marine sand
sinks following predicted accelerations of SLR from ongoing global
warming.

Notes: Volumes of littoral sand accretion (106 m3) in subaerial beach and dune
deposits (≥ 0 m NAVD88) are shown for corresponding timeline intervals
(summed polygon cell volumes) for each of the four CRLC subcells. Data are
from CentroidData.xls ﬁles in Supplementary Data.
a
N.A. – not applicable, timeline interval does not exist within the sub-cell.

5.1. Late-Holocene evolution of the barrier and beach plains in the CRLC
subcells
The relative timing and magnitude of littoral sand partitioning in
the four CRLC subcells (Table 7) are related, in part, to 1) local conditions of shoreline orientations and 2) interceptions of nearshore sand
transport by inshore sand sinks in the large marine-dominated estuaries, Willapa Bay and Grays Harbor (Fig. 12). The alongshore limits of
late-Holocene littoral sand distribution, or onshore accommodation
space, in the CRLC accreted barriers and beach plains are generally
controlled by paleo-shoreline orientations. The regional net-northward
longshore transport in the CRLC system is reversed in the Clatsop Plains
and North Beaches subcells, due to changes in shoreline orientations
relative to the west-southwest-west dominant direction of nearshore
wave energy. Northwest-southeast trends of latest-prehistoric shorelines located south of 1) the Columbia River delta/spit in the Clatsop
Plains subcell (paleo-shoreline bearing 343°TN) and 2) the Point
Grenville headland in the North Beaches subcell (350°TN) permit a net
southward nearshore transport in those two subcells. The localized
southward nearshore transport in the Clatsop Plains subcell is established by Columbia River sand mineralogy in prograded beach plain
deposits accreted between the PERK and ESTR transects in the Clatsop
Plains subcell (Peterson et al., 2010a). The westward progradation of
the Columbia River mouth barrier/delta in late-Holocene time (Fig. 7)
promoted southward littoral transport from the Columbia River mouth,
yielding the high beach and dune facies accretion rate of 27.4 × 106 m3
ka−1 in the Clatsop subcell during late-Holocene time (Table 7).
However, the shoreline ‘change’ to a northeast-southwest orientation
(28°TN) between the ESTR proﬁle and Tillamook Head, after 2.5 ka
(Peterson et al., 2010c), severely limited littoral sand transport and net
deposition, south of the ESR transect in latest-Holocene time.
A net southward onshore/nearshore littoral transport in the North
Beaches subcell is indicated by gravel clasts that were eroded from latePleistocene deposits in sea cliﬀs (Vanderburgh et al., 2010), located
between Grenville Point and the COPR transect (Fig. 12). The eroded
sea cliﬀ gravels were transported south, under traction transport in
onshore and nearshore settings, to the Grays Harbor tidal inlet, and
rarely, to the southernmost Grayland Plains proﬁle WARR (paleoshoreline bearing 352°TN), but not further south along the Long Beach
Peninsula (2° TN) (Peterson et al., 2010a). A net southward transport of
gravel, and presumed accompanying beach sand, in the North Beaches
subcell could account, in part, for the modest accretion rates
(11.7 × 106 m3 ka−1) in that subcell (Table 7). The apparent southward nearshore/onshore transport in the two northernmost subcells

Table 6
Mean shoreline-normalized volume accretion rates in the CRLC subcells.
Interval

Clatsop Plains
(m3 m−1 yr−1)

Long Beach
(m3 m−1 yr−1)

Grayland
Plains
(m3 m−1 yr−1)

North Beaches
(m3 m−1 yr−1)

I/J
H/I
G/H
F/G
E/F
D/E
C/D
B/C
A/B
Modern/A

–
2.60
2.87
4.96
15.05
5.15
4.14
19.53
0.89
12.49

23.52
2.21
3.34
2.87
5.93
3.12
4.70
12.16
0.80
12.45

–
–
–
0.90
7.45
1.96
4.80
6.17
1.12
10.68

–
–
–
–
1.65
1.84
2.53
3.13
0.56
12.38

Notes: Estimated shoreline volume accretion (m3) between deﬁned timelines
(Interval) are normalized for shoreline distance (m−1) and time (yr−1) in the
four CRLC subcells to yield mean shoreline-normalized volume accretion rates
(m3 m−1 yr−1). Polygon cell sand volumes, as averaged for each of the four
CRLC subcells, are from Table 5. Timeline interval age spans are from Table 2.
Data are graphically plotted below in Fig. 11.

By comparison, the Long Beach subcell began accretion, at GPR transect
PION, with the I/J interval (5.0–4.7 ka), and the Clatsop Plains subcell
began accretion, at GPR transect PERK, with the H/I interval
(4.7–4.0 ka). Those earliest onset ages of net progradation are associated with positions located next to the mouth of the Columbia River,
the source of littoral sand supply to the CRLC system (Fig. 1). But
alongshore distances of beach segments from the Columbia River
mouth do not entirely account for the diﬀerent beach-sand volume
accretion rates. Though the Grayland Plains subcell is the second most
distant subcell from the Columbia River mouth, the highest rate of
shoreline-normalized accretion in the C/D interval (4.8 m3 m−1 yr−1)
occurred in the Grayland Plains subcell. At the regional scale, the onset
ages of shoreline progradation in the CRLC system are likely controlled
by competing accommodation space in the inner-shelf (Peterson et al.,
2016). But at the subcell scale, both shoreline orientations and estuarine sand sinks played roles in controlling the local shoreline progradations, as discussed below. The origin(s) of the very-small values of
shoreline normalized accretion rates in all four subcells during the A/B

13

Marine Geology 427 (2020) 106272

Mean shore-normalized accretion rate
(m3 mˉ1 yrˉ1)

C.D. Peterson, et al.

Clatsop Plains

25

Long Beach

KEY

Grayland Plains

20

North Beaches

15
10
5

0
I/J

H/I

G/H

F/G
E/F
D/E
Timeline Intervals

C/D

B/C

A/B

Fig. 11. Mean shoreline-normalized volume accretion rates (m3 m−1 yr−1) are plotted for established timeline intervals (paired letters) in the four CRLC subcells.
Data are from Table 6. Very-high accretion rates for the I/J interval in the Long Beach subcell are suspect due to uncertain dating of the back-edge of the prograded
barrier spit/beach plain.

Harbor, inﬂuenced the late onset ages (2–3 ka) of net shoreline progradation in the northern end of the Long Beach subcell, the Grayland
Plains subcell, and the north spit of Grays Harbor (Table 7; Fig. 12).
Littoral sand accumulation rates in Willapa Bay (> 110 × 106
m3 ka−1) and Grays Harbor (117 × 106 m3 ka−1) (Peterson and
Vanderburgh, 2018a, 2018b) are about one order of magnitude (10×)
larger than the corresponding mean accretion rate in the North Beaches
subcell (11.7 × 106 m3 ka−1), and at least 5× larger than the mean
accretion rates in the Grayland subcell (15.9 × 106 m3 ka−1) and the
Long Beach subcell (19.8 × 106 m3 ka−1). As will be addressed later in
this article, the competitive roles of these large estuaries for littoral
sand accumulations should be signiﬁcant under conditions of potential
future accelerated SLR in the CRLC system.

Table 7
Late-Holocene mean accretion rates for the four CRLC subcell barriers/beach
plains and the large marine-dominated estuaries.
Subcell/estuary

Littoral sand
volume (m3)

Onset age of
beach accretion
(ka)

Mean accretion rate
(×106 m3 ka−1)

Clatsop Plains
Long Beach
Grayland Plains
North Beaches
Willapa Bay
estuary
Grays Harbor
estuary

5.97 × 108
7.03 × 108
2.32 × 108
2.10 × 108
–

4.7
5.0
2.8
2.5
–

27.4
19.8
15.9
11.7
> 110

–

–

117

Notes: Barrier and beach plain sand volume (m3) includes all intervals from the
onset of net accretion to the modern shoreline. Onset age of net accretion is
taken to be the earliest polygon cell bounding age or timeline from at least two
GPR transects within each subcell. Late-Holocene accretion rates (m3 ka−1) of
littoral sand are the averaged volume accretion rate as normalized to 1000 yr
(ka) for all polygon cells within each subcell, since the onset of subcell progradation (Linde, 2014). Measured littoral sand accumulation rates (m3 ka−1)
are also shown for Willapa Bay (Peterson and Vanderburgh, 2018a) and Grays
Harbor (Peterson and Vanderburgh, 2018b), as based on the latest-Holocene
time period (3–0 ka). Littoral sand accumulation rates in Willapa Bay do not
include subtidal channel settings, so the accumulation rate of littoral sand in
Willapa Bay is a minimum value.

5.2. Littoral sand accumulation rates in the CRLC inner-shelf during latestHolocene time
Inner-shelf sand deposits have been vibracored (1–5 m depth subsurface) and 14C dated from sites in the CRLC inner-shelf (Fig. 1)
(Kaminsky, 2006), as summarized in Peterson et al. (2016). Bedrock or
transgressive gravel deposits occur at the surface in some northernmost
inner-shelf localities (Twichell and Cross, 2001; Twichell et al., 2010),
which precluded vibracoring in those localities. Sedimentation rates in
those settings are based on seismic reﬂection proﬁling at identiﬁed
stations (Peterson et al., 2016). For this article, selected site and core
interval data, including subsurface depths and ages, are summarized in
a database ﬁle ShelfSedRates.pdf, as provided in Supplementary Data.
The selected vibracore data are used to establish representative sedimentation rates from the 14C dated vibracore intervals (n = 56) in
representative core/station sites (n = 35) from selected across-shelf
transects (n = 9) (Fig. 13). Sedimentation rates are computed for sand

contrasts with the northward transport of Holocene Columbia River
derived sand (hypersthene-rich) in the adjacent inner-shelf, under
conditions of orbital wave resuspension and northward geostrophic
ﬂow (Peterson et al., 2016).
The large marine-dominated estuaries, Willapa Bay and Grays
14

Marine Geology 427 (2020) 106272

C.D. Peterson, et al.

Fig. 12. Littoral sand sinks, as measured by littoral sand accretion or accumulation rates, are shown for 1) the prograded Clatsop Plains, Long Beach barrier, Grayland
Plains, and North Beaches, 2) the large marine dominated estuaries, Willapa Bay and Grays Harbor, and 3) the inner-shelf areas (0 m to ~40 m water depth) that
correspond to the four CRLC subcells (Fig. 1). Line widths approximate littoral sand accumulation rates (106 m3 ka−1) and line lengths represent the duration of
measured accumulation (time line 5–0 ka). Shoreline bearings are shown in degrees true north (°TN). Littoral sand net-transport directions/magnitudes (arrows) are
shown for the nearshore and the inner-shelf (bathymetric contours at 20 m intervals). Nearshore dominant wave direction (large arrow) is inferred from beach gravel
clast transport. Littoral sand accumulation rates are from Tables 1, 7, and 8.

values are averaged across each transect to yield averaged shelf transect
sedimentation rates (Table 8). Averaged shelf-transect sedimentation
rates range from 3.41 m ka−1 in transect SL33 to 0.17 m ka−1 in
transect SL3. The vibracore/station transects are also evaluated for
across-shelf gradient, average sand-ﬁlled cross-section area per
1000 yr (1 ka), and alongshore length between transect mid-points or
subcell boundaries. These data are used to compute the sand-ﬁlled
volume accumulation rate for each transect during latest-Holocene time

deposits from ~10 m to 45 m water depths, or from the nearshore to
~10 km distances oﬀshore. Interval sedimentation rates ranged from
11.3 m ka−1 in vibracore site LB 306 (31 m water depth) in transect
SL33, located just north of the Columbia River ebb tide delta, to
0 m ka−1 in all seismic reﬂection stations (> 8 m water depths) in
transects SL7 and SL3, located just south of Point Grenville. The 14C
dated vibracore intervals and seismic reﬂection station sedimentation
rates are averaged for each core site, as shown in Fig. 13, and those
15

Marine Geology 427 (2020) 106272

C.D. Peterson, et al.

Fig. 13. Inner-shelf seismic lines (SL), wave cut platform sites, as exposed at the surface (solid circles), and oﬀshore vibracore sites (open circles) are shown for the
CRLC system. Mean sedimentation rates (m ka−1) for exposed platform sites and vibracores in corresponding inner-shelf seismic line/transects are calculated from
Twichell and Cross (2001), Kaminsky (2006), and Peterson et al. (2016). Inner-shelf sediment (sand) thickness and radiocarbon ages are compiled in a summary ﬁle
ShelfSedRates.pdf, as presented in Supplementary Data. Shelf sand compositions, including percent sand fraction (generally > 90% by weight) and dominant grain
size distributions (medium to ﬁne), are provided in Twichell and Cross (2001), Kaminsky (2006), and Peterson et al. (2016). UTM northings are in kilometers.

Holocene time (Fig. 14) (ShelfSedRates.pdf in Supplementary Data).
The mean sedimentation rate in transect SL33 is excluded from this
analysis due to the anomalous high sedimentation rates in its deeperwater sites (31–41 m water depth). Those sites are directly associated
with northward littoral sand supply from the adjacent Columbia River
ebb tide delta (Fig. 13). The inﬁlling of available accommodation space
by abundant littoral sand supply in the inner-shelf eventually decreased
the across-shelf gradient in late-Holocene time. The shallowing gradient
increasingly favored shoreward sand transport across the inner-shelf,

(3–0 ka).
5.3. Inner-shelf sand sedimentation rates versus across-shelf transect
gradients
Mean sedimentation rates from 14C dated vibracore sand intervals
(Kaminsky, 2006; as summarized in Peterson et al., 2016) in eight
across-shelf transects are plotted against corresponding transect gradients to establish a positive correlation (R2 = 0.78) during latest16

Marine Geology 427 (2020) 106272

C.D. Peterson, et al.

Table 8
Averaged inner-shelf sedimentation rates and volume accumulation rates.
Transect

Elevation diﬀ. (m)

Distance (m)

Grad. (%)

Sed. rate (m ka−1)

Cross-area rate (m2 ka−1)

Length (m)

Acum. rate (×106 m3 ka−1)

SL23
SL33
SL36
SL41
SL43
SL12
SL11
SL7
SL3

39.5
47.6
33.6
44.1
47.4
43.5
36.8
40
40

11,000
11,500
8500
10,800
12,500
12,100
10,700
12,000
14,300

0.36
0.41
0.4
0.41
0.38
0.36
0.34
0.33
0.28

1.07
3.41
2.24
2.16
0.86
1.04
0.7
0.22
0.17

11,770
38,215
19,040
23,328
10,750
12,584
7490
2640
2431

29,000
22,000
32,000
13,500
14,500
9500
7500
10,000
18,000

341
841
609
319
156
119
562
264
437

Notes: Inner-shelf transect data include transect identiﬁcations (Twichell and Cross, 2001; Twichell et al., 2010), transect elevation diﬀerences (m) and across-shelf
distances (m), transect gradient (%), transect averaged sedimentation rate (m ka−1) (ShelfSedRate.pdf in Supplemental Materials), transect cross-sectional area
(vertical accretion) rate (m2 ka−1), transect alongshore length (m) between mid-points or subcell boundaries, and transect volume accumulation rate (x106 m3 ka−1).
The transect across-shelf distances and elevation diﬀerences are taken from paleo-shoreline positions and elevations at 1.5 ka to the estimated depositional surface at
1.5 ka in the terminal (seawardmost) site. The 1.5 ka inner-shelf transect surface represents the depositional mid-point during latest-Holocene time (3–0 ka).

Fig. 14. A plot of mean sedimentation rates (m ka−1) versus across-shelf gradient (%) in corresponding seismic line/vibracore transects in the CRLC system during
latest-Holocene time (Fig. 13). Data are from Table 8, excluding vibracores from seismic line SL33, located immediately north of the Columbia River ebb tide delta.

exceed the rate of late-Holocene SLR. For the central subcells in the
CRLC system the long-term (millennial) depth of closure is taken to be
between 35 and 40 m water depth during latest-Holocene time. For the
purposes of calculating inner-shelf sand accumulation rates following
potential near-future SLR (see Section 5.5) a most conservative depth of
closure (30 m water depth) is used to represent the oﬀshore extent to
which inner-shelf sedimentation rates would match the rate of SLR at
the shorter (one century) time scale.

leading to the upper-shoreface progradation in the Clatsop, Long Beach,
and Grayland subcells. Nevertheless, inner-shelf gradients exceeding
~0.40% in those subcells are associated with sedimentation rates that
continued to keep pace with latest-Holocene rates of net SLR
(~1 m ka−1) to oﬀshore distances of at least 8 km or 30 m water depth,
as interpreted below (Fig. 13). By comparison, the inner-shelf transect
gradients of < 0.35% in the North Beaches subcell generally failed to
accumulate signiﬁcant littoral sand deposits at oﬀshore distances
of > 2 km or > 10 m water depth.
Several vibracores in ~45 m water depths (LB304 in SL36, GL504 in
SL41 and GL507 in SL43) yield sedimentation rates of
0.32–0.45 m ka−1 (ShelfSedRates.pdf in Supplementary Data). These
vibracore sites are from representative transects with moderate innershelf gradients (0.38–0.41%) in the central subcell areas of the CRLC
system (Fig. 13). The long-term sedimentation rates in these deeper
water sites fall below the late-Holocene rate of SLR rise (1.0 m ka−1).
Vibracores from the same transects, but in shallower water depths of
27–36 m water depth (LB301, GL503, and GL506), demonstrate sedimentation rates (1.1–1.9 m ka−1). These sedimentation rates slightly

5.4. Oﬀshore, alongshore and onshore transfers of littoral sand in the CRLC
system
Columbia River-derived littoral sand likely faced many reversals of
alongshore/shelf and across-shore/shelf sand transport in the CRLC
system due to 1) variable wave energies and directions at seasonal to
inter-decadal (ENSO) time scales (Ruggiero et al., 2005; Ruggiero et al.,
2016) and 2) longer-term sea level change reversals at neotectonic cycle
time scales (decades to multi-century) (Fig. 2 Part B) (Meyer et al.,
1995; Peterson et al., 2010b; Linde, 2014). Nevertheless, a long-term
17

Marine Geology 427 (2020) 106272

C.D. Peterson, et al.

Willapa Bay and Grays Harbor (Fig. 12). The shallowing inner-shelf
gradient in the three southern subcells forced substantial (non-equilibrium) submarine sand accumulation in the uppermost shoreface
(Fig. 13) during latest-Holocene time. However, repeated catastrophic
shoreline retreat following the last several coseismic subsidence events
(Table 2) demonstrated eﬀective beach sand transports oﬀshore to the
inner-shelf following the abrupt sea level rises (1–2 m). Whereas the
alongshore transport and redistributions of beach sand within, and
between, the four subcells were measured at the multi-century time
scales (Tables 5 and ShelfSedRates.pdf, as provided in Supplementary
Data), the across-shore/shelf transport of beach sand following coseismic subsidence occurred at multi-decade time scales (Fig. 4). The
recorded transfers of beach sand from onshore to oﬀshore, following
the rapid sea level rises in latest-Holocene time (Table 2), could have
occurred at about 10× the rates of measured alongshore sand redistribution in the CRLC system. The coeval short-term (decadal) sand
accumulations in the inner-shelf, following coseismic subsidence, were
not resolved in the storm reworked deposits, as 14C dated by Kaminsky
(2006).

Table 9
Littoral sand accumulation rates in the CRLC system.
Inner-shelf
(subcell area)

Accumlation rate
measured (×106
m3 ka−1)

Modiﬁed
accumulation (sink)
from 1.0 m SLR
(×106 m3 m−1 SLR)

Total shelf and bay
accumulation (sink)
from 1.0 m SLR
(×106 m3 m−1 SLR)

Clatsop
Long Beach
Grayland
North Beaches
Subtotal

341
1450
475
246
2512

155
397
127
–
679

155
510
240

Marine-estuaries
Willapa Bay
110
Grays Harbor
117
Subtotal
227

110
117
227

Beach/dune
Clatsop
Long Beach
Grayland
North Beach
Subtotal
Grand Total

–
–
–
–
–
1002

27.4
19.8
15.9
11.7
74.8
2814

5.5. Littoral sand accumulations (sinks) in the inner-shelf and the large
marine-dominated estuaries

Notes: Measured littoral sand accumulation rates that occurred in CRLC subcells
during latest-Holocene time are from Tables 7 and 8. Measured littoral sand
accumulation rates, in Willapa Bay and Grays Harbor marine-dominated estuaries, from the last 3 ka, are from Peterson and Vanderburgh (2018a, 2018b).
Predicted (near-future) subcell oﬀshore accumulations of littoral sand from SLR
are based on conversions of measured littoral sand sedimentation rates of
1.0 m ka−1 to 1.0 m m−1 SLR, based on 1.0 m SLR ka−1 (Table 2) in latestHolocene time. A modiﬁed innermost-shelf area of accumulation is taken to be
between the nearshore ~10 m water depth, or 2.0 km distrance oﬀshore, and
the most conservartive 30 m depth of closure. The total shelf and bay accumulation (sink) includes the modiﬁed innermost-shelf area for all three
southern subcells, and an addtional 113 × 106 m3 m−1 SLR of littoral sand
accumlation each in the Long Beach and Grayland subcells, due to littoral sand
sinks in the adjacent estuaries, Willapa Bay and Grays Harbor.

The rates of littoral sand accretion or net storage in the CRLC oﬀshore (inner-shelf) and inshore (large marine-dominated estuaries) are
tied to Columbia River sediment supply and to increases in available
accommodation spaces. The increases in accommodation space are
proportional to the rate of SLR (1.0 m ka−1) in the study area during
latest-Holocene time (3–0 ka). However, excessive sedimentation rates
(> 1 m ka−1) in the uppermost-submarine shoreface, generally < 20 m
water depth (Fig. 13), resulted from upper-shoreface progradation
(lateral accretion) in late-Holocene time. As this article is focused on
predicted shoreline erosion impacts from future SLR, the lateral
shoreface progradation is neglected, but the measured vertical accretion rates (~1.0 m ka−1) in the deeper shoreface settings (20–30 m
water depth) are extended across the innermost-shelf to the nearshore
(~10 m water depth). As will be shown below in Section 5.6, the
nearshore (0–10 m water depth), extending about 2.0 km distance
seaward from the modern shoreline, could experience erosion during
ravinement, following substantial SLR. For this article the innermostshelf between the nearshore and the 30 m depth of closure is termed the
modiﬁed innermost-shelf area of littoral sand accumulation. Due to the
narrow beaches and very-low inner-shelf sedimentation rates in the
North Beaches subcell area during latest-Holocene time (Fig. 12), that
subcell is not included in the computations of predicted innermost-shelf
sand accumulation rates following potential future SLR. The modiﬁed
inner-shelf sand sinks, as estimated for 1.0 m of SLR, over the innershelf from the nearshore to 30 m water depth in other three subcells are
as follows; 155 × 106 m3 m−1 SLR (Clatsop Plains), 397 × 106 m3 m−1
SLR (Long Beach), 127 × 106 m3 m−1 SLR (Grayland Plains), as shown
in Table 9.
Could new river sand, as potentially supplied to the littoral system
from the large Columbia River, oﬀset some littoral sand loss to the large
marine sand sinks in the inner-shelf and marine dominated estuaries?
Littoral sand throughput from the Columbia River is estimated to have
been at least 2814 × 106 m3 ka−1 during late-Holocene time, as based
on the summing of littoral sand sinks in the inner-shelf and marinedominated estuaries, Willapa Bay and Grays Harbor (Table 9). This
value is smaller than that of 3720 × 106 m3 ka−1, as based on longer
early- to late-Holocene time intervals measured for the inner-shelf and
Grays Harbor (Table 1). The current throughput of Columbia River sand
is not known relative to prehistoric conditions, due the lack of sand
retention measurements in numerous impoundments (> 100 in
number) in the Columbia River tributaries. However, a modeled historic throughput value of 2.7 × 106 m3 yr−1 has been proposed by
Kaminsky et al. (2001). The estimated supplies of Columbia River sand

net northward transport in the inner-shelf delivered Holocene Columbia
River littoral sand to the northern subcells, where shoreward transport
then fed the large marine-dominated estuaries (Willapa Bay and Grays
Harbor) and the prograding barriers and beach plains (Fig. 12). The
shoreward across-shelf transport was eﬀective in supplying littoral sand
to the most distal beaches landward of transects SL5, SL7, SL11, and
SL12 in the North Beaches subcell (Fig. 13), due the very shallow
gradients of the shelf wave-cut platform (Twichell and Cross, 2001;
Twichell et al., 2010). However, a southward onshore/nearshore
transport stripped sand from the northernmost beaches, between the
Point Grenville and the CORP/SL7 transects, limiting onshore accommodation space there. A southward onshore/nearshore transport also
delivered Columbia River-derived littoral sand to the Clatsop subcell,
located south of the Columbia River mouth, where net oﬀshore sand
transport permitted progradation of the uppermost-submarine shoreface, < 20 m water depth in SL23 (Fig. 13) (Peterson et al., 2010a;
Peterson et al., 2016). The net northward transport in the inner-shelf
stripped sand from deeper water depths of the Clatsop subcell area
(Fig. 11), thereby resulting in the low-sedimentation rates
(~0.1 m ka−1) in the deeper submarine shoreface (> 25 m water
depth). In the subcells located north of the Columbia River mouth, the
northward inner-shelf transport delivered an abundance of Columbia
River-derived littoral sand that 1) ﬁlled the available inner-shelf accommodation space in middle-Holocene time, and then 2) initiated
submarine upper-shoreface progradation in late-Holocene time
(Peterson et al., 2016). Subaerial beach and dune facies deposition
trailed behind the prograding submarine shoreface deposits. However,
local delays in the onset of barrier spit and beach plain progradation
until latest-Holocene time (3–0 ka) apparently occurred due to competition for littoral sand with the large marine-dominated estuaries,

18

Marine Geology 427 (2020) 106272

C.D. Peterson, et al.

onshore (subaerial) eroded sand volumes for diﬀerent timeline intervals
(Table 5) are added to submarine eroded sand volumes (< 0 m
NAVD88) to yield a total sand volume loss for the analyzed subcells.
The onshore submarine eroded sand volume is calculated from translating the modern nearshore gradient (1.0%) landward. An equal volume is assumed for the nearshore eroded sand volume. Both distances
of submarine eroded sand volumes are scaled to the distance of onshore
retreat. The cross-sectional areas of submarine erosion are multiplied
by the subcell inner-shelf length to yield the subcell submarine erosion
volume. The subaerial erosion volume between timelines, and corresponding submarine erosion volumes, are combined to yield the subcell
total erosion volume. That is the subcell erosion volume needed to ﬁll
the subcell increased accommodation spaces corresponding to potential
future SLR. Because no post-submergence accumulation is assumed for
the eroded nearshore, ~2.0 km width estimated for 2 m SLR (as shown
in Fig. 15), that area is excluded from the inner-shelf accommodation
space. The relations above are compared to ﬁnd the closest match between the subcell increased submarine accommodation space and the
total subcell erosion volume, corresponding to the diﬀerent onshore
timeline intervals (Table 10). The closest matches yield subcell averaged retreat distances for potential for SLR of 1.0 m, 2.0 m and 3.0 m in
the CRLC system (Fig. 16). Predicted averaged shoreline retreat distances for 2.0 m of SLR are as follows; ~0.7 km (Clatsop Plains subcell),
~1.2 km (Long Beach subcell), and ~1.3 km (Grayland Plains subcell).
The averaged retreat distances (Table 10) are reported to 0.1 km
resolution, but they could be under-estimated by as much as 30%.
These retreat estimates are assumed to be minimum values for assumed
SLR for the following reasons 1) the subtidal area of littoral sand accumulation in Willapa Bay (40% of Willapa Bay surface area or about
25% of the total Willapa Bay and Grays Harbor surface area) is not
included in the estuarine littoral sand sinks (Table 9) and 2) the outermost inner-shelf areas (30–50 m water depth) are not included in the
inner-shelf sand sinks, as they might not accumulate littoral sand at the
rate of near-future SLR. The outermost inner-shelf deposits account for
less than 20% of the total inner-shelf littoral sand accumulations during
late-Holocene time (ShelfSedRates.pdf in Supplementary Data). Submarine littoral sink errors (inner-shelf and estuaries), as based on surface area uncertainties, could reach 30% (mostly under-estimated).
However, the largest uncertainties in shoreline retreat distances arise
from the amounts of SLR, posed here, as scenarios of 1.0 m, 2.0 m, and
3.0 m. The corresponding accumulation thicknesses of littoral sand
yield uncertainties of vertical accretion, reaching up to 300%. Given
assumed SLR values, and corresponding littoral sand thicknesses for the
diﬀerent scenarios, the estimated averaged retreat distances could vary
by 30% for each scenario. Site speciﬁc uncertainties could be greater
due to local variations from the mean conditions.
Averaged shoreline retreat values, as predicted by the volume accommodation space relations presented in this article (Table 10), are
compared to shoreline retreat distances that were predicted by the
modiﬁed Bruun method (Bruun, 1962; Komar et al., 1991), as utilized
for the CRLC beaches by Doyle (1996). The simpliﬁed Bruun method
parameters used by Doyle (1996) are; 1) across-shore beach retreat (R),
2) measured modern mid-beach berm heights (B = 3–4 m), 3) potential
SLR rise (S = 2.0 m), 4) oﬀshore interdecadal depths of closure
(h = 20 m), and 5) oﬀshore distances (L) to the point of depth of
closure, where R = S{L/(B + h)}. Representative proﬁles analyzed by
Doyle (1996) totaled n = 6 in the Clatsop subcell, n = 6 in the Long
Beach subcell, and n = 3 in the Grayland subcell. Averaged retreat
distances from 2.0 m of SLR are 300 m, 400 m and 400 m, respectively,
for the Clatsop Plains, Long Beach, and Grayland Plains subcells. Those
retreat distances (Doyle, 1996) are only ~30–50% as large as those
estimated by the measured accommodation relations as used in this
article (Table 10). The diﬀerences between the two approaches are
attributed to 1) the greater across-shelf distances of net sand accumulation assumed for the 1–2 century time scales and 2) the large marinedominated estuarine sand sinks (Willapa Bay and Grays Harbor) that

to the CRLC inner-shelf, estuary, and barrier/beach plain sinks are relatively similar for Holocene time (~3.7 × 106 m3 yr−1) and latestHolocene time (~2.8 × 106 m3 ka−1). Some sand loss occurred from
lithic fragment abrasion in the surf zone (~30%) and sand transport
north out of CRLC system (Peterson et al., 2016). Given those losses the
averaged sand sink accumulation rates ~3.3 × 106 m3 ka−1, are not
substantially diﬀerent from the previously reported Columbia River bed
load throughput rates of about 5 × 106 m3 ka−1, as based on timevarying bed load ﬁll rates in the Columbia River estuary (Peterson
et al., 2013). The estimated sand budgets in the CRLC system are relatively well-balanced throughout middle- to late-Holocene time.
For the purposes of predicting potential shoreline response to SLR is
this article, we assume a hypothetical Columbia River sand throughput
value of 3 × 106 m3 yr−1, which translates into 300 × 106 m3 per
100 yr or 600 × 106 m3 per 200 yr. In either case, most of the hypothetical new river sand supply (300–600 million cubic meters over
one to two centuries) would not reach the estuary mouth due to increasing accommodation in the estuary following an accelerated SLR.
For example, during late-Holocene time the bedload (sand) accommodation space in the Columbia River was 570 × 106 m3 ka−1
(Peterson et al., 2013) or 570 × 106 m−1 SLR, assuming a late-Holocene SLR of 1.0 m ka−1 (Fig. 2 Part A). The projected increase in accommodation space, due to potential SLR of 1–3 m, is similar in volume
to the hypothetical Columbia River sand production over a 1–2 century
time interval. Therefore, sand in-ﬁlling of increasing accommodation
space in the Columbia River estuary would largely cancel-out the potential river sand throughput to the littoral zone from a 1–3 m SLR
during the next century or two. Such inﬁlling of bedload accommodation space by new Columbia River sand would also largely preclude reentry of littoral (beach) sand back into the lower Columbia River estuary reaches, following a near-future SLR of ≤3 m.
5.6. Predicted shoreline retreat following potential future sea level rise
In this article, subcell averaged shoreline retreat distances are based
on onshore sand volume loss to increasing marine accommodation
space, following potential near-future SLR values of 1–3 m. Oﬀshore
(seaward) sand transport in the CRLC system by wave oscillatory currents, geostrophic ﬂow and down-welling, has been modeled by Kachel
and Smith (1986). Such oﬀshore transport and dispersal of Columbia
River sand, as modeled by Sternberg (1986) and Kachel and Smith
(1986), are directly conﬁrmed by shelf littoral sand sedimentation
rates, as summarized here in Table 8. The accommodation space mechanisms of littoral sand transport into and retention in the large
marine-dominated estuaries are discussed elsewhere for Willapa Bay
and Grays Harbor (Peterson and Vanderburgh, 2018a, 2018b). The
major marine sinks of littoral sand occur in the innermost-shelf
(total = 679 × 106 m3 m−1 SLR) (Table 9). However, the large marinedominated estuaries, Willapa Bay and Grays Harbor, are also signiﬁcant
sinks (total = 227 × 106 m3 m−1 SLR). For this article, the littoral sand
sinks in Willapa Bay and Grays Harbor estuaries are split between the
Long Beach and Grayland Plains subcells to yield combined marine
sinks of littoral sand for the Long Beach and Grayland Plains subcells.
No littoral sand is assumed to be lost or gained from the Clatsop subcell
to the Columbia River estuary during the near future period (1–2 centuries) of potential SLR (1–3 m), as previously discussed in Section 5.5.
For the purposes of this article, the potential sand transfers from onshore to oﬀshore are taken to occur within corresponding subcell areas
of the Clatsop, Long Beach and Grayland subcells. Sand volume loss or
gain between the four CRLC subcells could be addressed in future studies by combining the marine sinks and averaging the shoreline sand
loss proportionally across the four subcells on bases of identiﬁed
longshore transport parameters.
The predicted subcell beach sand losses to increasing oﬀshore/inshore marine accommodation spaces from SLR are used here to estimate
corresponding shoreline retreat distances (Fig. 15). Speciﬁcally, the
19

Marine Geology 427 (2020) 106272

C.D. Peterson, et al.

Fig. 15. Diagram of onshore and nearshore sand volume losses for a
hypothetical retreat interval (A-D), including subaerial eroded volumes
(lightly shaded) and submarine eroded volumes (darkly shaded) from a
2 m SLR in the Long Beach subcell. Some accumulation is shown for
the seaward portion of the submarine nearshore, but the transition
zone to 2.0 km seaward of the present shoreline is not included in
marine sink sand accumulation estimates. This transition zone (1/3 of
the innermost-shelf proﬁle distance) is assumed from equilibrium
proﬁle shifts (Bruun, 1962). The depth of truncation is based on a 1.0%
ravinement gradient and distance from the maximum retreat or timeline ‘D'. The onshore eroded basal slope (1.0% gradient) is translated
(shifted) landward from the modern nearshore gradient. The modern
nearshore gradient (1.0%) is taken from the modern beach toe (0 m
NAVD88) to 1.0 km distance oﬀshore in transects from the Clatsop,
Long Beach, and Grayland subcells. The 0 m NAVD88 elevation datum
is ~1.0 m below mean sea level (MSL) in the study area. Diagram
vertical exaggeration is 20×.

competing for available littoral sand supply. Some accommodation
spaces ﬁll before others, leading to predictable sequences of coastal
evolution. Under conditions of littoral sediment deﬁcit or when new
submarine accommodation space is added, such as during sea level rise
(SLR), it is expected that the onshore accommodation spaces, last to ﬁll,
will be the ﬁrst to empty, leading to predicted sequences of beach
erosion. In this article, the depositional evolutions of the prograded
barriers and beach plains in the CRLC system are explained in terms of
sequential ﬁlling of available accommodation spaces in diﬀerent depositional environments and in diﬀerent shoreline segments during
late-Holocene time. Filling of the inner-shelf and the large marinedominated estuaries, Grays Harbor and Willapa Bay, preceded uppershoreface progradation and accretion of subaerial beach and foredune
deposits in the Long Beach, Clatsop Plains, Grayland Plains, and North
Beaches subcells. Overlap of accommodation space ﬁlling in the innershelf and estuaries has continued into late-Holocene time. However,
measured accumulation rates decrease in magnitude from the innershelf to the large marine-dominated estuaries to the onshore beach
segments in the four CRLC subcell areas.
The order of accommodation space ﬁlling is hypothetically ‘reversed' to predict shoreline retreat in the four CRLC subcells following
potential SLR from ongoing global warming. The subcell beaches and
barriers will be eroded to ﬁll increasing accommodation space in the
inner-shelf and in the large marine-dominated estuaries. The high energies of waves, tides, and onshore-winds in the CRLC system permitted
the barrier and beach plain shorelines and large marine-dominated
estuaries to respond to episodic coseismic subsidence and interseismic
uplift, over time scales of several decades. In such high-energy littoral
systems, the shoreline erosion response to potential future accelerated
SLR is expected to ‘keep pace’ with the predicted rates of potential
eustatic sea level rise (1–3 m) over the next 100–200 years. Beach sand
losses to increasing oﬀshore (inner-shelf) and inshore (estuary) submarine accommodation spaces in the CRLC system, resulting from near
future SLR of 2.0 m, are estimated to result in shoreline retreat distances of 0.7–1.3 km, representing 25–50% of the widths of prograded
barrier spits and beach plains. The severe beach retreat estimates reported here are due to the wide-scale transfers of beach sand to increasing submarine accommodation spaces in the inner-shelf and in
large marine-dominated estuaries, following potential future SLR. Such
severe beach retreat estimates, as predicted for the CRLC system, should
serve as warnings about future catastrophic beach erosion resulting
from near-future SLR in other high-wave-energy sandy shorelines
worldwide.

Table 10
Shoreline volume loss and retreat distance.
Total
marine
sink
volume
(×106
m3 )

Onshore
timeline
interval

Eroded
subaerial
volume
(×106
m3 )

Eroded
submarine
volume
(×106 m3)

Eroded
combined
volume
(×106
m3)

Averaged
retreat
distance
(km)

155
310
465

Mod/A
Mod/C
Mod/D

110
160
240

30
120
390

140
280
640

0.3
0.7
1.2

Long Beach
1.0
510
2.0
1020
3.0
1530

Mod/B
Mod/D
Mod/E

170
310
390

310
800
1230

470
1110
1620

0.7
1.1
1.5

Grayland
1.0
240
2.0
480
3.0
720

Mod/A
Mod/B
Mod/D

80
100
160

120
360
520

190
460
670

0.7
1.3
1.5

Subcell/
SLR (m)

Clatsop
1.0
2.0
3.0

Notes: The total retreat volumes (subaerial and submarine) (x106 m3) are established for each subcell onshore timeline interval. The submarine sand sinks
include both the inner-self (all three subcells) and the large marine-dominated
estuaries, Willapa Bay (Long Beach subcell) and Grays Harbor (split between
the Long Beach and Grayland Plains subcells). The closest combined eroded
volume for each subcell is then found for the predicted total marine sink for a
given SLR value (1, 2, or 3 m). The matched timeline intervals yield the subcell
averaged retreat distances (m). Averaged retreat distances are reported to
0.1 km (resolution) but they have estimated uncertainties of as much as 30%.

are included in the measured accommodation space relations.
Ranasinghe et al. (2012) also reported much larger beach retreat values
in some settings, relative to the simple Bruun Method, due to sediment
sinks in tidal inlets. The measured accommodation space approach
proposed in this article, oﬀers advantages in that shoreline susceptibility to future SLR can be evaluated relative to measured values of
sand supply and sand loss, as calibrated to past conditions of sea level
change. The measured accommodation space relations demonstrated in
this article were found to be particularly applicable to the complex and
interactive settings in the large CRLC system. Such approaches are recommended for other similarly-complex littoral systems worldwide.
6. Conclusion
In complex littoral systems, such as the Columbia River Littoral Cell
(CRLC), accommodation spaces in diﬀerent depositional environments,
such as the inner-shelf, estuaries, and beaches can be thought of as
20

Marine Geology 427 (2020) 106272

C.D. Peterson, et al.

Fig. 16. Predicted shoreline erosion from beach sand transfer to oﬀshore (inner-shelf) and inshore (marine dominated estuaries) following predicted near-future SLR
at 1 m, 2 m, and 3 m. Data are from Table 10. Landward centroid positions used to map retreat interval distances are shown (solid circles). Retreat lines (dashed)
between identiﬁed centroids are guided by abandoned foredunes, as used to constrain polygon cell boundaries (Figs. 7 and 8). Predicted local retreat line positions
(solid circles) are shown to 0.4 km resolution or 30% of a representative retreat distance of 1.3 km for 2–3 m SLR. UTM northing coordinates are in kilometers. See
Fig. 1 for subcell locations in the CRLC system.

21

Marine Geology 427 (2020) 106272

C.D. Peterson, et al.

Declaration of Competing Interest

University of Sydney, New South Wales (633 p).
Kaminsky, G.M., Buijsman, M.C., Ruggiero, P., 2001. Predicting shoreline change at
decadal scale in the Paciﬁc Northwest, USA. In: Proceedings of the 27th International
Conference on Coastal Engineering. American Society of Civil Engineers, pp.
2400–2413.
Kaminsky, G.M., Ruggiero, Buijsman, M.C., McCandless, D., Gelfenbaum, G., 2010.
Historical evolution of the Columbia River littoral cell. Mar. Geol. 273, 96–126.
Komar, P.D., Lanfredi, N., Baba, M., Dean, R.G., Dyer, K., Healy, T., Ibe, A.C., Terwindt,
J.H.J., Thom, B.G., 1991. The response of beaches to sea-level changes-a review of
predictive models. J. Coast. Res. 7, 895–921.
Kopp, R.E., Horton, R.M., Little, C.M., Mitrovica, J.X., Oppenheimer, M., Rasmussen, D.J.,
Strauss, B.H., Tebaldi, C., 2014. Probabilistic 21st and 22nd century sea-level projections at a global network of tide-gauge sites. Earth’s Future 2, 383–406.
Kopp, R.E., Gilmore, E.A., Little, C.M., Lorenzo-Trueba, J., Ramenzoni, V.C., Sweet, W.V.,
2019. AGU Centennial Grand Challenge: Sea-level science on the frontier of usability.
Earth’s Future. AGU Adv. Earth Space Sci. 7, 1235–1269.
Linde, T., 2014. Regional Database Analysis of Dated Prehistoric Shorelines to Establish
Sand Partitioning in Late Holocene Barriers and Beach Plains of the Columbia River
Littoral Cell, Washington and Oregon, USA. M.S. Thesis. Portland State University,
Oregon (86 p).
Lorenzo-Trueba, J., Ashton, A.D., 2014. Rollover, drowning, and discontinuous retreat:
Distinct modes of barrier response to sea-level rise arising from a simple morphodynamic model. J. Geophys. Res. Earth Surf. Process. 119, 779–801.
Mengel, M., Levermann, A., Frieler, K., Robinson, A., Marzeion, B., Winkelmann, R.,
2016. Future Sea level rise constrained by observations and long-term commitment.
Proc. Natl. Acad. Sci. 113, 2597–2602.
Meyers, R.A., Smith, D.G., Jol, H.M., Peterson, C.D., 1996. Evidence for eight great
earthquake-subsidence events detected with ground-penetrating radar, Willapa barrier, Washington. Geology 24, 99–102.
Ortiz, A.C., Ashton, A.D., 2016. Exploring shoreface dynamics and a mechanistic explanation for a morphodynamic depth of closure. J. Geophys. Res. Earth Surf.
Process. 121, 442–464.
Peterson, C.D., Cruikshank, K.M., 2014. Quaternary tectonic deformation, Holocene paleoseismicity, and modern strain in the unusually-wide coupled zone of the central
Cascadia margin, Washington and Oregon, USA and British Columbia, Canada. J.
Geogr. Geol. 6 (33 p).
Peterson, C.D., Phipps, J.B., 2016. Accommodation space controls on incised-valley sediment accumulation rates during the Holocene marine transgression (0–11 ka) in
Grays Harbor, a large meso-tidal estuary, Washington, USA. Mar. Geol. 380, 1–16.
Peterson, C.D., Vanderburgh, S., 2018a. Tidal ﬂat depositional response to neotectonic
cyclic uplift and subsidence (1–2 m) as superimposed on latest-Holocene net sea level
rise (1.0 m/ka) in a large shallow mesotidal wave-dominated estuary, Willapa Bay,
Washington, USA. J. Geogr. Geol. 10, 109–139.
Peterson, C.D., Vanderburgh, S., 2018b. Interconnected accommodation space controls
between sand-charged shallow tidal channels and wind-wave truncated tidal ﬂats
during latest-Holocene Sea level rise (~3.0 m) in a large mesotidal wave-dominated
estuary, Grays Harbor, Washington, USA. J. Geogr. Geol. 10, 26–56.
Peterson, C.D., Doyle, D.L., Barnett, E.T., 2000. Coastal ﬂooding and beach retreat from
coseismic subsidence in the Central Cascadia margin, USA. Environ. Eng. Geol. 6,
255–269.
Peterson, C.D., Vanderburgh, S., Roberts, M.C., Jol, H.M., Phipps, J.P., Twichell, D.C.,
2010a. Composition, age, and depositional rates of Holocene shoreface deposits
under barriers and beach plains of the Columbia River littoral cell, USA. Mar. Geol.
273, 62–82.
Peterson, C.D., Jol, H.M., Vanderburgh, S., Phipps, J.B., Percy, D., Gelfenbaum, G.,
2010b. Dating of late-Holocene shoreline positions by regional correlation of coseismic retreat events in the Columbia River littoral cell. Mar. Geol. 273, 44–61.
Peterson, C.D., Jol, H.M., Horning, T., Cruikshank, K.M., 2010c. Paleotsunami Inundation
of a Beach Ridge Plain: Cobble Ridge Overtopping and Inter-ridge Valley Flooding in
Seaside, Oregon, USA. J. Geol. Res (22 p. Article ID 276989).
Peterson, C.D., Gates, E.B., Minor, R., Baker, D.L., 2013. Accommodation space controls
on the latest Pleistocene and Holocene (16-0 ka) sediment size and bypassing in the
lower Columbia River Valley: a large ﬂuvial-tidal system in Oregon and Washington,
USA. J. Coast. Res. 29, 1191–1211.
Peterson, C.D., Vanderburgh, S., Roberts, M., 2014. Late Holocene Geomorphology of the
Columbia River Estuary, Oregon and Washington, USA. J. Geogr. Geol. 6, 1–27.
Peterson, C.D., Twichell, D.C., Roberts, M.C., Vanderburgh, S., Hostetler, S.W., 2016.
Accommodation space in a high-wave-energy inner-shelf during the Holocene marine
transgression: correlation of onshore and oﬀshore inner-shelf deposits (0-12 ka) in
the Columbia River littoral cell system, Washington and Oregon, USA. Mar. Geol.
379, 140–156.
Phipps, J., Jol, H.M., Peterson, C.D., Vanderburgh, S., 2001. Sand dune reactivation and
subduction zone earthquakes in the Grayland area. Wash. Geol. 28, 31–33.
Ranasinghe, R.T., Duong, M., Uhlenbrook, S., Roelvink, D., Stive, M.J.F., 2012. Climatechange impact assessment for inlet-interrupted coastlines. Nat. Clim. Chang. 3,
83–87.
Rankin, D.K., 1983. Holocene Geologic History of the Clatsop Plains Foredune Ridge
Complex. M.S. Thesis. Portland State University, Portland, Oregon (175 p).
Reckendorf, F., Peterson, C., Percy, D., 2001. The Dune Ridges of Clatsop County, Oregon.
State of Oregon Department of Geology and Mineral Resources, Open-File Report O01-07, CDRom.
Ruggiero, P., Kaminsky, G.W., Komar, P.D., McDougal, W.G., 1997. Extreme waves and
coastal erosion in the Paciﬁc Northwest. In: Proceedings of Waves ‘97 ASCE, pp.
947–961.
Ruggiero, P., Kaminsky, G.M., Gelfenbaum, G., Voigt, B., 2005. Seasonal to interannual
morphodynamics along a high-energy dissipative littoral cell. J. Coast. Res. 553–578.

The authors declare that they have no known competing ﬁnancial
interests or personal relationships that could have appeared to inﬂuence the work reported in this paper.
Acknowledgements
Prehistoric sediment budget analyses and accommodation space
applications in the Columbia River Littoral Cell (CRLC) system were
initially encouraged by James Phipps, Kenneth Scheidegger, Larry
Sloss, Chris Kraft, Jerry Glenn, and Derald Smith. Graduate student
thesis studies to help support these analyses have been performed by
Edward Gates, Lorraine Woxell, April Herb, Diana Baker, and George
Kaminsky. Kenneth Cruikshank assisted with database design for the
CRLC prograded barrier spits and beach plains in this article. Marie
Ferland assisted with vibracoring, core logging, and radiocarbon sampling, in the CRLC inner-shelf. Kennett Peterson assisted with editing
early versions of this article. This work was funded, in part, by the U.S.
Geological Survey, Coastal and Marine Geology Program, under the
South West Washington Coastal Erosion Project, Coop #1434-HQ-96AG-01612 from 1996 to 2001. Additional funding (2001 to present) has
been provided by Fraser Valley College, Lethbridge College, Medicine
Hat College, and Kwantlen Polytechnic University.
References
Alder, J.R., Hostetler, S.W., 2015. Global climate simulations at 3000-year intervals for
the last 21,000 years with the GENMOM coupled atmosphere-ocean model. Clim.
Past 11, 449–471.
Atwater, B.F., Hemphill-Haley, E., 1997. Recurrence intervals for great earthquakes of the
past 3,500 years at northeastern Willapa Bay, Washington. In: US Geological Survey
Professional Paper 1576, (1–108 p).
Atwater, B.F., Tuttle, M.P., Schweig, E.S., Rubin, C.M., Yamaguchi, D.K., Hemphill-Haley,
E., 2004. Earthquake recurrence, inferred from paleoseismology. In: Gillespie, A.R.,
Porter, S.C., Atwater, B.F. (Eds.), The Quaternary Period in the United States.
Elsevier, Amsterdam, pp. 331–350.
Baker, D., Peterson, C., Hemphill-Haley, E., Twichel, D., 2010. Holocene sedimentation in
the Columbia River Estuary. Mar. Geol. 273, 83–95.
Bamber, J.L., Oppenheimer, M., Kopp, R.E., Aspinall, W.P., Cooke, R.M., 2019. Ice sheet
contributions to future sea-level rise from structured expert judgment. Proc. Natl.
Acad. Sci. 116, 11195–11200.
Bruun, P., 1962. Sea-level rise as a cause of shore erosion. J. Waterways Harbors Div. 88,
117–132.
Calib7.10, 2019. Radiocarbon Calibration Program. CALIB REV7.1.0. http://calib.org/
calib/calib.html (Accessed May 25, 2019).
Cruikshank, K.M., Peterson, C.D., 2017. Late-stage interseismic strain interval, Cascadia
subduction zone margin, USA and Canada. Open J. Earthq. Res. 6, 1–34.
Darienzo, M.E., Peterson, C.D., Clough, C., 1994. Stratigraphic evidence for great subduction zone earthquakes at four estuaries in Northern Oregon. J. Coast. Res. 10,
850–876.
DeConto, R.M., Pollard, D., 2016. Contribution of Antarctica to past and future sea-level
rise. Nature 531, 591–597.
Doyle, D.L., 1996. Beach Response to Subsidence Following a Cascadia Subduction Zone
Earthquake along the Washington-Oregon Coast. M.S. Thesis, Portland State
University, Portland, Oregon (113 p).
Gesch, D., Oimoen, M., Greenlee, S., Nelson, C., Steuck, M., Tyler, D., 2002. The National
Elevation Dataset: Photogrammetric Engineering and Remote Sensing. 68. pp. 5–32.
Herb, A.A., 2000. Holocene Stratigraphy and Sediment Volumes for the Columbia River
Littoral Cell, Paciﬁc Northwest, USA. M.S. Thesis, Portland State University, Portland,
Oregon (186 p).
Jevrejeva, S., Moore, J.C., Grinsted, A., 2012. Sea level projections to AD2500 with a new
generation of climate change scenarios. Glob. Planet. Chang. 80, 14–20.
Jol, H.M., Peterson, C.D., Vanderburgh, S., Phipps, J., 1999. GPR as a regional geomorphic tool: shoreline accretion/erosion along the Columbia River littoral cell. In:
Proceedings of the Seventh International Conference on Ground Penetrating Radar
(GPR’98). vol. 1. University of Kansas, Lawrence, Kansas, pp. 257–262.
Jol, H.M., Lawton, D.C., Smith, D.G., 2003. Ground penetrating radar: 2-D and 3-D
subsurface imaging of a coastal barrier spit, Long Beach, WA, USA. Geomorphology
53, 165–181.
Jurney, C., 2001. Reﬁning the Measured Frequency of Great Earthquakes along the
Cascadia Subduction Zone. M.S. thesis, Central Washington University, Ellensberg,
Washington (78 p).
Kachel, N.B., Smith, J.D., 1986. Geologic impact of sediment transporting events on the
Washington continental shelf. In: Knight, R.J., McLean, J.R. (Eds.), Shelf Sands and
Sandstones. Canadian Society of Petroleum Geologists, Memoir II, pp. 145–162.
Kaminsky, G.M., 2006. Shoreface Behavior and Equilibrium. Unpublished PhD Thesis.

22

Marine Geology 427 (2020) 106272

C.D. Peterson, et al.

Washington continental shelf, USA. J. Geol. Soc. Lond. 143, 945–956.
Twichell, D.C., Cross, V.A., 2001. Holocene evolution of the southern Washington and
northern Oregon shelf and coast: Geologic discussion and GIS data release. In: U.S.
Geological Survey, Open File Report 01–076, (27 p., and CDRom disk).
Twichell, D.C., Cross, V., Peterson, C.D., 2010. Partitioning of sediment on the shelf
oﬀshore of the Columbia River littoral cell. Mar. Geol. 273, 11–31.
Vanderburgh, S., Roberts, M.C., Peterson, C.D., Phipps, J.B., Herb, A., 2010. Holocene
transgressive and regressive deposits of the Columbia River littoral cell barriers and
beach plains. Mar. Geol. 273, 32–43.
Woxell, L.K., 1998. Prehistoric Beach Accretion Rates and Long-Term Response to
Sediment Depletion in the Columbia River Littoral Cell. M.S. Thesis, Portland State
University, Portland, Oregon (206 p).

Ruggiero, P., Kaminsky, G.M., Gelfenbaum, G., Cohn, N., 2016. Morphodynamics of
prograding beaches: a synthesis of seasonal-to century-scale observations of the
Columbia River littoral cell. Mar. Geol. 376, 51–68.
Satake, K., Shimazaki, K., Tsuji, Y., Ueda, K., 1996. Time and size of giant earthquake in
Cascadia inferred from Japanese tsunami records of January 1700. Nature 378,
246–249.
Schlichting, R.B., Peterson, C.D., 2006. Mapped Overland Distance of paleotsunami highvelocity inundation in back-barrier wetlands of the Central Cascadia margin, USA. J.
Geol. 114, 577–592.
Sherwood, C.R., Jay, D.A., Harvey, B., Hamilton, P., Simenstad, C.A., 1990. Historical
changes in the Columbia River estuary. Prog. Oceanogr. 25, 299–352.
Sternberg, R.W., 1986. Transport and accumulation of river-derived sediment on the

23