You are on page 1of 15

21699011, 2018, 3, Downloaded from https://agupubs.onlinelibrary.wiley.com/doi/10.1002/2017JF004202 by Cochrane Mexico, Wiley Online Library on [25/10/2022].

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
RESEARCH ARTICLE Turbidity Current Dynamics: 2. Simulating Flow Evolution
10.1002/2017JF004202
Toward Equilibrium in Idealized Channels
This article is a companion
to Traer et al. (2018),
https://doi.org/10.1002/2017JF004200.
M. M. Traer1 , A. Fildani2 , O. Fringer3 , T. McHargue1 , and G. E. Hilley1
1 Department of Geological Sciences, Stanford University, Stanford, CA, USA, 2 Statoil, Austin, TX, USA, 3 Department of Civil
Key Points: and Environmental Engineering, Stanford University, Stanford, CA, USA
• Turbidity currents simulated using
a modified four-equation model may
evolve to a single equilibrium state
regardless of initial conditions Abstract This study simulates turbidity currents through a number of idealized channels using a steady,
• Simulated turbidity currents generally one-dimensional, depth-averaged model to determine if modeled flows starting from a wide range of
required 1–40 km of downflow initial conditions might reach an equilibrium state where clear-water entrainment balances fluid mass lost
distance to adjust from one
equilibrium state to another to flow stripping or overspill processes. To accomplish this, we calculated flow dynamics based on 1,000 sets
• That modeled flows adjust to new of randomized initial conditions, identified flows that successfully traversed the system, and then extracted
equilibria over tens of kilometers the flow height, velocity, and sediment flux at the channel terminus. We then systematically changed
suggests that they are less sensitive
to upstream conditions in long channel and flow properties using a wide range of values and calculated the length required for the
channels simulated currents to reach the new equilibrium conditions. We found that modeled turbidity currents may
evolve to a single equilibrium state consistent with the physical channel and flow properties regardless
Supporting Information: of the initial flow conditions. Additionally, we found that simulated turbidity currents generally required
• Supporting Information S1 1–40 km of downflow distance to adjust to new equilibrium conditions resulting from changes in the
channel slope, channel height, channel width, coefficient of bed friction, and meander radius of curvature.
Correspondence to: Changes in sediment grain size and suspension cloud concentration resulted in adjustment lengths
M. M. Traer, exceeding 100 km. Shorter adjustment lengths resulting from flow stripping suggest that this process might
mtraer@stanford.edu
play an important role in flow filtering. The fact that modeled flows adjust to new equilibria over tens of
kilometers suggests that they are less sensitive to upstream flow conditions in longer channels.
Citation:
Traer, M. M., Fildani, A., Fringer, O.,
McHargue, T., & Hilley, G. E. (2018).
Turbidity current dynamics: 1. Introduction
2. Simulating flow evolution
toward equilibrium in idealized
One- and two-dimensional depth-averaged models for turbidity currents remain popular due to relatively
channels. Journal of Geophysical fast flow simulation times that allow for investigations of first-order flow properties over a broad range of
Research: Earth Surface, 123, 520–534. conditions (e.g., Cartigny et al., 2011; Kostic, 2011; Lamb et al., 2010; Pantin & Franklin, 2009; Sun & Parker,
https://doi.org/10.1002/2017JF004202
2005). These models have typically shown that flow heights grow quickly due to clear-water entrainment
when neglecting processes that remove fluid mass from the flow (e.g., Das et al., 2004; Traer et al., 2012, 2015).
Received 5 JAN 2017 The resulting currents often reach physically unrealistic values in relation to the channels the flows traverse
Accepted 10 JAN 2018
(Traer et al., 2015). In part 1 of this paper, we developed methods that allowed one-dimensional models to
Accepted article online 31 JAN 2018
Published online 12 MAR 2018
capture two-dimensional flow effects, such as flow stripping and overspill, both of which might play important
roles in regulating turbidity current dynamics (e.g., Hiscott et al., 1997; Straub et al., 2008; Traer et al., 2015).
Based on consistency of depositional patterns on channel levees (e.g., Hiscott et al., 1997; Skene et al., 2002),
consistency of deposits within channel elements (e.g., McHargue et al., 2011), and investigations of flow fil-
tering at channel meanders (e.g., Straub et al., 2008), a growing number of researchers have suggested that
uniform flow equilibrium conditions exist. We distinguish this state of equilibrium defined by the fluid mass
balance from the state of depositional equilibrium when erosion balances deposition. In the companion
paper, we found that turbidity currents could evolve toward an approximate dynamic equilibrium as fluid
incorporated into the flow due to turbulent mixing was balanced by fluid advected from the flow to the strip-
ping and overspill processes. This equilibrium condition might allow flows to extend into deepwater basins,
even on low slopes, complimenting previous work on the enigmatic long runout distances (e.g., Mohrig et al.,
1998; Pirmez & Imran, 2003; Talling et al., 2007).
In part 1 of this paper, we showed that flow equilibrium conditions were sensitive to changes in the local
channel and flow properties, specifically changes in slope, channel height, sediment grain size, and suspen-
©2018. American Geophysical Union. sion cloud concentration. The fact that equilibrium conditions changed due to alterations of the local channel
All Rights Reserved. and flow properties suggests that turbidity currents might evolve from one equilibrium state to another

TRAER ET AL. TURBIDITY CURRENT EQUILIBRIUM 520


21699011, 2018, 3, Downloaded from https://agupubs.onlinelibrary.wiley.com/doi/10.1002/2017JF004202 by Cochrane Mexico, Wiley Online Library on [25/10/2022]. 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 10.1002/2017JF004202

as they traverse longer systems. Several previous studies have found that
modeled turbidity currents are sensitive to the initial flow conditions
(e.g., Pantin & Franklin, 2009; Parker et al., 1986). Yet if turbidity currents
evolve through a series of equilibria, then flows would more likely be sensi-
Cs
tive to local conditions that exert a primary control on the equilibrium state
h than upstream or inlet flow conditions. Additionally, given that we identi-
Suspension cloud fied both Richardson supercritical and subcritical equilibrium values, it is
possible that transformations from one equilibrium state to another might
Turbidity current
accompany changes in the flow regime. Therefore, understanding how
C hc turbidity currents evolve through different states of equilibrium might
help better interpret patterns of erosion and deposition associated with
different flow regimes.

w=100m In this study, we use the model proposed and described in part 1 of this
paper to simulate turbidity currents through a series of idealized channels
Figure 1. Schematic cross section of the idealized channel and turbidity to examine whether or not turbidity currents evolve toward a constant
current with suspension cloud.
equilibrium state regardless of the initial flow conditions. We additionally
analyze how changes in the local channel and flow properties affect the
distance required for turbidity currents to evolve from one equilibrium state to another. We find that simu-
lated turbidity currents broadly converge to constant equilibrium values and that currents are generally more
sensitive to local channel and flow conditions than upstream or initial conditions, given the typical length
scales of submarine channels and the distance over which modeled flows adjust to changing conditions. Our
analysis helps elucidate the channel and flow properties that exert primary controls on flow evolution and
control the length and time scales over which turbidity currents might evolve from one equilibrium state
to another.

2. Methods
2.1. Synthetic Channel Geometries
To test the effects of flow stripping and overspill on turbidity currents evolving toward equilibrium, we simu-
lated flows through a number of synthetic channels (Figure 1). The first type of channel had a constant channel
height (hc ), channel width (w), slope (S), and radius of curvature (r), consistent with an infinitely spiraling channel
(hereafter referred to as the infinite spiral channel). The second type of channel had 15 identical, quarter-turn
(90∘ ) meanders with alternating senses of rotation, separated by identical, straight channel segments. This
channel type is hereafter referred to as the equal-spaced channel. The third channel type had 15 meanders
separated by straight channel segments, but the radii of curvature of the meanders and the length of the
channel segments were chosen at random from ranges of 100–10,000 m. This is hereafter referred to as the
random channel. All synthetic channels used for an individual flow simulation maintained constant channel
height, width, and slope (i.e., the channels did not get progressively shorter/taller, narrower/wider, less/more
concave as the flow progressed downslope). This ensured that any changes in flow height, and the resulting
changes in velocity and concentration, resulted from the mass advection processes.
2.2. Simulations
The governing equations and empirical relations required for their closure are described in part 1 of this con-
tribution. In the first part of our analysis, we investigated whether or not flows evolved to constant equilibirum
conditions regardless of the initial flow conditions. Initial flow heights (ho ), velocities (Uo ), sediment fluxes
(𝜓o ), and levels of turbulence (Ko ) were chosen randomly from ranges
thought reasonable based on previous field and laboratory measurements
Table 1 and numerical modeling studies (Table 1). We simulated 100 turbidity cur-
Range of Sampled Starting Flow Conditions rents traversing the infinite spiral, equal-spaced, and random channels. For
Flow property Range Units the flows traversing the infinite spiral channel, and for consistency with pre-
viously published work, we initially set the channel slope equal to 0.05, the
Height (h) 0–50 m
channel width equal to 100 m, and the channel aspect ratio (w∶hc ) to 10:1.
Velocity (U) 0–30 m/s
Based on a number of well-studied systems collected by Traer et al. (2015),
Concentration (C ) 0–0.2 —
we initially set the radius of curvature equal to 1 km. The equal-spaced chan-
Level of turbulence (K ) 10−3 – 101 m2 /s2
nel shared these characteristics, and the straight channel segments had

TRAER ET AL. TURBIDITY CURRENT EQUILIBRIUM 521


21699011, 2018, 3, Downloaded from https://agupubs.onlinelibrary.wiley.com/doi/10.1002/2017JF004202 by Cochrane Mexico, Wiley Online Library on [25/10/2022]. 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 10.1002/2017JF004202

Table 2
Base Case Equilibrium Values
Flow Property Flow stripping Overspill, 𝛾 = 10−1 Overspill, 𝛾 = 10−3
h (m) 10.19 19.34 43.45
U (m/s) 8.29 10.66 14.51
C 1.32 × 10−1 1.15 × 10−1 9.48 × 10−2
𝜓 (m3 s−1 ) 11.15 × 103 23.71 × 103 59.77 × 103
K (m2 s−2 ) 3.25 5.42 10.10
Ri 0.3156 0.3160 0.3163
Note. Base case values for channel and flow properties: S = 0.05, hc = 10 m, cD∗ = 0.004,
r = 1,000 m, Ds = 0.01 cm, Cs = 0.1 × C , 𝛼 = 0.1, w = 100 m.

a length equal to half the radius of curvature (=500 m). The infinite spiral channel shared the channel slope,
width, and height as the infinite spiral and equally spaced channel, but the radii of curvature of the mean-
ders and the length of the straight channel segments were chosen at random from ranges of 100–10,000 m.
We identified the turbidity currents that successfully traversed the various channel types and compared their
flow conditions (h, U, 𝜓, K ) at the channel terminus.
For the second part of our analysis, we investigated the distance over which simulated turbidity currents could
evolve from one equilibrium state to another. All simulated flows started with base case equilibrium values
(Table 2), appropriate for a flow at equilibrium traversing the infinite spiral channel described above. We then
systematically changed the channel (slope, channel height, width, bed friction coefficient, radius of curva-
ture) and flow (sediment grain size and suspension cloud concentration) properties across a range capturing
their physical extent (Table 3). We extracted the downflow distance required to adjust to the new equilibrium
values consistent with the altered channel and flow properties. Longer adjustment lengths indicated that the
initial conditions still exerted a strong influence over the flow dynamics, while shorter adjustment lengths
indicated that the flow dynamics were more sensitive to local channel and flow conditions. The simulated
flows traversed a channel 1,000 km in length to allow ample distance to adjust to the new equilibrium condi-
tions. For completeness, we extracted the distance and time (using the integrated flow velocity) required for
each of the flow conditions (height, velocity, sediment flux, and level of turbulence) to reach within 10 and
5% of the new equilibrium value relative to the difference between the new equilibrium value and the base
case equilibrium value, shown mathematically as

E10,5 = Enew ± 0.10, 0.05 ∗ |(Enew − Ebase )| (1)

where Enew is a flow condition at the channel terminus (equivalent to the new equilibrium value), Ebase is the
corresponding base case flow condition, and E10,5 is the extracted value used as a proxy for when the flow
reequilibrated. Flows that reached the E10,5 conditions are hereafter said to have reached within 10 and 5% of
the new equilibrium values.
To complete this study, we repeated the adjustment distance analysis but allowed each of the channel
and flow inputs to covary. We varied each input by 50 evenly spaced intervals between the minimum and
maximum values reported in Table 3. Each input was covaried with all other
inputs, and we extracted the resulting adjustment distance to investigate
Table 3
Range of Channel and Flow Properties Used When Determining the Turbidity how compounding changes in channel and flow properties affected the
Current Adjustment Length downstream evolution of the simulated turbidity currents and how these
changes affected the sensitivity to local and initial flow conditions. Given the
Channel/flow property Range Units
sensitivity analysis we explored in the companion paper, particular atten-
Slope (S) 0.01–0.05 —
tion was paid to joint changes in slope, channel height, width, sediment
Channel height (hc ) 5–20 m
grain size, and suspension cloud concentration.
Channel width (w) 50–200 m
Bed friction coefficient (cD∗ ) 0.004–0.02 — 2.3. Solution Methods
Downstream changes in flow dynamics were calculated using an implicit,
Radius of curvature (r) 10–10,000 m
adaptive space-stepping algorithm (MATLAB, 2011), benchmarked to flow
Grain size (Ds ) 0.004–0.0375 cm
simulations from part 1 of this paper. We utilized an implicit numerical inte-
Suspension cloud concentration (Cs ) 0.01–0.1 —
gration routine due to the stiff nature of the system of equations that would

TRAER ET AL. TURBIDITY CURRENT EQUILIBRIUM 522


21699011, 2018, 3, Downloaded from https://agupubs.onlinelibrary.wiley.com/doi/10.1002/2017JF004202 by Cochrane Mexico, Wiley Online Library on [25/10/2022]. 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 10.1002/2017JF004202

5 have required a small integration step when using a comparable explicit


method. The adaptive step integration algorithm adjusted the integration
h/equilh

step to preserve a prescribed relative error tolerance of 1 × 10−5 . In addi-


tion to the relative tolerance, we prescribed an absolute tolerance for each
flow property at least 1 order of magnitude below the expected values
0
3 (h = 1 × 10−4 m; U = 1 × 10−4 m/s; C = 1 × 10−5 ; K = 1 × 10−3 m2 ∕s2 ).
When any flow property fell below its tolerance value, the flow was consid-
U/equilU

ered to have arrested. If either error tolerance was not met, the integration
step size was decreased until they were. The combination of the implicit
solver and adaptive step algorithms allowed for numerical stability while
0
6 preserving a high order of accuracy.
C/equilC

3. Results
3.1. Uniqueness of Equilibria
To determine whether equilibrium values were sensitive to initial con-
0
0.6 ditions, we simulated 100 turbidity currents using random starting flow
heights, sediment fluxes, velocities, and turbulent kinetic energies drawn
from uniform distributions and extracted those flows that successfully
Ri

traversed the channel. For this study, we were only interested in flow char-
acteristics that might reasonably be measured and so we present analyses
0.2
0 10 20 30 40 50 of flow heights, velocities, and sediment fluxes (a full analysis of the level
Down−channel distance (km) of turbulence is provided in the supporting information). Seventy-six per-
cent of the flows modeled with random starting conditions successfully
Figure 2. Turbidity current evolution in the idealized, infinite spiral channel
traversed the infinite spiral channel (Figure 2). Regardless of the initial con-
for 25 randomly sampled flows modeled with flow stripping and beginning
from random starting (from top to bottom) flow height, concentration, ditions, all flows that successfully traversed the channel reached the same
velocity, and Richardson number. Each curve represents an individual flow. equilibrium height, velocity, sediment flux, and flow Richardson number.
These flows, similar to nearly every flow that reached equilibrium, had to
entrain significant sediment from the bed to maintain the required density
contrast. While the flow height and Richardson number reached equilib-
5 rium within several kilometers, the velocity and sediment flux required
longer distances (≈20–35 km) to adjust to the new equilibrium condition.
h/equilh

In the equal-spaced channel, the flow trajectories exhibited a sawtooth


pattern; stripping maintained equilibrium at the channel meanders, while
0 the flow expanded at the straight channel segments where stripping was
1.5
not present to balance clear-water entrainment (Figure 3). While the alter-
U/equilU

nating straight and curved channel segments produced flows that reached
an oscillating quasi-equilibrium, the flow height, velocity, sediment flux,
and Richardson number all effectively converged on the same values at
0 each point along the final few meanders, with the maximum and mini-
1.5
mum values differing by less than 1%. Sixty-two percent of flows traversed
C/equilC

the equal-spaced channel, slightly lower than flows simulated in the infi-
nite spiral channel due to the rapid loss of the suspension cloud at the first
channel meander that caused flows to arrest. Flow height and Richardson
0 number reached equilibrium earlier (≈1–3 km) than the velocity and
0.6
sediment flux (≈20–30 km).
To complete our analysis of flow stripping, we simulated a turbidity cur-
Ri

rent traversing 100 different random channels (Figure 4). We chose to


begin our flow at ignition (in the sense of Parker et al., 1986: ho = 2 m,
0.2
0 10 20 30 Uo = 0.801 m/s, 𝜓o = 0.976 m3 /s, Ko = 1.40 × 10−2 m2 /s2 ) to increase the
Down−channel distance (km) proportion of flows that would successfully traverse the different channels.
The heights, velocities, and sediment fluxes of flows traversing the random
Figure 3. Turbidity current evolution in the idealized, equal-spaced channel
for 25 randomly sampled flows modeled with flow stripping and beginning channels showed broad agreement with the equilibrium values modeled
from random starting (from top to bottom) flow height, concentration, in the infinite spiral channel and the equal-spaced channel but contained
velocity, and Richardson number. Each curve represents an individual flow. quantitatively higher variance. Within the final meander, the minimum

TRAER ET AL. TURBIDITY CURRENT EQUILIBRIUM 523


21699011, 2018, 3, Downloaded from https://agupubs.onlinelibrary.wiley.com/doi/10.1002/2017JF004202 by Cochrane Mexico, Wiley Online Library on [25/10/2022]. 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 10.1002/2017JF004202

5 and maximum recorded flow heights differed by 9%. Additionally, the


mean flow height recorded in the random channel was approximately 3%
h/equilh

higher than the base case equilibrium flow height. The minimum and max-
imum flow velocities differed by 10%, while the minimum and maximum
sediment flux values differed by 24%. The mean flow velocity and sedi-
0
1.5 ment flux within the final meander were approximately 6 and 14% higher
than the base case values, respectively. The flow Richardson number also
U/equilU

showed higher variance, with the minimum and maximum values differ-
ing by 10%. Yet despite the variance between these modeled flow heights,
velocities, and sediment fluxes, flows simulated in the random channels
0
1.5 effectively reached the same Richardson number as flows modeled in the
infinite spiral channel and the equal-spaced channel.
C/equilC

As with our flow stripping analysis, we simulated 100 turbidity currents


with overspill and used random starting flow heights, sediment fluxes,
0 velocities, and turbulent kinetic energies drawn from uniform distributions
0.6 to determine whether the equilibrium conditions were consistent across
a range of starting flow conditions (Figures 5 and 6). Using the smaller 𝛾
Ri

value, 82% of flows successfully traversed the channel, while when using
the larger 𝛾 value, 79% of flows successfully traversed the channel. Using
0.2
the smaller 𝛾 value, the flow height and Richardson number reached equi-
0 30 60 90 librium between 30 and 75 km, while velocity and sediment flux required
Down−channel distance (km) between 90 and 130 km to reach equilibrium. All flow heights, velocities,
Figure 4. Turbidity current evolution in the idealized, random channel for
and sediment fluxes simulated using the smaller 𝛾 value converged to the
25 randomly sampled flows modeled with flow stripping and beginning same equilibrium values. Using the larger 𝛾 value, the flow height and
from random starting (from top to bottom) flow height, concentration, Richardson number reached equilibrium between 10 and 20 km, while
velocity, and Richardson number. Each curve represents an individual flow. velocity and sediment flux required approximately 50–60 km to reach
equilibrium. All flow heights, velocities, and sediment fluxes simulated
using the larger 𝛾 value converged to the same equilibrium values. Flows
5 modeled with overspill required longer distances to reach equilibrium
than flows modeled with flow stripping.
h/equilh

3.2. Equilibrium Adjustment


3.2.1. Varying Model Inputs
0 To determine how along-flow changes in channel and flow properties
3
affected flow dynamics, we simulated turbidity currents starting with the
U/equilU

base case equilibrium values, systematically changed the input channel


and flow properties using ranges reported in Table 3, and then calculated
the distance required for the current’s height, velocity, and sediment flux
0 to reach within 10 and 5% of the new equilibrium values. For each of
3
the channel and flow properties, we extracted the maximum adjustment
C/equilC

length, referred to as the adjustment length, to isolate those variables


that exerted a primary control on this length scale. The ranges reported in
Table 3 highlight the uncertainties associated with the model inputs. For
0 individual systems it might be easier to constrain measurable channel or
0.6 flow properties. Yet other values, such as the bed friction coefficient or sus-
pension cloud concentration, remain difficult to constrain. Our analysis is
Ri

focused on more general flow behavior not associated with any particular
system, so we explored the entire range of each of these model inputs.
0.2
0 50 100 150 Using flow stripping, the adjustment distance was broadly on the order
Down−channel distance (km) of tens of kilometers (Table 4). Flows took longest to adjust to changes
in the fractional suspension cloud concentration, followed by changes in
Figure 5. Turbidity current evolution in the idealized, infinite spiral channel
channel height and grain size. The flow height reequilibrated in less than
for 25 randomly sampled flows modeled with overspill, 𝛾 = 10−3 , and
beginning from random starting (from top to bottom) flow height, 1.5 km for changes in all variables except grain size and fractional sus-
concentration, velocity, and Richardson number. Each curve represents an pension cloud concentration, which required between 50 and 95 km to
individual flow. reequilibrate. Flow velocity required between 19 and 49 km to reach 5%

TRAER ET AL. TURBIDITY CURRENT EQUILIBRIUM 524


21699011, 2018, 3, Downloaded from https://agupubs.onlinelibrary.wiley.com/doi/10.1002/2017JF004202 by Cochrane Mexico, Wiley Online Library on [25/10/2022]. 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 10.1002/2017JF004202

5 of the new equilibrium value, depending upon the input variable changed.
The sediment flux required similar distances to reequilibrate as the flow
h/equilh

velocity and took 23–54 km to reach within 5% of the new equilibrium


values depending upon the input variable changed. The Richardson num-
ber required 0.5–71 km to reequilibrate. All flows reached within 10% of
0
3 the new equilibrium values within 82 km, with a median value of 20 km.
Often, the distance required to reach within 5% of the new equilibrium
U/equilU

state was approximately 1.25 times the distance required for the flows to
reach within 10% of the new equilibrium state.
0 When isolating overspill processes, the adjustment distances were greater
4 than those for flows modeled with flow stripping, with the largest adjust-
ment distances associated with simulations using the smaller 𝛾 value
C/equilC

(Tables 5 and 6). Similar to flow stripping, flows took longest to reequi-
librate with changes in the suspension cloud concentration, followed by
0 changes in channel height, width, and grain size. Yet unlike the flow strip-
0.6 ping simulations, changes in the suspension cloud concentration resulted
in adjustment lengths that were consistently 2 to 5 times longer than all
Ri

other adjustment lengths. Flow heights generally reequilibrated in the


shortest distances, reaching within 5% of the new equilibrium values at
0.2 distances of 121–295 km using the smaller 𝛾 value and 53–78 km using
0 20 40 60 80
the larger 𝛾 value as we varied the suspension cloud concentration and
Down−channel distance (km) grain size. Changes to all other variables produced adjustment lengths less
Figure 6. Turbidity current evolution in the idealized, infinite spiral channel than 9 km and less than 3 km for the smaller and larger 𝛾 values, respec-
for 25 randomly sampled flows modeled with overspill, 𝛾 = 10−1 , and tively. Flow velocities reached within 5% of the new equilibrium values at
beginning from random starting (from top to bottom) flow height, distances of 98–484 km using the smaller 𝛾 value and 52–152 km using
concentration, velocity, and Richardson number. Each curve represents an
individual flow.
the larger 𝛾 value as we varied the suspension cloud concentration, grain
size, and channel height. Changes to all other input variables produced
adjustment lengths <91 km and <37 km for the smaller and larger 𝛾 values,
respectively. Sediment fluxes reached within 5% of the new equilibrium values at distances of 100–569 km
using the smaller 𝛾 value and 45–176 km using the larger 𝛾 value as we varied the suspension cloud concentra-
tion, grain size, slope, and channel height. Changes to all other input variables prompted adjustment lengths
<103 km and <45 km for the smaller and larger 𝛾 values, respectively. The Richardson numbers reached within
5% of the new equilibrium values at distances of 176–701 km using the smaller 𝛾 value and 85–224 km
using the larger 𝛾 value as we varied the suspension cloud concentration, grain size, and channel height.

Table 4
Flow Stripping: Maximum Adjustment Distance (km) From Base Case Equilibrium
Flow Property S hc cD∗ r Ds Cs
Height
5% 1.40 0.80 0.80 0.60 50.45 94.29
10% 1.20 0.80 0.60 0.40 45.05 81.68
Velocity
5% 19.22 33.43 21.42 19.82 26.43 48.65
10% 14.61 25.03 17.42 15.02 21.02 36.24
Sediment flux
5% 26.83 36.04 24.02 20.02 23.02 53.45
10% 20.82 27.63 18.62 15.42 17.62 40.84
Richardson number
5% 1.20 62.66 0.60 37.04 41.24 70.27
10% 1.00 54.25 0.40 31.83 35.84 57.86
Note. Distances show when the new flow reached within 10 and 5% of the new equilibrium
values resulting from changes in the model inputs.

TRAER ET AL. TURBIDITY CURRENT EQUILIBRIUM 525


21699011, 2018, 3, Downloaded from https://agupubs.onlinelibrary.wiley.com/doi/10.1002/2017JF004202 by Cochrane Mexico, Wiley Online Library on [25/10/2022]. 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 10.1002/2017JF004202

Table 5
Overspill, 𝛾 = 10−3 : Maximum Adjustment Distance (km) From Base Case Equilibrium
Flow Property S hc cD∗ Ds Cs w
Height
5% 5.41 8.61 5.81 121.52 294.29 3.20
10% 4.20 2.40 4.20 97.70 178.18 2.60
Velocity
5% 57.66 98.70 61.86 118.92 483.48 91.29
10% 38.44 75.48 42.84 95.10 357.36 69.67
Sediment flux
5% 148.95 100.70 103.50 103.50 568.57 92.49
10% 130.13 77.48 82.28 79.88 441.44 70.87
Richardson number
5% 3.80 176.98 3.40 182.18 700.70 169.17
10% 2.80 162.16 2.80 170.37 590.59 153.95
Note. Distances show when the new flow reached within 10 and 5% of the new equilibrium values
resulting from changes in the model inputs.

Changes to all other input variables prompted adjustment lengths <4 km and <2 km for the smaller and larger
𝛾 values, respectively.
3.2.2. Covarying Model Inputs
We repeated the above analysis but extracted the adjustment distance for all combinations of covarying
model inputs. In part 1 of this paper, we showed that the equilibrium conditions are most sensitive to changes
in the physical channel properties of slope, width, and channel height and are most sensitive to changes in
the flow properties of sediment grain size and suspension cloud concentration. Additionally, by varying the
model inputs individually, we have determined that the adjustment length is sensitive to these same param-
eters. We therefore present the adjustment length analysis for these pairs of model inputs and provide all
additional analysis in the supporting information. In our analysis, we chose to vary each model input using 50
equally spaced steps dividing the parameter space from the minimum and maximum values (Table 3). Due
to the division of the parameter space, we acknowledge a small level of numerical noise represented in the
contours of Figures 7–10 representing the adjustment distance but do not believe that the numerical noise
obscures our results to any meaningful degree.

Table 6
Overspill, 𝛾 = 10−1 : Maximum Adjustment Distance (km) From Base Case Equilibrium
Flow Property S hc cD∗ Ds Cs w
Height
5% 1.80 1.20 2.40 53.25 78.08 0.80
10% 1.20 1.00 1.40 42.24 44.04 0.60
Velocity
5% 28.63 52.05 32.43 52.45 151.15 36.24
10% 20.82 39.24 24.02 41.44 111.11 27.63
Sediment flux
5% 49.85 54.45 44.84 45.45 175.18 36.44
10% 39.44 41.64 35.04 34.83 133.13 27.83
Richardson number
5% 1.60 100.90 1.40 85.89 223.22 69.47
10% 1.20 87.69 1.20 75.28 183.18 60.86
Note. Distances show when the new flow reached within 10 and 5% of the new equilibrium
values resulting from changes in the model inputs.

TRAER ET AL. TURBIDITY CURRENT EQUILIBRIUM 526


21699011, 2018, 3, Downloaded from https://agupubs.onlinelibrary.wiley.com/doi/10.1002/2017JF004202 by Cochrane Mexico, Wiley Online Library on [25/10/2022]. 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 10.1002/2017JF004202

20 1.2 70
1 60

60
a 0.8 b c

40

30

50
0.6 50
20

40
0.4 40
50
0.2 40 30
0.4 0.2 30
20 20
0.4 0.2 10 1 10
1 1 10
10
5
20
Channel Height (m)

200
200
10
d 1 e f

20
10 00

30

100
0
50

5
0 100
50
30 10
10 0
75
75
50 50
5 25 25
25

75
50
25 50
5
20

110
100

90
g

75

90
1
h i 70
1

70
50
5
75 50
20 10
20
10
50
5 30
25
1 1 10

30
1 10
1 25
5
0.01 0.05 0.01 0.05 0.01 0.05
Slope
Figure 7. Adjustment lengths for covarying slope and channel height using flow stripping (a–c), overspill with 𝛾 = 10−3
(d–f ), and overspill with 𝛾 = 10−1 (g–i). Contours represent the adjustment distance (km) for flow height (a, d, and g),
velocity (b, e, and h), and sediment flux (c, f, and i). Dashed curves represent combinations of input channel and flow
parameters that did not change the equilibrium values.

In Figures 7–9, we plotted the adjustment distance required for flow height, velocity, and sediment flux to
reach within 5% of the new equilibrium values. In Figure 10, we plotted the adjustment distance required for
the flow Richardson number to reach within 5% of the new equilibrium value. We plotted the adjustment
distances in all figures as contours spanning the channel and flow parameter space. Adjustment lengths less
than 1 km resulted from two distinct conditions: (1) flows quickly adjusted to the new equilibrium conditions,
and (2) the new equilibrium conditions were within 5% of the base case equilibrium conditions, and therefore
the flows required no distance to effectively reequilibrate. To help distinguish between these two conditions,
we provided unmarked, dashed curves representing combinations of input channel and flow parameters that
did not change the equilibrium values. Closely spaced contours approximately parallel to these curves rep-
resent the extent of the parameter space where the new equilibrium conditions were ±5% the base case
equilibrium values.
When we covaried the geometrical properties of slope and channel height (Figure 7), we focused our analysis
on two regions of the parameter space: (1) regions of high slope and larger channel heights more consis-
tent with submarine canyon morphologies and (2) regions of low slope and lower channel heights more
consistent with leveed channels in deepwater basins. Broadly, adjustment distances were longer in regions
consistent with canyons than in regions consistent with basin channels, and adjustment distances were

TRAER ET AL. TURBIDITY CURRENT EQUILIBRIUM 527


21699011, 2018, 3, Downloaded from https://agupubs.onlinelibrary.wiley.com/doi/10.1002/2017JF004202 by Cochrane Mexico, Wiley Online Library on [25/10/2022]. 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 10.1002/2017JF004202

0.1

20
30 40
a b c
0.08

30
50
30
0.06

40

10

10
20
10

20
0.04

50

30
20

30
30
40 0

40
Suspension cloud concentration (xC)

5
0.02

0.1

100
100
50 100 50 d e f
0.08 20 10

50
20 10

100

50
0.06 10
50

10
0 0
0.04
10

50
20 20
0

0 0
0.02 20 50 50
500 0 0 0

0.1
25
50 g h i
50

50

50
0.08
25 10 5 1 5

20
2

20
0.06

0.04

50
50
25

10
10

10

0
50

0.02 75
0

20

20
100 0

0
0.01 0.02 0.03 0.01 0.02 0.03 0.01 0.02 0.03
Grain−size (cm)
Figure 8. Adjustment lengths for covarying sediment grain size and suspension cloud concentration using flow
stripping (a–c), overspill with 𝛾 = 10−3 (d–f ), and overspill with 𝛾 = 10−1 (g–i). Contours represent the adjustment
distance (km) for flow height (a, d, and g), velocity (b, e, and h), and sediment flux (c, f, and i). Dashed curves represent
combinations of input channel and flow parameters that did not change the equilibrium values.

comparatively shorter when using flow stripping and longer when simulating overspill using the smaller 𝛾
value. In the regions consistent with submarine canyons, flow heights generally adjusted in the shortest dis-
tances requiring approximately 1–30 km (Figures 7a, 7d, and 7g), with all adjustment distances greater than
5 km associated with overspill using the smaller 𝛾 value. This region corresponded to adjustment times rang-
ing from 1 to 30 min based on the integrated velocity profiles. Flow velocities in the canyon region required
adjustment distances of 35–175 km, with all adjustment distances greater than 60 km associated with over-
spill using the smaller 𝛾 value (Figures 7b, 7e, and 7h). These distances corresponded to adjustment times
ranging from 60 to 180 min. Sediment fluxes in the canyon region adjusted over similar distances to flow veloc-
ity, ≈20–100 km (Figures 7c, 7f, and 7i). The longest adjustment distances were found in narrow bands of the
parameter space surrounding the parameter values associated with adjustment lengths of 0, shown as closely
spaced contours parallel to the dashed curves in Figure 7. In the regions consistent with basin channels, flow
heights required adjustment lengths ranging from approximately 0 to 5 km, corresponding to adjustment
times of 1–15 min. In the basin channel region of the parameter space, the flow velocities and concentrations
required adjustment lengths approximately ranging from 1 to 50 km depending on whether we modeled flow
stripping or overspill. Using the integrated velocity profiles, these distances again correspond to adjustment
times on the order of tens of minutes.

TRAER ET AL. TURBIDITY CURRENT EQUILIBRIUM 528


21699011, 2018, 3, Downloaded from https://agupubs.onlinelibrary.wiley.com/doi/10.1002/2017JF004202 by Cochrane Mexico, Wiley Online Library on [25/10/2022]. 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 10.1002/2017JF004202

20

3.
a b c

5
10
0 10
0

3
2.5
90 90

2
Channel Height (m)
70 70 80
80
5
20
1.4
50 e
1.2 d 50 f
1
0.8 40 40

0.6 30
30

5
50 200 50 200 50 200
Width (m)
Figure 9. Adjustment lengths for covarying channel properties using flow overspill with 𝛾 = 10−3 (a–c), and 𝛾 = 10−1
(d–f ). Contours represent the adjustment distance (km) for flow height (a and d), velocity (b and e), and sediment flux
(c and f ). Dashed curves represent combinations of input channel parameters that did not change the equilibrium
values.

When we covaried flow properties (grain size and suspension cloud concentration), the adjustment lengths
across much of the parameter space were broadly similar to those resulting from changes in channel prop-
erties (Figure 8). Yet adjustment lengths tended to be significantly larger (≈2–2.5 times) for flows simulated
with lower grain sizes and suspension cloud concentrations. Similar to the covariance analysis with channel
properties, the adjustment lengths associated with changing flow properties were largest when we modeled
currents with overspill and the smaller 𝛾 value and were smallest when we modeled currents with flow strip-
ping. Flow height adjustment lengths exhibited the most complicated pattern across the parameter space,
with values generally ranging from tens to hundreds of kilometers (Figures 8a, 8d, and 8g). The adjustment
lengths broadly increased as the grain size decreased, with more complicated patterns of sharp increases and
decreases in adjustment distance appearing in narrow bands of the parameter space surrounding parameter
values associated with adjustment lengths of 0. Adjustment lengths for flow velocities (Figures 8b, 8e, and 8h)
and sediment fluxes (Figures 8c, 8f, and 8i) exhibited similar patterns across the parameter space, with adjust-
ment lengths increasing as both grain size and suspension cloud concentration decreased. Simulated flows
of fine sand to very fine sand more consistent with basin channels broadly required adjustment distances of
25–150 km when suspension cloud concentration remained high. Based on the integrated velocity profiles,
we determined that adjustment lengths greater than 50 km, which comprise much of the parameter space
associated with overspill, approximately correspond to flow durations in excess of 1 h.
When we covaried the channel properties (channel height and width), the adjustment lengths were gener-
ally considerably smaller than those resulting from changes in the geometrical and flow properties (Figure 9).
The primary effect of changing the channel width is to constrict or expand the flow vertically as it adjusts
to its new container, and the primary mass advection processes acting upon a taller or shorter current is
flow overspill. Thus, we display the overspill results in Figure 9 and provide the flow stripping analysis in the
supporting information. We focused our analysis on two regions of the parameter space: (1) channels with
lower widths and higher heights, more consistent with submarine canyon morphologies, and (2) channels
with higher widths and lower channel heights, more consistent with submarine basin morphologies.
Adjustment lengths broadly increased with increasing channel heights and widths, regardless of the 𝛾 value
used. Flow heights generally required adjustment lengths of 1–2.5 km in the canyon regions and 0.75–2.7 km

TRAER ET AL. TURBIDITY CURRENT EQUILIBRIUM 529


21699011, 2018, 3, Downloaded from https://agupubs.onlinelibrary.wiley.com/doi/10.1002/2017JF004202 by Cochrane Mexico, Wiley Online Library on [25/10/2022]. 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 10.1002/2017JF004202

Bed Friction Coefficient


0.02
a b c

0.9
0.016

0.7
0.012 5
0. 3
3 2
0.008 0. 1

0.004
0.02 0.05 0.02 0.05 0.02 0.05
Slope
Figure 10. Adjustment lengths for covarying slope and bed friction coefficient using (a) flow stripping, (b) overspill with
𝛾 = 10−3 , and (c) overspill with 𝛾 = 10−1 . Contours represent the adjustment distance (km) for the flow Richardson
number. Gray areas represent parameter space where flows are arrested before they could reequilibrate.

in the basin regions depending on the 𝛾 value (Figures 9a and 9d). Flow velocities generally required adjust-
ment lengths of 90 km in canyon regions and 85 km in the basin regions when using the smaller 𝛾 value,
distances that represent significant portions of some submarine systems (Figures 9b and 9e). When using the
larger 𝛾 value, the velocities required more moderate adjustment lengths of 45 km in the canyon regions and
30 km in the basin regions. Sediment flux showed nearly identical behaviors across the parameter space as the
flow velocity for the respective 𝛾 values (Figures 9c and 9f ). Using the smaller 𝛾 value, sediment flux required
adjustment lengths of approximately 90 km in the canyon regions and 85 km in the basin regions, while when
using the larger 𝛾 value, sediment flux required adjustment lengths of approximately 45 km in the canyon
regions and 30 km in the basin regions. This consistency seems to suggest that flow velocity exerts a primary
control over the sediment flux when changing the channel dimensions.
Equilibrium flow Richardson number was most sensitive to changes in channel slope and the coefficient of bed
friction (part 1 of this work). As a result, we complete our analysis by covarying these parameters (Figure 10).
In all modeled cases, the flow Richardson number adjusted to the new equilibrium values within 5 km and
usually adjusted within 1 km. As channel slope decreased and the coefficient of bed friction increased, the
adjustment lengths also increased. These flows still entrained sediment from the bed to maintain density
contrast, even as they approached the Richardson supercritical to subcritical transition. However, for rapid
drops in channel slope in concert with rapid increases in bed friction, the simulated turbidity current subsided
before they could reequilibrate.

4. Discussion
This study shows that turbidity currents that start from a variety of starting conditions may evolve to an equi-
librium height, velocity, and sediment flux that is consistent with physical channel and flow properties. While
we showed in the companion paper that flows could possibly evolve to equilibrium conditions, this simula-
tion of flows with random initial conditions suggest that only one set of equilibrium conditions exist for a
given set of channel and flow properties and that the majority of realistic, randomly generated flows evolve
to this equilibrium state. This finding supports the work of Hu et al. (2015), who derived unique asymptotic
analytic solutions to the three-equation model for steady turbidity currents for given channel and flow prop-
erties. Many of our simulated equilibrium flows were somewhat analogous to the igniting flows of Parker
et al. (1986), in that they eroded sediment to maintain excess density. Yet the removal of fluid and accompa-
nying sediment mass from the suspension cloud prevented the flows from self-accelerating. These types of
flows are generally large and likely form a subset of flows that traverse natural channels. In our simulations,
many flows did not reach equilibrium, primarily for two reasons: (1) they lacked the appropriate excess den-
sity to ignite and grow beyond the channel height, thereby neglecting the effects of stripping and overspill
that make equilibrium possible, and (2) sudden mass removal at channel meanders led to rapid deceleration
and flow collapse. The conditions under which flows reach these respective states can be broadly summa-
rized as (1) flows with low initial velocities and/or sediment fluxes, generally but not exclusively less than 0.5
and 0.05 times their respective base case values, and (2) flows with large initial flow heights and/or sediment

TRAER ET AL. TURBIDITY CURRENT EQUILIBRIUM 530


21699011, 2018, 3, Downloaded from https://agupubs.onlinelibrary.wiley.com/doi/10.1002/2017JF004202 by Cochrane Mexico, Wiley Online Library on [25/10/2022]. 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 10.1002/2017JF004202

fluxes, generally but not exclusively greater than 3 and 10 times their respective base case values. Equilibrium
attained by mass advection might thus provide one of several mechanisms for natural flows to traverse all or
part of submarine systems.
As the local channel and flow properties change, the flows adjust to the new equilibrium conditions, often
taking tens of kilometers to reach within 5% of the new equilibrium values. Even relatively short adjustment
lengths of 10–20 km represent significant portions of some submarine systems such as the Var canyon and
channel (Savoye et al., 1993) and the La Jolla channel (Covault et al., 2011), suggesting that flow equilibrium
might only exist for turbidity currents traversing larger systems where changes in channel morphology are
more likely to occur over longer distances (e.g., Babboneau et al., 2002; Bouma et al., 1983; Pirmez & Imran,
2003). Given the relative consistency of more distal submarine channel morphologies, our results suggest that
the adjustment distances are more likely to arise due to changes in the flow properties (e.g., grain size and
suspension cloud concentration) than changes in the geometrical or channel properties (e.g., slope, channel
height and width). Furthermore, the adjustment of turbidity current equilibrium to more gradual changes in
channel properties found in distal submarine systems provides a mechanism for sustaining flows for great
distances despite low slopes.
Several recent studies have suggested that flow stripping at channel meanders acts as a filter that effectively
normalizes turbidity current flow dynamics beyond the first several meanders (e.g., McHargue et al., 2011;
Straub et al., 2008). Our analyses of simulated turbidity currents that include the flow stripping process affirms
this view. Specifically, we observed a general convergence of flow dynamics toward equilibrium conditions
in infinitely spiraling, equal-spaced, and random channels, despite disparate, randomized starting conditions
of individual flows. This approach to a common equilibrium also arose with simulations of turbidity currents
using overspill effects; however, flows in which overspill was the dominant mode of fluid mass ejection often
took between 50 and 100 km to reach equilibrium values. In contrast, flows simulated with random starting
conditions and stripping effects often took between 20 and 30 km to converge to their equilibrium values.
This further affirms the view that stripping likely plays an important role in flow filtering, provided that the
turbidity currents traverse meandering channels.
The sensitivity of the four-equation model to the initial flow conditions has been well established (e.g., Pantin
& Franklin, 2009; Parker et al., 1986). Yet accounting for processes that remove fluid mass from the flow allowed
simulated turbidity currents to reach constant equilibrium values regardless of the initial conditions, suggest-
ing that the new model is less sensitive to the initial conditions than models that do not account for these
processes. Additionally, Traer et al. (2012) noted that simulated turbidity currents were highly sensitive to the
clear-water entrainment process which made predictions of flow behavior beyond several hundred meters
difficult. By accounting for the mass advection processes, our new model limits this uncertainty and suggests
that flow prediction under equilibrium conditions becomes significantly easier.
Furthermore, by comparing the adjustment distance to the length of submarine channels, our adjustment
length analysis helps elucidate conditions under which flow evolution reflects changes to local channel and
flow properties as opposed to upstream flow conditions. To illustrate this, we use the following thought exper-
iment: Temporarily considering adjustment lengths not associated with changes in grain size or suspension
cloud concentration, we found that flows generally reequilibrated within approximately 50 km. The Carlsbad
channel, offshore California, is approximately 25 km long from canyon to fan (Covault et al., 2011), and there-
fore any flow traversing the system would not be able to adjust to changing equilibrium conditions despite
changes in the channel properties. Our results suggest that modeled flow evolution in the Carlsbad system
would more strongly reflect the initial, or inlet, flow conditions. Considering a larger system, the Amazon chan-
nel is approximately 800 km long from canyon to fan (Pirmez & Imran, 2003), and therefore flows traversing
the system could adjust to new equilibrium conditions approximately 16 times over the channel’s length. Our
results suggest that modeled flows in the Amazon channel would therefore more strongly reflect changes in
the local channel and flow properties represented by the changing equilibrium conditions.
Changes in the grain size and suspension cloud concentration produced the largest adjustment distances.
Adjustment distances associated with changes in grain size ranged from 20 to 180 km, while adjustment
distances associated with changes in the suspension cloud concentration ranged from 40 to 700 km. Rapid
changes in grain size suggest rapid changes in the flow’s capacity to turbulently suspend grains of a cer-
tain size. Increased flow turbulence would allow the current to suspend coarser material while decreased
flow turbulence would likely allow coarser grains to settle out of suspension. Given that numerous studies

TRAER ET AL. TURBIDITY CURRENT EQUILIBRIUM 531


21699011, 2018, 3, Downloaded from https://agupubs.onlinelibrary.wiley.com/doi/10.1002/2017JF004202 by Cochrane Mexico, Wiley Online Library on [25/10/2022]. 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 10.1002/2017JF004202

of turbidite outcrops have interpreted rapid deposition rates (e.g., Jobe et al., 2012; Mutti & Lucchi, 1978;
Sinclair & Tomasso, 2002), large changes in grain size and the resulting long adjustment distances cannot be
ruled out. However, rapid deposition likely corresponds with nonuniform flows where decreasing turbulence
intensity over short distances results in abrupt loss of capacity (Hiscott, 1994; Kneller, 1995), and such flows
are not likely to maintain equilibrium conditions.
Rapid decreases in the suspension cloud concentration suggest dilution resulting from increased clear-water
entrainment, while rapid increases in the suspension cloud concentration suggest an increase in the sediment
carrying capacity of the suspension cloud. Increased carrying capacity would likely arise due to increased
turbulence (e.g., Garcia & Parker, 1991, 1993), which would likely lead to higher rates of clear-water entrain-
ment (e.g., Parker et al., 1987). It is unclear how the joint effects of these two processes might affect capacity.
However, flow stratification models that predict upward decreasing concentration profiles are supported
by measurements from natural turbidity currents (e.g., Normark, 1987) and field observations of decreasing
grain size with increasing levee height (e.g., Hess & Normark, 1976; Klaucke et al., 1997). To our knowledge,
rapid changes in the concentration profiles of long-lived turbidity currents have not been documented at
the laboratory or field scale, suggesting that rapid changes in the suspension cloud concentration might
be uncommon. Despite the long adjustment distances associated with changing grain size and suspension
cloud concentration, our analyses suggest that modeled flows are less sensitive to upstream flow conditions
in longer channels that may allow turbidity currents to evolve from one equilibrium state to another.
Given the changes in equilibrium conditions and the different adjustment lengths for flow height, velocity,
and sediment flux, the flow Richardson number often required longer adjustment lengths. We were unable
to identify changes in the channel and flow properties that allowed the simulated turbidity currents to
evolve from a supercritical to subcritical state. We found in our companion paper that subcritical equilibrium
flows were possible on low slopes with higher coefficients of bed friction. Yet when we covaried the slope
and bed friction, flows arrested before reequilibrating in this parameter space. The simulations that caused
flows to arrest represented slope changes that were equivalent to an instantaneous change from submarine
canyon slope values to basin floor values. Using laboratory experiments and numerical models, Garcia and
Parker (1989) and Kostic and Parker (2006) examined turbidity currents traversing similar slope transitions and
found that rapid changes in slope often led to a change in the flow regime, which created hydraulic jumps
within the turbidity current. Such flows are not consistent with equilibrium conditions. Therefore, more subtle
changes in the channel and flow properties might be more favorable for supercritical to subcritical equilibrium
transitions.

5. Conclusions
Modeled turbidity currents that successfully traverse submarine systems appear to reach an equilibrium
between entrained fluid mass and fluid mass ejected from the flow by flow stripping or overspill, regardless of
the initial conditions. This equilibrium was approached over shorter distances when flow stripping was more
important than overspill in removing fluid and sediment mass from the current. This suggests that flow strip-
ping may play a more dominant role in regulating turbidity current heights, velocities, and concentrations.
Additionally, these simulations suggest that flow stripping might act as a flow filter that normalizes the flow
conditions beyond the first several channel meanders. When we systematically changed the channel and flow
properties and calculated the distance required for the turbidity currents to reequilibrate, we found that the
adjustment lengths generally ranged from 1 to 40 km. Changes in grain size and suspension cloud concen-
tration resulted in longer adjustment distances exceeding 100 km. Longer adjustment lengths indicated that
the initial conditions still exerted a strong influence over the flow dynamics, while shorter adjustment lengths
indicated that the flow dynamics were more sensitive to local channel and flow conditions. Our results sug-
gest that turbidity currents traversing shorter systems (e.g., La Jolla, Carlsbad) might fail to reequilibrate to the
new channel and flow properties, while turbidity currents traversing longer systems (e.g., Amazon, Mississippi,
Indus) might evolve through several progressive equilibrium states.

Notation
C depth-averaged flow concentration
h flow height (L)
hc channel depth (L)

TRAER ET AL. TURBIDITY CURRENT EQUILIBRIUM 532


21699011, 2018, 3, Downloaded from https://agupubs.onlinelibrary.wiley.com/doi/10.1002/2017JF004202 by Cochrane Mexico, Wiley Online Library on [25/10/2022]. 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 10.1002/2017JF004202

Riflow Richardson number (= RgCh∕U2 )


Schannel slope
U flow velocity (depth-averaged) (LT −1 )
w channel width (L)
𝛾 nonphysical flow overspill scaling parameter (TL−3 ); see part 1 of this paper for expanded definition of 𝛾
and nondimensionalization
𝜓 sediment flux (L3 T −1 )

Acknowledgments References
The authors wish to thank three
anonymous reviewers for their Babboneau, N., Savoye, B., Cremer, M., & Klein, B. (2002). Morphology and architecture of the present canyon and channel system of the
thoughtful comments on an earlier Zaire deep-sea fan. Marine and Petroleum Geology, 19, 445–467.
version of this manuscript that Bouma, A. H., Stelting, C. W., & Coleman, J. M. (1983). Mississippi Fan: Internal structure and depositional processes. Geo-Marine Letters, 3,
helped add focus and clarity to this 147–153.
work. All data used are available in Cartigny, M. J., Postma, G., van den Berg, J. H., & Mastbergen, D. R. (2011). A comparative study of sediment waves and cyclic steps based on
the tables. geometries, internal structures and numerical modeling. Marine Geology, 280, 40–56.
Covault, J. A., Fildani, A., Romans, B. W., & McHargue, T. (2011). The natural range of submarine canyon-and-channel longitudinal profiles.
Geosphere, 7(2), 313–332.
Das, H. S., Imran, J., Pirmez, C., & Mohrig, D. (2004). Numerical modeling of flow and bed evolution in meandering submarine channels.
Journal of Geophysical Research, 109, C10009. https://doi.org/10.1029/2002JC001518
Garcia, M., & Parker, G. (1991). Entrainment of bed sediments into suspension. Journal of Hydraulic Engineering, 117, 414–435.
Garcia, M., & Parker, G. (1989). Experiments on hydraulic jumps in turbidity currents near a canyon-fan transition. Science, 245, 393–396.
Garcia, M., & Parker, G. (1993). Experiment on the entrainment of sediment into suspension by a dense bottom current. Journal of
Geophysical Research, 98, 4793–4807.
Hess, G. R., & Normark, W. R. (1976). Holocene sedimentation history of the major fan valleys of Monterey Fan. Marine Geology, 22, 233–251.
Hiscott, R. N. (1994). Loss of capacity, not competence, as the fundamental process governing deposition from turbidity currents. Journal of
Sedimentary Research, 64, 209–214.
Hiscott, R. N., Hall, F. R., & Pirmez, C. (1997). Turbidity-current overspill from the Amazon channel: Texture of the silt/sand load, paleoflow
from anisotropy of magnetic susceptibility and implications for flow processes. In Proceedings of the Ocean Drilling Program: Collage
Station (Vol. 155, pp. 53–78). TX: Texas A and M University.
Hu, P., Pähtz, T., & He, Z. (2015). Is it appropriate to model turbidity currents with the three-equation model?. Journal of Geophysical Research:
Earth Surface, 120, 1153–1170. https://doi.org/10.1002/2015JF003474
Jobe, Z. R., Lowe, D. R., & Morris, W. R. (2012). Climbing-ripple successions in turbidite systems: Depositional environments, sedimentation
rates and accumulation times. Sedimentology, 59, 867–898.
Klaucke, I., Hesse, R., & Ryan, W. (1997). Flow parameters of turbidity currents in a low-sinuosity giant deep-sea channel. Sedimentology, 44,
1093–1102.
Kneller, B. (1995). Beyond the turbidite paradigm: Physical models for deposition of turbidites and their implications for reservoir
prediction. Geological Society of London Special Publication, 94, 31–49.
Kostic, S. (2011). Modeling of submarine cyclic steps: Controls on their formation, migration, and architecture. Geosphere, 7(2), 294–304.
Kostic, S., & Parker, G. (2006). The response of turbidity currents to a canyon-fan transition: Internal hydraulic jumps and depositional
signatures. Journal of Hydraulic Research, 44(5), 631–653.
Lamb, M. P., McElroy, B., Kopriva, B., Shaw, J., & Mohrig, D. (2010). Linking river-flood dynamics to hyperpycnal-plume deposits: Experiments,
theory, and geological implications. GSA Bulletin, 122(9/10), 1389–1400.
MATLAB (2011). version 7.12.0 (R2011a). Natick, Massachusetts: The Mathworks Inc.
McHargue, T., Pyrcz, M. J., Sullivan, M. D., Clark, J. D., Fildani, A., Romans, B. W., … Drinkwater, N. J. (2011). Architecture or turbidite channel
systems on the continental slope: Patterns and predictions. Marine and Petroleum Geology, 28, 728–743.
Mohrig, D., Whipple, K. X., Hondzo, M., Ellis, C., & Parker, G. (1998). Hydroplaning of subaqueous debris flows. GSA Bulletin, 110(3), 387–294.
Mutti, E., & Lucchi, F. R. (1978). Turbidites of the northern Apennines: Introduction to facies analysis. International Geology Review, 20(2),
125–166.
Normark, W. R. (1989). Observed parameters for turbidity current flow in channels, Reserve Fan, Lake Superior. Journal of Sedimentary
Petrology, 59, 423–431.
Pantin, H. M., & Franklin, M. C. (2009). Predicting autosuspension in steady turbidity flow: Ignition points and their relation to Richardson
numbers. Journal of Sedimentary Research, 79, 862–871.
Parker, G., Fukushima, Y., & Pantin, H. M. (1986). Self-accelerating turbidity currents. Journal of Fluid Mechanics, 171, 145–181.
Parker, G., Garcia, M., Fukushima, Y., & Yu, W. (1987). Experiments on turbidity currents over an erodible bed. Journal of Hydraulic Research,
25, 123–147.
Pirmez, C., & Imran, J. (2003). Reconstruction of turbidity currents in Amazon channel. Marine and Petroleum Geology, 20, 823–849.
Savoye, B., Piper, D. J. W., & Droz, L. (1993). Plio-Pleistocene evolution of the Var deep-sea fan off the French Riviera. Marine and Petroleum
Geology, 10, 550–571.
Sinclair, H. D., & Tomasso, M. (2002). Depositional evolution of confined turbidite basins. Journal of Sedimentary Research, 72(4), 451–456.
Skene, K. I., Piper, D. J. W., & Hill, P. S. (2002). Quantitative analysis of variations in depositional sequence thickness from submarine channel
levees. Sedimentology, 49, 1411–1430.
Straub, K. M., Mohrig, D., McElroy, B., Buttles, J., & Pirmez, C. (2008). Interactions between turbidity currents and topography in aggrading
sinuous submarine channels: A laboratory study. Geological Society of America Bulletin, 120, 368–385.
Sun, T., & Parker, G. (2005). Transportational cyclic steps created by flow over an erodible bed. Part 2. Theory and numerical simulation.
Journal of Hydraulic Research, 43(5), 502–514.

TRAER ET AL. TURBIDITY CURRENT EQUILIBRIUM 533


21699011, 2018, 3, Downloaded from https://agupubs.onlinelibrary.wiley.com/doi/10.1002/2017JF004202 by Cochrane Mexico, Wiley Online Library on [25/10/2022]. 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 10.1002/2017JF004202

Talling, P. J., Amy, L. A., Wynn, R. B., Blackbourn, G., & Gibson, O. (2007). Evolution of turbidity currents deduced from extensive
thin turbidites: Marnoso Arenacea Formation (Miocene), Italian Apennines. Journal of Sedimentary Research, 77, 172–196.
https://doi.org/10.2110/jsr.2007.018
Traer, M. M., Fildani, A., Fringer, O., McHargue, T., & Hilley, G. E. (2018). Turbidity current dynamics: Part 1. Model formulation and
identification of flow equilibrium conditions resulting from flow stripping and overspill. Journal of Geophysical Research: Earth Surface,
123. https://doi.org/10.1002/2017JF004200
Traer, M. M., Fildani, A., McHargue, T., & Hilley, G. E. (2015). Simulating depth-averaged, one-dimensional turbidity current dynamics
using natural topographies. Journal of Geophysical Research: Earth Surface, 120, 1485–1500. https://doi.org/10.1002/
2015JF003638
Traer, M. M., Hilley, G. E., Fildani, A., & McHargue, T. (2012). The sensitivity of turbidity currents to mass and momentum exchanges between
these underflows and their surroundings. Journal of Geophysical Research, 117, F01009. https://doi.org/10.1029/2011JF001990

TRAER ET AL. TURBIDITY CURRENT EQUILIBRIUM 534

You might also like