J
OURNAL OF
S
EDIMENTARY
R
ESEARCH
,V
OL
. 74, N
O
.1,J
ANUARY
, 2004,
P
. 148–155
Copyright
q
2004, SEPM (Society for Sedimentary Geology) 1527-1404/04/074-148/$03.00
SURGING VERSUS CONTINUOUS TURBIDITY CURRENTS: FLOW DYNAMICS AND DEPOSITS IN AN
EXPERIMENTAL INTRASLOPE MINIBASIN
MICHAEL P. LAMB,
1
* THOMAS HICKSON,
2
JEFFREY G. MARR,
1
BEN SHEETS,
1
CHRIS PAOLA,
1
AND
GARY PARKER
1
1
St. Anthony Falls Laboratory, University of Minnesota, Minneapolis, Minnesota 55414, U.S.A.
email: mike@ocean.washington.edu
2
Department of Geology, University of St. Thomas, 2115 Summit Avenue, OWS 153, St. Paul, Minnesota 55105, U.S.A.
A
BSTRACT
: Small intraslope basins (
;
100 km
2
), or ‘‘minibasins,’’
such as those found on the continental slope of the Gulf of Mexico,
have been filled predominantly by turbidity currents. Each minibasin
is the result of local subsidence and is partially or completely isolated
from neighboring basins by ridges formed from compensational uplift.
We undertook a series of experiments to investigate the relationship
between the flow dynamics of turbidity currents entering a minibasin
and the stratal architecture of their deposits. The experiments were
performed using continuous-feed turbidity currents and surge-feed
turbidity currents. A dimensionless ponding number is developed to
compare the geometry of the deposits with the dynamics of the flows
that filled the basins. The experimental surging turbidity currents cre-
ated a deposit that was notably more ponded than the deposits of con-
tinuous turbidity currents.
INTRODUCTION
Intraslope salt-withdrawal basins, or ‘‘minibasins,’’ are an important fea-
ture of the northwestern continental slope of the Gulf of Mexico (Pratson
and Ryan 1994). Similar basins have been encountered in both modern and
ancient sedimentary successions throughout the world (e.g., Trinidad and
Tobago, Brami et al. 2000; Angola, Schollnberger and Vail 1999). Mini-
basins are formed from diapirism due to the buoyant instability of a mobile
substrate (e.g., salt body) overlain by a load of denser sediment. These
basins typically have several hundred meters of relief and span areas on
the order of 10
1
–10
2
square kilometers. An example of a seismic section
from such a basin is shown in Figure 1. Depositional turbidity currents
appear to have completely or partially filled many of the minibasins in the
Gulf of Mexico. Many, but not all, of these minibasins are connected by
submarine channels that form a drainage network that eventually discharges
onto the abyssal plain of the Gulf of Mexico (Liu and Bryant 2000).
Turbidity currents are believed to have filled many minibasins and erod-
ed channels into neighboring ridges through a process known as ‘‘fill and
spill’’ (e.g., Winker 1996; Beauboeuf and Friedman 2000), creating sand-
rich deposits within minibasins that constitute prime targets for oil explo-
ration. We undertook a series of experiments to investigate fill-and-spill
processes by analyzing the relationship between the dynamics of the tur-
bidity currents entering a minibasin and the stratal architecture of their
associated deposits. We focused on two possibilities. Firstly, we considered
large, continuous turbidity currents that traverse multiple neighboring ba-
sins, ‘‘filling and spilling’’ from basin to basin across several ridges, or
through partially excavated canyons between ridges, in a single flow event.
In this case, ridge incision and basin filling would occur during the same
time interval for several successive basins and ridges. Secondly, we ex-
amined small, pulse-like flows that fill one basin at a time. These flows are
unable to travel over the downstream lip of the basin until the basin is
substantially filled with sediment. Once the basin is filled, successive tur-
bidity current pulses may downcut the ridge and begin filling the next
minibasin downstream.
* Present Address: School of Oceanography, University of Washington, Seattle,
Washington 98195, U.S.A.
EXPERIMENTAL APPROACH
The experiments reported here represent an extension of an initial effort
by Hickson et al. (2000). We used a flume at St. Anthony Falls Laboratory,
University of Minnesota, specifically designed for the study of both surging
and continuous turbidity currents. The tank is 0.31 m wide, 10.6 m long
and 0.76 m deep. A model minibasin was installed in the flume (Fig. 2).
The installed minibasin floor had a length of 4.1 m and a maximum relief
of 0.15 m. The dimensions of the experimental minibasin were constrained
by the average dimensionless length-to-depth profiles and the average ratios
of basin area to basin capacity (
ø
9) of thirteen minibasins from the Gulf
of Mexico, as determined from full-resolution multibeam data from the
National Geophysical Data Center (Divins 2001). Here, basin capacity is
defined as the maximum volume of fluid a basin can hold before spilling
over its downstream lip.
The flume was equipped with a mixing tank and a damping tank (Fig.
2). The mixing tank was filled with a slurry of sediment and water, which
was continuously circulated up to a constant head tank and back down to
the mixing tank. During an experiment a portion of the slurry was diverted
from the head tank into the flume (Fig. 3A). The slurry entered the mini-
basin flume through a headgate that restricted the flow thickness to 3 cm.
Whenever the turbidity current overflowed the minibasin, it emptied into
the damping tank, which was drained from the bottom. The water level in
the damping tank and flume was held constant by a fresh water supply at
the top of the damping tank. In addition to keeping the ambient water in
the flume above the turbidity current free of sediment, the damping tank
served to entirely suppress reflection waves, which would have resulted
had the turbidity current collided with a closed barrier at the downstream
end of the flume. It should be pointed out, however, that turbidity currents
could and did fully or partially reflect from the downstream lip of the
minibasin itself, a feature that was intentionally designed into the experi-
ments.
We performed four continuous-feed turbidity current experiments (c1–
c4) and one surge-feed turbidity current experiment consisting of twelve
successive surges (s1–s12) and analyzed their resultant deposits. The input
flow rates and the slurry compositions are summarized in Table 1. All of
the turbidity currents had an initial sediment concentration of 5% by vol-
ume, with the remainder being tap water. The sediment used in all exper-
iments was composed of 50% kaolinite clay and 50% silica silt. At the
small scale of the experiments reported here, silt tended to deposit out
easily, with little or no resuspension. The suspended clay acted as an extra
‘‘driver’’ for the turbidity current, maintaining its density excess above
clear water even as the silt settled out (Salaheldin et al. 2000).
For the first three continuous-feed experiments (c1, c2, c3) the sediment
consisted of 50% kaolinite clay and 50% poorly sorted 20
m
m silt (silica
flour). The sediment used in the fourth experiment (c4) consisted of 50%
kaolinite clay and 50% poorly sorted 45
m
m silt (silica flour). Input flow
rates and concentrations were held constant throughout the runs, which
ranged from 14 minutes to 55 minutes in duration. The bed deposit was
profiled with a point gauge at the end of each run.
The surge-feed experiment consisted of twelve pulsed turbidity currents
(s1–s12), the deposits of which were allowed to stack on top of one another.
The sediment mix was 50% kaolinite clay and 50% poorly sorted 20
m
m
silt, which is the same mix used in the first three continuous-turbidity-
149SURGING AND CONTINUOUS TURBIDITY CURRENTS IN AN EXPERIMENTAL INTRASLOPE BASIN
F
IG
. 1.—High-resolution seismic dip profile of a salt-withdrawal minibasin on the northern continental slope of the Gulf of Mexico, showing differentiation of major
seismic facies. Diagram courtesy of Winker (1996).
F
IG
. 2.—Schematic of the experimental flume.
The slurry was kept in suspension in the mixing
tank and pumped to a constant-head tank in
order to maintain a steady sediment and slurry
discharge in the flume. Any slurry that reached
the end of the flume was vented out of the
system from a bottom drain in the damping tank.
The damping tank also suppressed any
reflectional waves from the downstream end of
the flume.
current experiments. For each surge, the input flow rate was held constant
at 1 l/s for a duration of 15 s. The sizes of the surges were designed such
that they resulted in little to no basin spillover. This corresponded to a
surge volume that was 15 liters, or approximately 10% of the basin capac-
ity. The capacity of the experimental minibasin was 141 liters with no
sediment in the basin. The time between surges was on the order of 8
hours, a value chosen to allow for nearly complete settling of sediment. At
the end of each surge, ultrasonic sonar was used to record the elevation of
the sediment bed.
RESULTS
A dimensionless deposit ponding index number (Po) is used to describe
the experimental deposits:
150 M.P. LAMB ET AL.
F
IG
. 3.—Photographs showing A) the head of a continuous-feed turbidity current entering the basin and B) the same flow after setup of a quasi-steady dammed turbidity
current. The flow was from left to right. The internal hydraulic jump is not shown in the image.
T
ABLE
1.—Input parameters (sediment concentration, discharge, velocity, and den-
simetric Froude Number) as well as the deposit ponding index and trapping effi-
ciency of the deposits from the four continuous flows (c1–c4) and the cumulative
surge deposit (s1–s12).
Experiment c1 c2 c3 c4 s1–s12
Input flow rate
(liters/second)
0.5 0.85 2 1 1
Sediment concentration
by volume
0.05 0.05 0.05 0.05 0.05
Kaolinite concentration
by volume
0.025 0.025 0.025 0.025 0.025
Silt concentration
by volume
0.025 0.025 0.025 0.025 0.025
Input velocity (cm/s) 5.4 9.1 21.5 10.8 10.8
Approximate input Fr
d
0.35 0.59 1.38 0.69 0.69
Mean silt grain size
(microns)
20 20 20 45 20
Duration (seconds) 3300 1860 840 1680 15 per surge
Ponding index (Po)
Trapping efficiency
0.37
0.15
0.16
0.17
0.22
0.14
0.11
0.32
0.54
0.99
L
1 d
u
Po
52
dx (1)
E
Ld
h
0
where x is the downdip coordinate, L denotes the streamwise length of the
basin,
u
(x) denotes the thickness of the deposit and
h
(x) denotes the ele-
vation of the initial bed. A deposit ponding index equal to one represents
a completely ponded deposit (Fig. 4A), a deposit ponding index greater
than unity represents a mounded deposit (Fig. 4B), and a deposit ponding
index equal to zero represents a purely draped deposit (Fig. 4C). A deposit
with accentuated highs, meaning that the flow deposits preferentially on
the slopes rather than the center of the basin, would have a negative deposit
ponding index (Fig. 4D).
Continuous Turbidity Currents
The flows for all four continuous-turbidity-current experiments were
strongly influenced by the minibasin topography. The head of a continuous
turbidity current traversed the basin and collided with the minibasin lip,
such that part of the head overflowed the basin and the rest formed a
backward-migrating bore (Fig. 5). The portion of the head that overflowed
the basin resembled a surge-feed turbidity current filling the next down-
stream basin, even though its source was from a continuous flow. As shown
in Figure 5, the bore eventually either stabilized at a point toward the
upstream end of the basin, forming a well-defined internal hydraulic jump
(experiment c3), or drowned out the entrance headgate (experiments c1,
c2, and c4). The position of the bore and the basin overflow stabilized
within about 60–80 seconds, after which setup of the continuous flow was
complete and the flow changed only in response to the changing bed to-
pography as sediment was deposited.
Garcı´a and Parker (1989) and Garcı´a (1993) have emphasized the role
of internal hydraulic jumps in regard to turbidity current dynamics. A hy-
draulic jump is a shock by which a shallow, swift flow, i.e., a supercritical
flow in terms of bulk densimetric Froude number, is converted to a deep,
slow, subcritical flow. The bulk densimetric Froude number (Fr
d
) can be
defined as follows:
U
Fr
5
(2)
d
Ï
RgCH
where H denotes turbidity current layer thickness, U denotes layer-averaged
flow velocity, C denotes layer-averaged volume sediment concentration in
the turbidity current, g denotes the acceleration of gravity, and R denotes
the submerged specific gravity of the sediment, which was about 1.65 for
the sediment used in the present study. Fr
d
is greater than unity for a
supercritical flow, and Fr
d
is less than unity for a subcritical flow. At an
internal hydraulic jump, the bulk densimetric Froude number undergoes a
rapid change from a value in excess of unity to a value below unity over
a short distance.
Values of Fr
d
at the headgate for all experiments are reported in Table
1. In experiment c3, the turbidity current was supercritical as it entered the
flume. The downstream lip of the minibasin acted to dam the flow, forcing
a transition to subcritical flow via an internal hydraulic jump, so that the
flow upstream of the hydraulic jump was supercritical and the flow down-
stream of the jump but within the basin was highly subcritical (Fig. 5).
This would likely be the case for minibasins in nature. This is because
turbidity currents, and indeed all dense bottom currents, can be expected
to attain a densimetric Froude number of unity because of hydraulic control
as they pass over a bed-elevation maximum and drain into a deep basin
(e.g., Turner 1973). Thus, natural turbidity currents can be expected to
make a transition from subcritical to supercritical flow as they pass over a
minibasin lip. In addition, if a continuous supercritical flow enters a con-
fined basin, the confinement sets up a hydraulic jump. In the cases of
experiments c1, c2, and c4 the incoming flow was subcritical, so that the
bore migrated upstream to the headgate before stabilizing, creating a
151SURGING AND CONTINUOUS TURBIDITY CURRENTS IN AN EXPERIMENTAL INTRASLOPE BASIN
F
IG
. 4.—Four regimes of deposition described by the deposit ponding index are
A) a perfectly ponded deposit: Po
5
1, B) a mounded deposit: Po
.
1, C) a
perfectly draped deposit: Po
5
0, and D) a deposit with accentuated highs: Po
,
0.
F
IG
. 5.—Schematic diagram of a continuous turbidity current flowing into the experimental flume after the flow reached the basin lip but before setup was complete
(solid line) and after setup of a dammed turbidity current (dashed lines). Note the upstream migrating bore (line 1), which eventually stabilized within the first 80 seconds
of each continuous-feed experiment into either a quasi-steady hydraulic jump for the case of experiment c3 (line 2a) or a drowned inlet for the case of experiments c1,
c2, and c4 (line 2b).
drowned inlet, and again resulting in a quasi-steady, highly subcritical tur-
bidity current (Fig. 5). Owing to the presence of the drowned inlet, each
of these runs can be interpreted in terms of an equivalent run with super-
critical flow and a hydraulic jump to highly subcritical flow specified far-
ther upstream of the headgate.
In the experiments the damming effect was strong enough to produce a
dammed turbidity current, i.e., a current that because of an obstruction is
very deep, very slow-flowing, and has a very low densimetric Froude num-
ber. This flow damming was manifested in the nonturbulent, glassy inter-
face between the turbid water and the clear water above in Figure 3B,
evidencing very slow, subcritical flow. In the zone of dammed flow, the
interface between turbid water and clear water above was horizontal from
either the downstream side of the jump (experiment c3) or the headgate
(experiments c1, c2, and c4) to a point near the basin lip, beyond which it
dropped downward as the current flowed over the downstream basin lip
(Fig. 3B). This is in strong contrast to the head of the turbidity current of
Figure 3A before setup was complete, which showed strong turbulent de-
formation. It should be noted that the continuous turbidity currents in these
experiments still overflowed the downstream lip of the basin, i.e., the flows
were partially rather than fully dammed. Although the experimental basin
was never completely filled with sediment, damming of the turbidity cur-
rent would likely have persisted until the basin was nearly full of sediment,
so that the lip no longer presented an obstacle.
The turbidity-current deposits, or turbidites, resulting from the continu-
ous flows had a drape-like geometry as shown in Figure 6. The deposits
had a deposit ponding index ranging from 0.11 to 0.37, indicating a deposit
closer to a draped deposit than a ponded deposit (Table 1). The deposit
ponding index varied inversely with the input flow rate for runs c1, c2,
and c4, indicating that higher discharges led to more draped deposits. Run
c3, however, did not follow this trend.
Three distinct depositional zones are recognizable in Figure 6. A prox-
imal zone extends from the headgate to about 1.25 m downstream. In this
zone the bed has a slope of about 0.084, or a slope angle of about 4.8
8
.
Beyond this is a basin zone extending from 1.25 m to 3.5 m, over which
the average bed slope is 0.0022 and the slope angle is 0.012
8
. Beyond this
is the zone of the basin lip, over which the bed rises by 12.7 cm over a
length of 0.5 m.
Each of the deposits from runs c1–c3, which used a slurry composed of
kaolinite and 20
m
m silt, had a nearly constant thickness in the proximal
zone and in the corresponding basin zone, as shown in Figure 6. In the
case of run c4, which used kaolinite and 45
m
m silt, the proximal deposit
formed a wedge tapering down to the deposit thickness in the basin zone.
This tapering of deposit thickness in the streamwise direction occurred even
in the basin zone. Evidently the coarser sediment tended to deposit pref-
erentially in the proximal zone. In all cases very little deposition occurred
in the lip zone.
152 M.P. LAMB ET AL.
F
IG
. 6.—Graph of the final deposit elevation for the four continuous-feed runs (c1–c4). Runs c1, c2, and c3 had a 20
m
m silt fraction and c4 had a 45
m
m silt fraction.
The slurry input flow rate was 0.85 l/s, 0.5 l/s, 2 l/s and 1 l/s, for c1–c4, respectively. The deposits for all of the continuous turbidity currents had deposit ponding indices
most representative of a drape.
For the continuous flows presented here, the glassy interface between
sediment-laden water and clear water stabilized at an elevation above the
downstream lip of the basin after setup of dammed flow was complete, as
can be seen in Figure 3B. This led to significant basin overflow and inef-
ficient trapping of sediment within the basin. The trapping efficiency of the
minibasin is here defined as the ratio of the sediment deposited in the basin
during the experiment to the total sediment fed in during the same exper-
iment. As shown in Table 1, runs c1, c2, and c3 had approximately equal
trapping efficiencies of 15%, in spite of the fact that inflow slurry discharge
took the corresponding values 0.5, 0.85, and 2.0 l/s. The inflow slurry in
all these runs consisted of 20
m
m silt and kaolinite clay. Run c4, on the
other hand, had a trapping efficiency of 32%; the inflow slurry discharge
was 1 l/s and the sediment consisted of 45
m
m silt and kaolinite clay.
Evidently grain size exerts a much stronger control on trapping efficiency
than does flow discharge in these experiments.
Surging Turbidity Currents
The surging flows had turbulent heads that were identical to those of the
continuous flows. However, the cessation of slurry input after 15 seconds
caused the surging turbidity currents to have discrete, small bodies in com-
parison with the continuous turbidity currents. The surges were chosen to
be small enough (approximately 10% of the basin capacity) to prevent
much overflow from the basin. As a result, the turbidity-current head ap-
proached the lip, climbed up it, and then reflected backward with little
spillover. The surging flows did not undergo a hydraulic jump because the
input discharge was turned on for only 15 seconds, a time insufficient for
setup of quasi-steady flow with a hydraulic jump. Repeated reflection from
one end of the basin to the other again led to the formation of a muddy
pond in the basin, but this time with almost no overflow. The muddy pond
was allowed to settle out for approximately eight hours before the next
surge event.
An ultrasonic sonar system was used to measure the bed elevation be-
tween surges. Vertical stacking of these profiles led to the generation of
the reconstructed depositional record shown in Figure 7, which character-
ized the stratigraphic architecture of the basin fill. Each of the deposits
from the twelve surge events formed a thin, normally graded bed, char-
acterized by a cap of slightly darker fine sediment (Fig. 8).
The final deposit, consisting of twelve stacked turbidites, each resulting
from an individual surging turbidity current, was notably more ponded than
the deposits from the continuous turbidity currents (Table 1). Most of the
individual surge deposits were slightly ponded (Fig. 9), with an average
ponding index for the individual surge deposits of 0.11. This resulted in a
ponding index very close to zero for the cumulative surge deposit after the
first few surges, indicating a draped deposit. However, the addition of each
slightly ponded surge deposit resulted in a cumulative deposit that became
more ponded with each successive surge, reaching a value of 0.57 after the
last surge event (s12). Therefore, the ponding index of the cumulative surge
deposit varied directly with the surge number.
The proximal deposits of the surge-feed turbidity currents tended to be
very thin relative to the deposits in the basin zone (Fig. 7), in contrast to
those of the continuous runs (Fig. 6). The surging flows had very short
bodies (in length or duration), and thus were dominated by the turbulent
head of the turbidity current. In contrast, the continuous flows had long
bodies, and therefore were dominated by the less turbulent body of the
turbidity current. The implication is that within this proximal region the
heads of the turbidity currents were more competent and thus kept more
sediment in suspension than the bodies of the turbidity currents.
