of 27
1.
Introduction
River deltas are home to diverse ecosystems, valuable resources, and nearly half a billion of the human popula
-
tion (Gleick,
2003
; Olson & Dinerstein,
1998
; Vörösmarty et al.,
2009
). Many deltas develop a triangular-shaped
planform morphology through natural diversions in channel course to the shoreline, known as river avulsions,
which occur periodically every
∼10−1000
years depending on the delta (Figures
1a
and
1b
) (Slingerland &
Smith,
2004
). Avulsions are a hazard to human life and property (Sinha,
2009
; Syvitski & Brakenridge,
2013
) and
have been responsible for some of the deadliest flood disasters in human history (Kidder & Liu,
2017
; Soong &
Zhao,
1994
). At the same time, avulsions are necessary to counter land lost by nourishing subsiding wetlands with
sediment, nutrients, and carbon (Figure
1b
) (Edmonds et al.,
2009
; Richards et al.,
2002
). Anthropogenic inter
-
ference and greenhouse effects are contributing to relative sea-level rise across the globe (Pachauri et al.,
2014
),
and engineered avulsions are important elements of billion-dollar coastal restoration plans to combat rise and
sustain coastal cities and ecosystems (Brakenridge et al.,
2017
; Coastal Protection and Restoration Authority of
Louisiana,
2007
). To increase the success of such plans, we need to understand where and when river avulsions
naturally occur on deltas.
Abstract
Lowland deltas experience natural diversions in river course known as avulsions. River avulsions
pose catastrophic flood hazards and redistribute sediment that is vital for sustaining land in the face of sea-level
rise. Avulsions also affect deltaic stratigraphic architecture and the preservation of sea-level cycles in the
sedimentary record. Here, we present results from an experimental lowland delta with persistent backwater
effects and systematic changes in the rates of sea-level rise and fall. River avulsions repeatedly occurred
where and when the river aggraded to a height of nearly half the channel depth, giving rise to a preferential
avulsion node within the backwater zone regardless of sea-level change. As sea-level rise accelerated, the
river responded by avulsing more frequently until reaching a maximum frequency limited by the upstream
sediment supply. Experimental results support recent models, field observations, and experiments, and suggest
anthropogenic sea-level rise will introduce more frequent avulsion hazards farther inland than observed in
recent history. The experiment also demonstrated that avulsions can occur during sea-level fall—even within
the confines of an incised valley—provided the offshore basin is shallow enough to allow the shoreline to
prograde and the river to aggrade. Avulsions create erosional surfaces within stratigraphy that bound beds
reflecting the amount of deposition between avulsions. Avulsion-induced scours overprint erosional surfaces
from sea-level fall, except when the cumulative drop in sea-level is greater than the channel depth and less than
the basin depth. Results imply sea-level signals outside this range are removed or distorted in delta deposits.
Plain Language Summary
Rivers on deltas are unstable and naturally change course every
∼10–1,000 years through a process known as avulsion. Avulsions are responsible for some of the deadliest
natural disasters in human history and for depositing sediment that builds sedimentary records and sustains
coastal landscapes in the face of anthropogenic sea-level rise. Sea level can affect where and when channels
shift course, but understanding these effects is difficult because avulsions occur infrequently. Here, we present
findings from a laboratory experiment that produced avulsions on a delta in miniature. As sea level rises more
quickly, we found that avulsions occur more often and at locations farther upstream. The record of avulsions
is recorded in the pattern of sedimentary layers within the delta, but this pattern may be confused with erosion
during sea-level fall. Findings from this experiment support recent predictions and observations of deltas in
nature.
CHADWICK ET AL.
© 2022. American Geophysical Union.
All Rights Reserved.
Effect of Sea-Level Change on River Avulsions and
Stratigraphy for an Experimental Lowland Delta
A. J. Chadwick
1
, S. Steele
1
, J. Silvestre
1
, and M. P. Lamb
1
1
Division of Geological and Planetary Sciences, California Institute of Technology, Pasadena, CA, USA
Key Points:
A delta experiment shows sea-level
rise causes more frequent avulsions
at a constant distance upstream of the
river mouth
During sea-level fall, avulsions occur
if progradation into the offshore basin
is fast enough to counteract incision
and cause aggradation
In stratigraphy, avulsion-induced
incision overprints sea-level scours
smaller than the channel depth or
greater than the basin depth
Supporting Information:
Supporting Information may be found in
the online version of this article.
Correspondence to:
A. J. Chadwick,
austin.chadwick23@gmail.com
Citation:
Chadwick, A. J., Steele, S., Silvestre, J.,
& Lamb, M. P. (2022). Effect of sea-level
change on river avulsions and stratigraphy
for an experimental lowland delta.
Journal of Geophysical Research: Earth
Surface
,
127
, e2021JF006422.
https://doi.
org/10.1029/2021JF006422
Received 7 SEP 2021
Accepted 24 JUN 2022
Author Contributions:
Conceptualization:
A. J. Chadwick, M.
P. Lamb
Funding acquisition:
A. J. Chadwick,
M. P. Lamb
Investigation:
A. J. Chadwick, S. Steele,
J. Silvestre, M. P. Lamb
Supervision:
A. J. Chadwick, M. P. Lamb
Writing – review & editing:
A. J.
Chadwick, S. Steele, J. Silvestre, M.
P.
Lamb
10.1029/2021JF006422
RESEARCH ARTICLE
1 of 27
Journal of Geophysical Research: Earth Surface
CHADWICK ET AL.
10.1029/2021JF006422
2 of 27
Interpreting ancient delta deposits depends in part on our ability to disentan
-
gle stratigraphic patterns of sea-level change and river avulsions. Fluviodel
-
taic deposits are major building blocks of the sedimentary record on Earth,
and are important reservoirs of hydrocarbon and freshwater resources
(Bohacs & Suter,
1997
; Hariharan et al.,
2021
). On Mars, the Perseverance
rover has begun exploration of ancient deltaic deposits to unravel lake-level
history and assess conditions for ancient life (Farley et al.,
2020
). Repeat
-
ing sequences in sedimentary strata are thought to generally reflect changes
in sea level over time, with deposition during sea-level rise and erosion
during
sea-level fall (i.e., sequence stratigraphy) (Allen & Posamentier,
1993
;
Van Wagoner,
1998
). However, deciphering this record is not always clear:
avulsions and floods can create erosional surfaces, regardless of sea-level
change, which appear similar to scour generated during sea-level fall (Ganti
et al.,
2019
; Trower et al.,
2018
). In addition, deposition can occur during
sea-level fall if sediment supply and basin geometry allow for rapid shoreline
progradation (Bijkerk et al.,
2016
; Wang et al.,
2019
) or if the river is tempo
-
rarily disconnected with the shelf edge (Van Heijst & Postma,
2001
).
Global compilations of lowland river deltas show avulsions preferentially
occur at a relatively fixed location, termed the avulsion node (Figures
1a
and
1b
). The avulsion node is found at a distance upstream of the shoreline
that scales with the backwater length-scale,
퐿퐿
퐴퐴
퐿퐿
푏푏
(1)
where
퐴퐴퐴퐴
퐴퐴
is the avulsion length measured along the channel from the river
mouth to the avulsion location, and
퐴퐴퐴퐴
푏푏
is the backwater length-scale, defined
as the ratio of channel depth,
퐴퐴퐴퐴
푐푐
,
to average bed slope,
퐴퐴퐴퐴
(i.e.,
퐴퐴퐴퐴
푏푏
=
퐻퐻
푐푐
푆푆
;
Figure
1c
) (Chatanantavet et al.,
2012
; Jerolmack & Swenson,
2007
). The
backwater length-scale approximates the river reach where sea level influ
-
ences river hydraulics—termed the backwater zone—which extends for
hundreds of kilometers for large, low-sloping rivers (Lamb et al.,
2012
;
Paola & Mohrig,
1996
). Deltas exhibiting this scaling are referred to as
backwater-scaled deltas. In this study, we focus on backwater-scaled deltas
and the regional backwater effects that can span entire deltas, rather than
smaller scale backwater effects that can influence more local sedimentation
processes (Shaw & McElroy,
2016
; Van Dijk et al.,
2012
).
Recent models have reproduced backwater-scaled deltas (cf. Chatanantavet
et al.,
2012
; Moodie et al.,
2019
; Wu & Nitterour,
2020
). Simulations show
cycles of low flows and high flows in the backwater zone are responsible
for creating a spatial maximum in deposition that determines the avulsion
site (Chadwick et al.,
2019
; Chatanantavet et al.,
2012
). Models predict
sea-level rise does not affect the avulsion length
퐴퐴
(
퐿퐿
퐴퐴
)
, but could induce
higher aggradation rates at the avulsion node, resulting in more frequent
avulsions (Chadwick et al.,
2020
; Ratliff et al.,
2018
). Avulsion frequency
is expected to increase when sea-level rise is comparable to the aggradation
rate; in this regime, sea-level rise facilitates faster delta-top aggradation and
stifles delta-front progradation. At slower rise rates, avulsion frequency is
unaffected because aggradation associated with progradation outpaces the
effect of sea-level rise. At much higher rise rates, there is simply not enough
sediment to keep pace with sea level; progradation ceases, channels aggrade
as fast as possible given the sediment input, and avulsion frequency reaches
a maximum. Avulsions may cease altogether if the delta drowns (Chadwick
et al.,
2020
; Wu & Nitterour,
2020
) or if offshore basin geometry hinders
Figure 1.
(a) Huanghe delta, China, illustrating modern (solid line) and
abandoned (dotted line) channel pathways and avulsion node (yellow star).
(b) Mississippi River delta, USA. Thin dashed line indicates approximate
shoreline in 1900 A.D. (Gagliano et al.,
1981
). (c) Correlation between
measured avulsion length and computed backwater length-scale (Equation
1
),
which approximates the length of the backwater zone where sea level
causes gradually varied flow (Chatanantavet et al.,
2012
; Jerolmack &
Swenson,
2007
).
21699011, 2022, 7, Downloaded from https://agupubs.onlinelibrary.wiley.com/doi/10.1029/2021JF006422 by California Inst of Technology, Wiley Online Library on [06/10/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Journal of Geophysical Research: Earth Surface
CHADWICK ET AL.
10.1029/2021JF006422
3 of 27
aggradation (Bijkerk et al.,
2016
; Wang et al.,
2019
). Avulsions typically take decades to centuries or longer to
occur in nature, so direct observations that test these predictions are sparse.
Experimental studies offer an opportunity to observe repeated avulsions at reduced scale. Hypotheses can
be tested under controlled, simplified conditions, often reducing the complexity associated with vegetation,
cohesive floodplains, offshore waves and tides, and other factors capable of influencing individual avulsions
in nature (Finotello et al.,
2019
; Nicholas et al.,
2018
; Piliouras & Kim,
2019
). Generally, experiments have
demonstrated that avulsions are associated with cycles of local shoreline progradation and lobe construction (de
Haas et al.,
2016
; Reitz et al.,
2010
; Van Dijk et al.,
2012
). However, until recently, most experiments featured
relatively steep surface slopes, braided rivers, and minimal backwater effects (cf. Muto & Steel,
2004
; Reitz &
Jerolmack,
2012
; Wickert et al.,
2013
), producing hydrodynamics and sediment transport more akin to alluvial
fans and fan deltas than they are to lowland, backwater-influenced deltas (Ganti et al.,
2014
). In a pioneering
experiment, Hoyal and Sheets (
2009
) utilized a constant input water discharge and cohesive sediment mixture
that naturally produced gentle slopes and single-thread channels. Avulsion in their experiment was caused by
shoreline progradation: as the shoreline advanced seaward, the river profile lengthened and its slope declined,
leading to a reduction in sediment transport capacity that promoted aggradation and avulsion. Later experiments
and models supported and expanded upon this finding, demonstrating that progradation—as well as relative
sea-level rise—ultimately set the pace of aggradation and avulsion on lowland deltas (Martin et al.,
2009
; Moodie
et al.,
2019
; Yu et al.,
2017
).
Although progradation may set the pace of avulsion, it does not necessarily control where avulsion takes place.
In the experiments of Hoyal and Sheets (
2009
), avulsion location coincided with weak points in the levee profile
that allowed for maximum fluid shear stress during overbank flows, which could manifest at the scale of the
backwater length (Edmonds et al.,
2009
; Hoyal & Sheets,
2009
). Later models reproduced similar behavior
(Chadwick et al.,
2019
; Ratliff et al.,
2021
), showing that for deltas subjected to a constant water discharge, the
levee and floodplain topography is a dominant control on avulsion location. Levee and floodplain topography
are sensitive to basin boundary and initial conditions (Pierik et al.,
2017
); avulsions are prone to occur where
the basin enforces a downstream increase in levee and floodplain width or decrease in floodplain slope (Prasojo
et al.,
2022
; Ratliff et al.,
2021
). However, in the absence of external forces and with a constant water discharge,
levee and floodplain topography can equilibrate to a state where avulsions have a similar likelihood everywhere
along the river profile (Chadwick et al.,
2019
). In contrast, major avulsions on natural deltas tend to occur at or
near the delta-apex, which in turn tends to scale with the backwater length (Equation
1
).
Avulsion location in nature is thought to originate from variable flow regimes that produce persistent sedimen
-
tation patterns in the backwater zone (Chadwick et al.,
2019
; Chatanantavet et al.,
2012
), but few experiments
have been conducted under such conditions. Chatanantavet and Lamb (
2014
) conducted one such lowland river
experiment with subcritical flows and a variable flood regime; in their experiment, backwater effects caused a
downstream migrating wave of aggradation during low flows, and an upstream migrating wave of incision during
high flows. Over many flow events, this interplay resulted in a spatial maximum in aggradation in the upstream
part of the backwater zone. This finding was reproduced in a later experiment that utilized a larger facility allow
-
ing for repeated river avulsions (Ganti, Chadwick, Hassenruck-Guidpati, Fuller, & Lamb,
2016
; Ganti, Chadwick,
Hassenruck-Gudipati, & Lamb,
2016
). Results demonstrated the spatial maximum in aggradation rate coincided
with the avulsion node, leading to the emergence of a backwater-scaled avulsion length (Equation
1
; Figure
1c
).
Importantly, backwater-scaled aggradation and avulsion were only reproduced in the presence of a variable flood
regime; in a comparison experiment without floods, backwater-scaled avulsions did not occur (Ganti, Chadwick,
Hassenruck-Guidpati, Fuller, & Lamb,
2016
). These experimental investigations have supported numerical
modeling efforts of deltas in nature (Chatanantavet et al.,
2012
; Moodie et al.,
2019
), but did not explore the
effects of sea-level rise and fall on river avulsion patterns.
Here, we build on this recent experimental work (Chatanantavet & Lamb,
2014
; Ganti, Chadwick, Hassenruck-
Gudipati, Fuller, & Lamb,
2016
; Ganti, Chadwick, Hassenruck-Gudipati, & Lamb,
2016
) and present results from
a backwater-scaled lowland delta experiment subjected to systematic sea-level rise and fall. First, we provide
an overview of experimental design, scaling, and methods. Second, we present a walkthrough of experimental
results for different sea-level conditions. Third, we use experimental data to test models of river-delta avulsion
location and frequency, and compare the erosional signatures of avulsions and sea-level change preserved in
21699011, 2022, 7, Downloaded from https://agupubs.onlinelibrary.wiley.com/doi/10.1029/2021JF006422 by California Inst of Technology, Wiley Online Library on [06/10/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Journal of Geophysical Research: Earth Surface
CHADWICK ET AL.
10.1029/2021JF006422
4 of 27
delta stratigraphy. Finally, we discuss implications for predicting future avul
-
sion hazards on densely populated deltas and for interpreting the deposits of
ancient deltas on Earth and Mars.
2.
Materials and Methods
2.1.
Experimental Design
We conducted a laboratory flume experiment called CROF18 in the
Caltech River-Ocean Facility, the same facility used by Ganti, Chadwick,
Hassenruck-Gudipati, Fuller, and Lamb (
2016
), Ganti, Chadwick,
Hassenruck-Gudipati, and Lamb (
2016
); their experiment is referred to as
CROF16. The experimental flume consisted of a 7-m-long, 14-cm-wide
fixed-width river section that flowed into a 5-m-long, 3-m-wide unconfined
ocean basin (Figure
2
). Water and sediment were supplied at the upstream
end, and sea level was controlled using a programmable standpipe at the
downstream end. The basin was initially free of sediment, and over several
hours the flow naturally deposited sediment to form an alluvial river and
delta. We widened the width of the fixed walls in the river section (14 cm)
compared to CROF16 (7 cm) to match the width of naturally formed channels
on the delta top, in order to mitigate flow expansion and associated avulsion
at the river-basin boundary. Further details regarding the experimental facility
can be found in Ganti, Chadwick, Hassenruck-Gudipati, and Lamb (
2016
).
The CROF18 experiment was designed to simulate the simplest possible scenario
that could reproduce backwater-scaled avulsions under changing sea level. We
intentionally excluded complexities associated with cohesive
sediment, flood
-
plains, vegetation, waves, tides, and anthropogenic modification (Brakenridge
et al.,
2017
; Caldwell & Edmonds,
2014
; Finotello et al.,
2019
; Nicholas
et al.,
2018
). These factors can influence avulsions in nature (cf. Piliouras &
Kim,
2019
; Ratliff et al.,
2021
), but previous modeling and simplified experiments
suggest they are not necessary to reproduce backwater-scaled avulsion patterns
to first order (Chadwick et al.,
2019
, Ganti, Chadwick, Hassenruck-Gudipati,
Fuller, & Lamb,
2016
) (Figure
1c
). To produce a laboratory-scale river with
similarities to natural lowland rivers, we selected water discharges such that
flow was subcritical
퐴퐴
(
퐹퐹퐹퐹퐹
1)
and turbulent
퐴퐴
(
푅푅푅푅푅
1000)
(Table
1
) (Kleinhans
et al.,
2014
,
2015
; Paola et al.,
2009
). To achieve sediment transport at such
low discharges, we used low-density sediment: crushed, non-cohesive walnut
shells
퐴퐴
(1300kg∕m
3
)
with near-uniform particle diameter
퐴퐴
(0
.
7mm)
. This sedi
-
ment mixture yielded a normal-flow Shields number that was similar to chan
-
nels in nature (
퐴퐴퐴퐴
∼0
.
2
; Table
1
) and that was high enough to avoid current
ripples and associated scour holes in the laboratory (Kleinhans et al.,
2017
).
Non-cohesive sediment is sufficient to produce avulsions and stratigraphy simi
-
lar to natural avulsive systems (Paola et al.,
2009
; Straub et al.,
2012
), though we
note that channel lateral migration and channel geometry can exhibit differences
compared to lowland river channels in nature with cohesive, vegetated banks
(Dunne & Jerolmack,
2020
; Tal & Paola,
2007
).
To reproduce laboratory-scale backwater zones, we implemented a varia
-
ble flood regime following the example of CROF16. Water and sediment
supply oscillated between a low-flow and high-flow discharge (Table
1
).
Water discharges associated with each flow were selected to produce signif
-
icantly different normal flow depths (7.5 mm during low flows and 11.7 mm
during high flows, a ∼40% difference) to facilitate significant changes in
water-surface slope as flow approached sea level in the backwater zone
(Chatanantavet et al.,
2012
). Furthermore, the durations for each flow
Figure 2.
(a) Schematic of the Caltech River-Ocean Facility. The tank
coordinate system was defined such that the primary flow direction was
eastward. (b) Sea level as a function of run time over the course of the
experiment. Alternate background shading indicates Phases A through F, each
of which featured a different rate of sea-level change.
Low flow
High flow
Water discharge
퐴퐴
[
liters
∕min]
14.4
20.4
Sediment supply
퐴퐴
[g∕min]
30.4
69.4
Normal-flow depth,
퐴퐴퐴퐴
푛푛
[
mm
]
7.5
11.7
Flow duration,
퐴퐴퐴퐴
푓푓푓푓푓푓푓푓
퐴퐴
[min]
22
8
Channel-adjustment timescale,
퐴퐴퐴퐴
푎푎푎푎푎푎
[min]
45
20
Normal-flow transport slope,
퐴퐴퐴퐴
[−]
0.0042
0.0042
Backwater length-scale
*
,
퐴퐴퐴퐴
푏푏
=
퐻퐻
푛푛
푆푆
[m]
1.8
퐴퐴
Froude number,
퐴퐴퐴퐴퐴퐴
[-]
0.59
0.43
Reynolds number,
퐴퐴퐴퐴퐴퐴
[-]
1,200
1,700
Shields number,
퐴퐴퐴퐴
[-]
0.18
0.28
*
The backwater length-scale well approximates the length of the backwater
zone but can be more accurately constrained by accounting for
퐴퐴퐴퐴퐴퐴
and
variable discharges (Equation
3
) (Bresse,
1860
; Lamb et al.,
2012
).
Table 1
Variable Flood Regime of the CROF18 Experiment
21699011, 2022, 7, Downloaded from https://agupubs.onlinelibrary.wiley.com/doi/10.1029/2021JF006422 by California Inst of Technology, Wiley Online Library on [06/10/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Journal of Geophysical Research: Earth Surface
CHADWICK ET AL.
10.1029/2021JF006422
5 of 27
(
퐴퐴퐴퐴
푓푓푓푓푓푓푓푓
=
22 min for low flows,
퐴퐴퐴퐴
푓푓푓푓푓푓푓푓
=
8 min for high flows) were selected to be significantly shorter than the
time required to adjust the backwater zone to uniform-flow conditions (∼45 min for low flow, ∼20 min for high
flow) estimated using the channel-adjustment timescale,
퐴퐴퐴퐴
푎푎푎푎푎푎
=
퐿퐿
푏푏
퐵퐵
푐푐
Δ
퐻퐻
푄푄
푠푠
(
1−
휆휆
푝푝
)
,
where
퐴퐴퐴퐴
푏푏
is backwater length,
퐴퐴퐴퐴
푐푐
is channel width,
퐴퐴퐴퐴
푝푝
is sediment porosity,
퐴퐴퐴퐴
푠푠
is the sediment supply for a given flow event, and
퐴퐴
Δ
퐻퐻
is the
amplitude of bed-elevation adjustments in the backwater zone (estimated by the difference between the high-flow
and low-flow normal flow depths; Chatanantavet & Lamb,
2014
). The channel-adjustment timescale is based on
the simplified scenario where the full sediment supply
퐴퐴
(
푄푄
푠푠
)
is deposited within the backwater zone, representing
a shortest-possible adjustment time; for the more general case where part of the sediment supply is deposited
offshore or on the floodplain, channels should take longer to adjust. Because flow duration
퐴퐴퐴퐴
푓푓푓푓푓푓푓푓
was shorter
than the adjustment time
퐴퐴퐴퐴
푎푎푎푎푎푎
(Table
1
), the backwater zone was kept in a state of perpetual adjustment similar
to natural deltaic rivers: riverbed elevation and slope oscillated relative to a mean, convex-upward profile that
reflected ongoing competition between low-flow deposition and high-flow erosion (Arkesteijn et al.,
2019
; Blom
et al.,
2017
). Previous numerical simulations have demonstrated that this condition
퐴퐴
(
푇푇
푓푓푓푓푓푓푓푓
푇푇
푎푎푎푎푎푎
<
1)
is necessary for
backwater-scaled avulsions, and that dynamics are relatively insensitive to the flow duration so long as this condi
-
tion is upheld (Chadwick et al.,
2019
). With this in mind, we designed the experiment to satisfy this condition,
and relaxed the precise ratio between flow duration and adjustment (
퐴퐴
푇푇
푓푓푓푓푓푓푓푓
푇푇
푎푎푎푎푎푎
∼0
.
5;
Table
1
) compared to what is
usually observed in nature (
퐴퐴
푇푇
푓푓푓푓푓푓푓푓
푇푇
푎푎푎푎푎푎
∼0
.
0001−0
.
1
; Chadwick & Lamb,
2021
) to ensure the experiment was physi
-
cally and logistically feasible.
At the upstream end, we covaried sediment supply with water discharge to produce the same normal-flow bed
slope
퐴퐴
(
푆푆
=0
.
0042)
for high flow and low flow, which was gentle enough to maintain subcritical flow
퐴퐴
(
퐹퐹퐹퐹퐹
1)
and a single-thread channel. The flow depth and slope resulted in a backwater length-scale of
퐴퐴퐴퐴
푏푏
=
퐻퐻
푐푐
푆푆
=1
.
8m
. It
was necessary that
퐴퐴퐴퐴
푏푏
be much greater than the channel width (∼0.2 m) to observe a significant backwater reach,
but also small enough to fit within the laboratory basin (Figure
2a
). Water and sediment discharges in CROF18
were similar to CROF16, but were adjusted slightly to improve dynamic scaling with sea-level rise rates.
To isolate the effect of sea-level change on backwater-scaled delta dynamics, we systematically raised and lowered
sea level at six different speeds during six distinct phases of the experiment, lettered A–F (Table
2
). Rise rate was
constant during each phase. Following previous work, we scaled laboratory sea-level rise and fall rates relative
to the characteristic rate of aggradation (Chadwick et al.,
2020
). The dimensionless sea-level rise rate is given by
휎휎
휎휎
퐻퐻
푐푐
푛푛푛푛
푐푐
(2)
where
퐴퐴퐴퐴
푐푐
is the channel-filling timescale approximated by
퐴퐴퐴퐴
푐푐
퐻퐻
푐푐
퐵퐵
푐푐
퐿퐿
푏푏
푄푄
푠푠
(
1−
휆휆
푝푝
)
(Chatanantavet & Lamb,
2014
;
Reitz & Jerolmack,
2012
),
퐴퐴퐴퐴
=(
푁푁
+1)∕2
is the number of avulsions before a given deltaic lobe is reoccupied,
and
퐴퐴퐴퐴
is the average number of delta lobes, for simplicity here taken to be
퐴퐴퐴퐴
=4
(Ganti et al.,
2019
). The
denominator of Equation
2
is a first-order estimate of the maximum possible aggradation rate, in the hypothetical
limit where all sediment is deposited uniformly across the backwater-scaled topset. Most modern deltas fall in
the range
퐴퐴
0
<휎휎
<
1
(Chadwick et al.,
2020
). Using estimates of Milankovich-scale sea-level cycles over the
Pleistocene and Miocene (
∼120m
of rise and fall over
∼100ky
for Pleistocene, and
∼30m
of rise and fall over
Phase A
Phase B
Phase C
Phase D
Phase E
Phase F
Dimensionless sea-level rise rate,
퐴퐴퐴퐴
[−]
0
0.08
0.33
1.33
−0.03
−0.67
Sea-level rise rate,
퐴퐴퐴퐴
[
mm
hr
]
0
0.25
1
4
−0.1
−2
Run time
퐴퐴
[hr]
0–43.5
43.5–82
82–101
101–105
105–140
140–163.5
Duration,
퐴퐴
[hr]
43.5
38.5
19
4
35
23.5
Number of low flows [-]
87
77
38
8
70
47
Number of high flows [-]
87
77
38
8
70
47
Table 2
Phases of the CROF18 Experiment and Associated Sea-Level Change
21699011, 2022, 7, Downloaded from https://agupubs.onlinelibrary.wiley.com/doi/10.1029/2021JF006422 by California Inst of Technology, Wiley Online Library on [06/10/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Journal of Geophysical Research: Earth Surface
CHADWICK ET AL.
10.1029/2021JF006422
6 of 27
∼40ky
for Miocene; Li et al.,
2016
), we estimate many deltas fall in the range
−1
<휎
<
1
. CROF18 Phases A,
B, C, E, and F were selected to fall within this range, and correspond to rise rates of
−3mm∕hr
<휎<
3mm∕hr
(Table
2
). Phase D was designed to explore the case where sea level rises faster than the maximum aggradation
rate
퐴퐴
(
휎휎
>
1)
, a possible scenario for many lowland deltas in the next century (Chadwick et al.,
2020
). While
we did not incorporate subsidence into the experiment, delta response to uniform subsidence is expected to be
mechanically similar to that of sea-level rise (González & Tornqvist,
2006
; Reitz et al.,
2015
). We allowed each
phase to continue long enough to allow for many avulsions. Phases were also kept brief enough that the offshore
basin depth did not change by more than a factor of two (Figure
2b
); this allowed us to mitigate the effect of
changing basin depth on avulsion patterns (Carlson et al.,
2018
; Wang et al.,
2019
), and therefore better isolate
the effect of sea-level rise rate.
2.2.
Data Collection
Overhead images of delta evolution were collected every minute using six mounted cameras that bordered the
experimental facility. Photos from each camera were concatenated to ensure a wide field of view that extended
beneath railings in the facility. The water was dyed using a fluorescent green dye, allowing for visual distinction
between subaerial and submerged surfaces even for shallow (
∼1
cm) water depths under ultraviolet light fixtures.
Before starting a flow event, we inserted
∼0
.
5
gallons of dye into the end tank. We ran the flow using a very low
discharge
퐴퐴
(
푄푄
=0
.
002L∕min)
with no sediment feed for
∼12
hours of standby to allow the dye to disperse evenly
through the tank without mobilizing sediment or disturbing the delta.
An ultrasound distance meter (Massa M-5000/220, Massa, Hingham, MA) and laser displacement sensor
(Keyence LK-G5000, Keyence Corporation of America, Itasca, IL) measured water surface elevation at 1-mm
vertical resolution and bed topography at 0.1-mm resolution. Before each flow event, we adjusted the weir siphon
such that sea level followed the curve in Figure
2b
; we verified sea level remained in equilibrium with the weir
within
퐴퐴
0
.
1mm
by scanning a thin piece of floating wax paper at the downstream end using the laser sensor. At
the beginning of each flow event, we collected water surface elevation data, and after each event we switched off
the flow and measured bed surface elevation. Topographic scans included a long profile through the river section
and basin section along the flume centerline (3-mm horizontal resolution), as well as a series of cross sections
in the basin perpendicular to the flume axis (3-mm resolution in the cross-stream direction, spaced every 15 cm
in the downstream direction). The basin water level was maintained during bed topography scans, and data for
submerged parts of the delta were corrected for the refraction index of the laser beam through still water. The
green color of the fluorescent dye in the basin ensured transmission of the red laser signal. Summed errors asso
-
ciated with vertical precision of the instrument and the refraction-index correction were
±0
.
1mm
. Raw data from
topographic scans were denoised using a median filter of kernel size 1.5 cm.
2.3.
Data Analysis
Concatenated overhead images were used to map channel avulsions and shoreline evolution in the experiment.
Following previous work, we identified avulsions as the establishment of a new channel (the daughter channel)
that captured the majority of flow through consecutive flow events, and the old channel (the parent channel) was
partially or completely abandoned (Ganti, Chadwick, Hassenruck-Gudipati, Fuller, & Lamb,
2016
). Avulsion
location and time were measured as the location and time when the levee breach in the parent channel initiated.
Manual identification of avulsions involved a degree of subjectivity; still, our measurements for avulsion location
have an uncertainty of less than a channel width and much less than a backwater length-scale, and measurements
for avulsion time have an uncertainty of approximately 1 minute (Ganti, Chadwick, Hassenruck-Gudipati, Fuller,
& Lamb,
2016
; Ganti, Chadwick, Hassenruck-Gudipati, & Lamb,
2016
). Avulsion length
퐴퐴
(
퐿퐿
퐴퐴
)
was computed
as the distance along the parent channel from the river mouth to the avulsion location. Avulsion frequency was
calculated using
퐴퐴퐴퐴
퐴퐴
=1∕
푇푇
퐴퐴
, where
퐴퐴퐴퐴
퐴퐴
is the time between the current avulsion event and the previous avulsion
event. We manually mapped the delta shoreline, following the boundary of fluorescent green water and the brown
sediment surface. We also independently mapped the location of the topset-foreset break; during sea-level rise the
shoreline sometimes retreated upstream from the topset-foreset break.
We measured channel aggradation and erosion using the time series of topographic scans collected by the laser
sensor at the end of each flow event. For cases when topographic scans were collected within
퐴퐴
4min
of an avulsion
21699011, 2022, 7, Downloaded from https://agupubs.onlinelibrary.wiley.com/doi/10.1029/2021JF006422 by California Inst of Technology, Wiley Online Library on [06/10/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License