Document Actions

IJEGE-11_BS-McCoy-et-alii

background image
Italian Journal of Engineering Geology and Environment www.ijege.uniroma1.it © 2011 Casa Editrice Università La Sapienza
715
DOI: 10.4408/IJEGE.2011-03.B-078
OBSERVATIONS OF DEBRIS FLOWS AT CHALK CLIFFS, COLORADO, USA:
PART 1, IN SITU MEASUREMENTS OF FLOW DYNAMICS, TRACER PARTI-
CLE MOVEMENT AND VIDEO IMAGERY FROM THE SUMMER OF 2009
S.W. M
C
COY
(*)
, J.A. COE
(**)
, J.W. KEAN
(**)
, G.E. t
uCkeR
(*)
,
D.M. s
taley
(**)
& T.A. w
asklewiCz
(***)
(*)
University of Colorado, CIRES & Department of Geological Sciences - Boulder, Colorado 80309, USA
Email: scott.mccoy@colorado.edu
(**)
U.S. Geological Survey, Denver Federal Center, MS 966 - Denver, Colorado 80225, USA
(**)
East Carolina University, Department of Geography - Greenville, North Carolina 27858, USA
to a lesser extent, peak surge depth and velocity. Our
growing database of natural debris-flow events can be
used to develop linkages between observed debris-flow
transport and entrainment responses and the controlling
rainstorm characteristics and flow properties.
K
ey
words
: debris flow, initiation, entrainment, sediment tran-
sport, monitoring, instrumentation, runoff, tracer particles
INTRODUCTION
Debris flows are rapidly gaining recognition as a
serious hazard to expanding development, as impor-
tant agents of mountain landscape evolution, and as
a significant mode of sediment transport to rivers in
rugged terrain. This recognition has made apparent
the need for complete and quantitative datasets from
well-constrained natural debris flows to test theoreti-
cal and computational approaches, confirm experi-
mental results and stimulate new ideas.
In steep, rocky, and sparsely vegetated terrain, de-
bris flows initiated by surface-water runoff in response
to short duration, moderate- to high-intensity rainfall
are common (e.g. f
Ryxell
& H
oRbeRG
, 1943; w
oHl
&
P
eaRtHRee
, 1991; m
eyeR
et alii, 1995; C
annon
et alii,
2001; b
eRti
& s
imoni
, 2005; l
aRsen
et alii, 2006; C
oe
et alii, 2008). Runoff-initiated debris flows can form
from small quantities of rain, with very short lag times
from the beginning of the rainstorm to the beginning of
the debris-flow event (b
eRti
et alii, 2000). They start
as water-rich flows that rapidly entrain and concentrate
ABSTRACT
Debris flows initiated by surface-water runoff dur-
ing short duration, moderate- to high-intensity rainfall
are common in steep, rocky, and sparsely vegetated
terrain. Yet large uncertainties remain about the po-
tential for a flow to grow through entrainment of loose
debris, which make formulation of accurate mechani-
cal models of debris-flow routing difficult. Using a
combination of in situ measurements of debris-flow
dynamics, video imagery, tracer rocks implanted with
passive integrated transponders (PIT) and pre- and
post-flow 2-cm resolution digital terrain models (ter-
rain data presented in a companion paper by s
taley
et
alii, 2011), we investigated the entrainment and trans-
port response of debris flows at Chalk Cliffs, CO, USA.
Four monitored events during the summer of 2009 all
initiated from surface-water runoff, generally less than
an hour after the first measurable rain. Despite reach-
scale morphology that remained relatively constant, the
four flow events displayed a range of responses, from
long-runout flows that entrained significant amounts of
channel sediment and dammed the main-stem river, to
smaller, short-runout flows that were primarily depo-
sitional in the upper basin. Tracer-rock travel-distance
distributions for these events were bimodal; particles
either remained immobile or they travelled the entire
length of the catchment. The long-runout, large-entrain-
ment flow differed from the other smaller flows by the
following controlling factors: peak 10-minute rain in-
tensity; duration of significant flow in the channel; and
background image
S.w. M
c
COY, J.A. COE, J.w. kEAN, G.E. t
ucKer
, D.M. S
tAley
& T.A. w
ASKlewicZ
716
5th International Conference on Debris-Flow Hazards Mitigation: Mechanics, Prediction and Assessment Padua, Italy - 14-17 June 2011
and debris-flow deposits (0°-45°) cover the remaining
40% of the basin (C
oe
et alii, 2008). Between debris-
flow events dry ravel from the steep colluvium and rock
fall from the bedrock cliffs rapidly fill the channels with
debris. Bedrock slopes in the basin are devoid of veg-
etation, and colluvial slopes are sparsely vegetated.
This basin has a high frequency of debris flows (one
to four events per year between the months of May and
October) generated by rainfall related surface-water
runoff (C
oe
et alii, 2008), making it ideal for long-term
monitoring. Basin topographic characteristics are sum-
marized in Table 1. A more complete description of the
geologic setting and site selection can be found in ear-
lier work (C
oe
et alii, 2008, C
oe
et alii, 2010).
MONITORING SYSTEM
The automated monitoring system, established
through collaboration between the U.S. Geological
Survey and the University of Colorado, consists of
three instrumented cross sections (upper, middle and
lower stations) and two video cameras filming at 2
frames per second (one filming the middle station, the
other filming the upper station) (Fig. 1A-D and H).
The upper station is located in the west channel, 38
m above the junction with the east channel and ~590
m downstream from the drainage divide. The middle
station is 92 m downstream of the upper station and
54 m downstream of the east-west channel junction.
The lower station is 319 m downstream of the middle
station and is 220 m from the mouth of the basin and
the apex of the depositional fan (Fig. 1H).
As of May 2009 each station was equipped with: an
ultrasonic stage sensor (accuracy +/- 1 cm) suspended
over the channel to measure the sediment and/or flow
height above a fixed datum (the datum is the stable
large quantities of debris and sediment into hazardous
debris-flow fronts (C
oe
et alii, 2008) and can have peak
discharges many times that of a comparable water-flood
(v
andine
, 1985). Yet, for this type of debris flow, large
uncertainties remain about the mechanics of initiation
and the potential for a flow to grow through entrainment
of loose debris. These uncertainties make it difficult to
formulate accurate mechanical models of debris-flow
initiation, routing, and deposition in natural settings and
thereby add large uncertainties to hazard assessments.
To relate the entrainment and transport response of
a debris-flow event to potential controlling factors such
as physical flow properties, rainstorm characteristics,
and basin topographic parameters we used a combina-
tion of in situ measurements of debris-flow dynamics,
video imagery, tracer rocks implanted with passive
integrated transponders (PIT) and pre- and post-flow
2-cm resolution digital terrain models acquired through
terrestrial laser scanning (TLS). Terrain data are pre-
sented in a companion paper in this volume by s
taley
et alii. This combination of readily available tech-
nologies captured a comprehensive record of the four
debris-flow events that occurred during the summer
of 2009 in a natural debris-flow basin at Chalk Cliffs,
Colorado, USA. We documented the initial and final
topographic state, the hydrologic conditions leading up
to the event, net transport distances of tracer particles
and continuous time series of physical flow properties
and flow dynamics for each debris-flow event.
In this paper and the companion paper by s
taley
et alii (2011), we compare the four debris-flow events
that occurred during the summer of 2009 (2 June, 26
July, 6 September, 15 September) to highlight dramat-
ic differences in entrainment and transport behavior
possible from a single basin with similar topographic
initial conditions for each event, yet driven by rain-
storms of different intensities and durations.
CHALK CLIFFS STUDY SITE
CATCHMENT DESRIPTION AND GEOLOGIC SETTING
The Chalk Cliffs study area is located on the south-
ern flank of Mount Princeton in the Sawatch Range of
Central Colorado, USA. The steep, 0.3 km
2
semi-arid
basin is incised into pervasively fractured and hydro-
thermally altered quartz monzonite, adjacent to the
range-bounding Sawatch normal fault (m
illeR
, 1999).
The steep headwaters of the basin are exposed bedrock
(generally > 45°) (Fig. 1A). Unconsolidated colluvium
Tab. 1 - Chalk Cliffs study basin topographic character-
istics at each monitoring station shown in Fig.
1A and H. The Chalk Creek location is on the fan
where the debris-flow channel enters Chalk Creek
(not a station). Elevation, slope, and length pa-
rameters correspond to the west channel profile
in Figure 1H (inset). Local slopes at the stations
were taken over 10-horizontal meters
background image
OBSERVATIONS OF DEBRIS FLOWS AT CHALK CLIFFS, COLORADO, USA: PART 1, IN SITU MEASUREMENTS OF FLOW DYNAMICS,
TRACER PARTICLE MOVEMENT AND VIDEO IMAGERY FROM THE SUMMER OF 2009
Italian Journal of Engineering Geology and Environment www.ijege.uniroma1.it © 2011 Casa Editrice Università La Sapienza
717
complete sensor specifications). All these sensors were
sampled every 2 seconds during a rainstorm.
At the upper station only, we mounted a 15.24 cm
by 15.24 cm (232 cm
2
) metal plate attached to a single-
axis load cell (accuracy +/- 2.5 kg) flush with the bedrock
channel to measure total normal stress at the sediment-
bedrock interface. This load cell was sampled at 250 Hz
through June 2009; we subsequently reduced the sam-
bedrock channel at the upper and middle stations and a
buried pressure transducer at the lower station); one or
two unvented pressure transducers (accuracy +/- 3 mm
water pressure head) to measure the pore-fluid pressure
at the bedrock-sediment interface (upper and middle
stations) or at ~ 0.25 m depth in the bed sediment (low-
er station); a siphoning rain gauge (accuracy +/- 2%);
and a temperature sensor (see m
C
C
oy
et alii, 2010 for
Fig. 1 - Chalk Cliffs study area with photographs of the monitoring stations and shaded relief images (from 1 m ALSM data pro-
vided by NCALM) with tracer-particle positions marked by white circles. white rectangle in H encompasses zoomed-in
view shown in E, F, and G. A) Line drawing of study basin showing station locations and the distribution of bedrock and
colluvium (redrawn from m
c
c
oy
et alii, 2010) B) Upper station with sensors. C) Middle station. D) Lower station. E)
Starting tracer-particle positions 20 May 2009. F) Positions 11 June 2009, 9 days after 2 June 2009 event. G) Positions
4 August 2009, 9 days after 26 July 2009 event. H) Positions 26 September 2009, 11 days after 15 September 2009 event
(note no survey was completed after the 06 September 2009 event). Inset: Long profile of east and west channel
background image
S.w. M
c
COY, J.A. COE, J.w. kEAN, G.E. t
ucKer
, D.M. S
tAley
& T.A. w
ASKlewicZ
718
5th International Conference on Debris-Flow Hazards Mitigation: Mechanics, Prediction and Assessment Padua, Italy - 14-17 June 2011
error, +/- ~0.5 m, in the surveyed position was pri-
marily due to the read-range of the antenna.
To measure the total curvilinear distance travelled
(streamwise and cross-stream) by a tracer rock during
a single debris-flow, we projected tracer-particle posi-
tions onto the longitudinal profile of the debris-flow
channel. The longitudinal profile was extracted from a
1 m digital elevation model derived from airborne la-
ser swath mapping (ALSM) data collected by the Na-
tional Center for Airborne Laser Mapping (NCALM).
We then calculated along slope travel distance by dif-
ferencing the post-event and pre-event slope distance
along the longitudinal profile. All tracer rocks that
moved less than 1 m were put in the 1 m bin for plot-
ting on the log-log scale in Fig. 2.
pling rate to 100 Hz. Additional measurements of rain-
fall and soil moisture (accuracy +/- 5% volumetric water
content) were made on adjacent hillslopes within the
basin and within the channel below the middle station.
Near-realtime data and a daily photograph from
the upper station were sent to a web server using a cell
phone modem. This system allowed us to know when
a debris-flow event had occurred (http://landslides.
cr.usgs.gov/monitoring/chalk_cliffs/index.php
).
METHODS
TRACER ROCkS
We implanted 32-mm passive integrated trans-
ponders (PIT) in 99 rocks to measure the travel dis-
tances of cobbles during a debris-flow event. PIT
tags “labelled” each rock with a unique identification
number. Using radio-frequency identification (RFID)
technology, we could find and identify these tracer
rocks, buried up to ~ 0.25-0.5 m depth, without dis-
turbing the bed surface (b
Radley
, 2010). We selected
the tracer rocks from the area surrounding the upper
station. Rock strength determined the size of selected
rocks; rocks that were too small broke when drilled
with the hammer drill. The mean length of the a, b,
and c axes of the tracer rocks were 13 cm, 9 cm, and
6 cm, respectively. For comparison, the mean length
of the a, b, and c, axes of 100 randomly selected rocks
from an October 2008 debris-flow deposit forming the
levees and bed at the upper station were 10 cm, 7 cm,
and 4 cm, respectively. We prepared the tracer rocks by
drilling a 4.76 mm diameter hole with a hammer drill,
inserting a 32 mm long PIT tag and then sealing the
hole with marine epoxy. To aid in the recovery of sur-
face rocks and the initial survey, we spray-painted all
tracers red. We then placed the rocks in groups of three,
along the centerline of the channel, in the reach by the
upper station, with each group separated by ~2 m (Fig.
1E), and surveyed their positions using a total station.
After each debris-flow event, we searched the
entire travel-path of the debris flow using a 0.5 m di-
ameter RFID detection antenna. When a tracer rock
was passed over by the antenna, the capacitor in the
PIT tag was charged, which enabled the tag to trans-
mit its unique identification number to the receiver
to be recorded. Once the tracer rock was recorded,
we surveyed its location using a total station. Many
of the tracer rocks were buried below the deposit
surface and never visually identified. The resulting
Fig. 2 - Histograms of tracer-particle travel distance be-
tween consecutive surveys. Solid line is the travel
distance from the mean starting position on 20
May to the junction with the perennial stream
Chalk Creek. Dashed line is the distance trav-
elled by the flow front. In C, both lines overlap.
A) Distance travelled due to 2 June 2009 event. B)
Distance travelled due to 26 July 2009 event. C)
Distance travelled due to combination of 6 Sep-
tember 2009 and 15 September 2009 events
background image
OBSERVATIONS OF DEBRIS FLOWS AT CHALK CLIFFS, COLORADO, USA: PART 1, IN SITU MEASUREMENTS OF FLOW DYNAMICS,
TRACER PARTICLE MOVEMENT AND VIDEO IMAGERY FROM THE SUMMER OF 2009
Italian Journal of Engineering Geology and Environment www.ijege.uniroma1.it © 2011 Casa Editrice Università La Sapienza
719
STATION-DATA PROCESSING
To remove much of the +/- 2.5 kg noise in the force
time series, we downsampled the 250 Hz and 100 Hz
force plate data to 10 Hz by binning data in 0.1 s bins
and taking the mean of each bin. The stress axis in Figs.
3-6 was scaled such that stage H and basal stress σ su-
perpose if σ
=
ρgH cos(θ) with ρ=2040 kg/m
3
, where
ρ is the wet bulk density of the sediment bed and flow
combined, g is the gravitational acceleration, H is the
bed-normal height of material above the force plate,
equal to stage, and θ is the bed inclination (i.e., σ plots
above H when ρ
actual
>2040 kg/m
3
and σ plots below H
when ρ
actual
<2040 kg/m
3
). Note that accurate bulk densi-
ties can only be determined from this method when the
assumption of a lithostatic stress state is valid.
The load cell was buried beneath sediment for the
majority of 2009, which precluded taking a new tare val-
ue before each event. To minimize error associated with
a tare value that is a function of environmental factors
(primarily temperature), we took advantage of the fact
that before rainstorms the bed sediment was dry (~10%
volumetric water content) and thus should have a rela-
tively constant bulk density before each event. We set
the force tare value using a pre-storm dry bed sediment
bulk density of 1750 kg/m
3
; this value was determined
just after installation. We estimated the approximate un-
certainty resulting from this assumption as +/- 100 kg/
m
3
or +/- 500 Pa if 0.5 m of sediment is present.
We filtered the 2 s ultrasonic stage data based on
echo return strength to remove low strength returns.
Persistent increases or decreases in stage were due to
sediment deposition or erosion.
To measure pore-fluid pressure using the unvent-
ed, temperature-compensated pressure transducers,
we subtracted the pre-storm pressure (i.e., atmos-
pheric pressure) from the total pressure. For most of
the events, the bed sediment between the flow and the
pressure transducer was not saturated. Given these
conditions, the pressure transducer did not measure
positive pressure head due to the overriding flow (not
shown in our figures) or the pressure head was attenu-
ated. The only exception to these behaviours was re-
corded at the middle station on 2 June.
We calculated surge front velocities (each event
was composed of multiple surge fronts) by manually
tracking surge front positions in the spatially referenced
video imagery recorded by the upper station camera.
Surge-front velocities Ū were then plotted as a function
of maximum surge depth h to develop a rating curve
for the upper station (ū
=
3.6h
1/2
). To calculate maxi-
mum flow depth, the height of the bed sediment and the
maximum stage during the surge must be known. We
used the period of no or little flow in between succes-
sive surges to determine the bed height through time.
We automated this procedure by taking a moving-win-
dow minimum of the stage data, with a window width
equal to the maximum surge duration for each event.
It has been demonstrated by others (P
ieRson
, 1986;
s
uwa
et alii, 1993; a
Rattano
et alii, 2000) that the peak
flow velocity can occur after the peak flow depth and
coarse-grained flow front have passed. Since many of
the surges were of short duration and we only measured
the surge front velocities, we did not detect this phe-
nomena, but if present, it would make our rating curve
underestimate the actually velocities in the more shal-
low, dilute portions of the surge.
For each debris-flow event, we calculated an esti-
mate of the total event volume passing the upper sta-
tion (sediment plus water) using
where the summation was taken from the beginning of
the flow event t
i
to the end t
f
, ū is the mean cross sec-
tional velocity from the rating curve, A(t) is the current
active cross sectional area at a given time calculated us-
ing the surveyed bedrock cross section, the current bed
height, and current flow depth, and Δt is the length of
time between two stage measurements (typically 2 s).
To determine the total time spent above certain
transport stages during the coarse of a debris-flow
event we summed the total time over which the flow
depth was greater than 0.1 m, 0.2 m, and 0.3 m.
RESULTS AND DISCUSSION
Monitoring results from the four debris-flow
events indicated that 2 June and 26 July events had
relatively short travel distances, stayed within the
basin and were primarily depositional in nature. In
sharp contrast, the 15 September event was prima-
rily erosional in nature and traveled all the way to
the modern fan at Chalk Creek, where it temporar-
ily dammed the stream. The 6 September event was
intermediate to these two end members, exiting the
basin, but not reaching Chalk Creek. Below, we first
provide details on the tracer-particle travel distance
distributions and then describe the conditions that
background image
S.w. M
c
COY, J.A. COE, J.w. kEAN, G.E. t
ucKer
, D.M. S
tAley
& T.A. w
ASKlewicZ
720
5th International Conference on Debris-Flow Hazards Mitigation: Mechanics, Prediction and Assessment Padua, Italy - 14-17 June 2011
the perennial stream Chalk Creek (which flows down
a low-gradient glacial valley). The distance from the
mean tracer starting position on 20 May to the end
of the debris-flow dominated system (the junction of
the debris-flow channel and Chalk Creek) was 1620
m. For the 2 June and 26 July events, the mean travel
distances were 10 m and 4 m, or only 0.6% and 0.2%
of the channel length dominated by debris flows. The
mean tracer travel distance was also considerably
shorter than the maximum flow-front travel distance
for these two flows, which was on the order of 300
m. In contrast, the mean tracer travel distance for the
15 September event, which had a maximum flow-front
travel distance of 1620 m, was 1100 m or 68% of the
1620 m debris flow dominated channel length. Sixty-
seven percent of the recovered population travelled a
distance equal to the debris flow dominated channel
length in one event. Moreover, both the mean travel
distance of 1100 m and the percent of particles that
travelled the entire length of the debris flow dominated
channel length are underestimated for the 15 Septem-
ber event due to undercounting of tracer particles. Up-
stream of the fan, few debris-flow deposits had depths
greater than the read range of the antenna (s
taley
et
alii, 2011), whereas the reach from county road 162 to
Chalk Creek had significant areas of aggradation great-
er than 2 m. Had all particles been found, the measured
15 September distribution would be even more heavily
skewed toward travel distances closer to the length of
the debris flow dominated portion of the catchment.
Travel-distance distributions were bimodal: parti-
cles either remained nearly immobile or they travelled
the entire length of the debris flow dominated portion
of the catchment (Fig. 2). Although we only have
tracer data from three events, if this bimodal distribu-
tion is characteristic of travel distances for particles
transported by debris flows, it sets them apart from
many other modes of sediment transport, in which the
single-event travel distance is often a small fraction of
the system size (b
Radley
, 2010). The existence of par-
ticle travel distances that are equal to the system size
calls into question the ability to write sediment trans-
port equations that depend only on local topographic
and flow parameters (t
uCkeR
& b
Radley
, 2010).
STATION DATA
Monitoring station data for the four debris-flow
events are shown in Figs. 3-6. Key storm character-
caused this dramatic difference in entrainment and
transport behavior between the 4 events.
TRACER ROCkS
Of the initial 99 PIT-tagged tracer rocks (Fig. 1E),
the number of tracer rocks with known pre-event and
post-event positions was 84, 68, and 33 after the 2
June, 26 July, and 15 September events, respectively.
We were not able to conduct a survey after the 6 Sep-
tember event do the rapid arrival of the 15 September
event. These recovery rates are much smaller than
those obtained using the same techniques and tech-
nology in a fluvial environment, where the rates can
be up to 98% (b
Radley
, 2010). At Chalk Cliffs, the
high losses were due to burial of many of the tracers,
especially in the 15 September event, to depths greater
than the read range of the antenna (~0.25-0.5 m).
In our early surveys, we found most of the tracer
rocks in the seed reach near their origins (Fig.1 E-G).
However, our final survey on 26 September 2009, fol-
lowing the 15 September event, showed that the high-
est concentration of tracer rocks had shifted ~1.6 km
downstream to the fan at the junction of the debris-
flow channel and the perennial stream Chalk Creek
(Fig. 1H). The 6 September event did not reach county
road 162 (Fig. 1H), however 67% of tracer particles
were found below the road on 26 September. Because
extended sediment transport was due to the 15 Sep-
tember event, and only small amounts of erosion were
measured at the monitoring stations after the 6 Sep-
tember event (Fig. 5), we grouped the 6 September
event with the earlier lower-transport events.
The fraction of particles that moved greater than 1
m after the 2 June, the 26 July, and the 15 September
events were 67%, 25% and 94%, respectively. Distri-
butions of travel distance are plotted in Fig. 2 and sum-
marized in Table. 2. Travel distances measured after
the 2 June event and the 26 July event (Fig. 2A and B)
differ greatly from the distribution for the 15 Septem-
ber event (Fig. 2C). Travel distances < 1 m dominate
the distributions for the low transport events (Fig. 2A
and B), whereas travel distances > 1500 m dominate
the distribution for the 15 September event (Fig. 2C).
The maximum possible travel distance for a tracer
particle transported by a debris flow is set by its start-
ing position and the downstream end of the debris-flow
dominated portion of the catchment, which in this case
is near the junction of the debris-flow channel and
background image
OBSERVATIONS OF DEBRIS FLOWS AT CHALK CLIFFS, COLORADO, USA: PART 1, IN SITU MEASUREMENTS OF FLOW DYNAMICS,
TRACER PARTICLE MOVEMENT AND VIDEO IMAGERY FROM THE SUMMER OF 2009
Italian Journal of Engineering Geology and Environment www.ijege.uniroma1.it © 2011 Casa Editrice Università La Sapienza
721
RAINFALL
Cumulative rainfall amounts since the beginnings
of each debris-flow producing storm event are plot-
ted as a dashed line in panel A of Figs. 3-6. On the
upper axis of panel A in Figs. 3-6, 10-minute rainfall
intensities are plotted. With exception of the 2 June
event, all storms had durations less than 2 hours. The
15 September event had the largest cumulative storm
rainfall (24.6 mm) and the highest peak 10-minute in-
tensity (38.1 mm/hr). The 6 September event had the
smallest storm total (6.3 mm), and the 2 June event
istics, flow properties, and tracer travel distances are
summarized in Table. 2. By comparing the measured
rainfall and flow properties between the four 2009
events, we can gain insight into why the September
15 flows were so much more effective, in terms of
travel distance, clast transport, and bed entrainment,
than the previous three events.
Fig. 3 - Sensor data 2 June 2009. A) Upper axis:
10-minute rainfall intensities measured at the up-
per station. Lower axis: Complete event duration
measured at the upper and middle station. Flows
stopped upstream of the lower station. The first
debris flows initiated in the west channel, arrived
at the upper station at 16:22:54 and stopped be-
fore reaching the middle station. The second set
of debris flows initiated in the east channel and
arrived at the middle station at 17:09:14. Pre-flow
sediment depth at the upper station was 0.47 m
and 0.26 m at the middle station. Net change in
sediment depth was +0.54 m and +0.78 m at the
upper and middle station, respectively. Due to rain
at 11:00, pre-event pore pressures were not zero.
B) 5-minute time slice encompassing the arrival of
the first 5 debris-flow surges at the upper station.
C) 20-minute time slice from the middle station
Fig. 4 - Sensor data 26 July 2009. A) Upper axis:
10-minute rainfall intensities measured at the up-
per station. Lower axis: Complete event duration
measured at the upper and middle station. Flows
stopped upstream of the lower station. The first
surge front reached the upper station at 7:12:42
and the middle station at 7:17:32. Pre-flow sedi-
ment depths at the upper and middle stations were
1.04 m, and 1.01 m, respectively. Net changes
in sediment depth at the upper and middle sta-
tions were +0.19 and +0.07 m, respectively. B)
25-minute time slice from the upper station. C)
25-minute time slice from the middle station
background image
S.w. M
c
COY, J.A. COE, J.w. kEAN, G.E. t
ucKer
, D.M. S
tAley
& T.A. w
ASKlewicZ
722
5th International Conference on Debris-Flow Hazards Mitigation: Mechanics, Prediction and Assessment Padua, Italy - 14-17 June 2011
had the lowest peak 10-minute intensity (9.1 mm/hr).
The first debris-flow surge generally reached the up-
per station in less than 45 minutes after the onset of
rain. There was a poor correlation in time between the
storm maximum 10-min intensity and flow initiation,
but a tight correlation, often less than a few minutes,
with local peaks in 10-minute rainfall intensity and
flow initiation (Figs. 3-6). The short temporal lag be-
tween the beginning of rain and the first debris-flow
surge (as short as 10 minutes) and the close correlation
in time with local peaks in the 10-minute rainfall in-
tensity reinforce the role of surface-water runoff as the
debris-flow generation process, as opposed to longer
response time, infiltration related, land sliding. Simi-
lar short response times from the beginning of rain to
the initiation of a debris flow have been observed at
other field sites (e.g. b
eRti
et alii, 2000)
To quantify the role of cumulative rainfall in
determining event volume, we normalized the event
volume (sediment plus water) measured at the upper
station by total rain volume that fell upstream of the
upper station (calculated as the product of upstream
Fig. 5 - Sensor data 6 September 2009. a) Upper axis:
10-minute rainfall intensities measured at the upper
station. Lower axis: Complete event duration meas-
ured at the upper, middle and lower stations. Stage
at middle station shifted -0.3 m in all plots. The first
surge front reached the upper, middle and lower
stations at 14:05:18, 14:06:38, and 14:09:41, re-
spectively. Pre-flow sediment depths at the upper,
middle and lower stations were 1.14 m, 1.06 m, and
0.22 m, respectively. Net change in sediment depth
at the upper, middle and lower stations were -0.04
m, - 0.06 m, and 0.0 m, respectively. b) 6-minute
time slice from the upper station. c) 10-minute time
slice from the upper, middle and lower stations
Fig. 6 - Sensor data 15 September 2009. A) Upper axis:
10-minute rainfall intensities measured at the up-
per station. Lower axis: Complete event duration
measured at the upper station. B) Complete event
duration measured at the middle and lower sta-
tions. Stage at middle station shifted +0.2 m. The
first surge front reached the upper, middle and low-
er stations at 17:38:18, 17:38:40, and 17:42:41,
respectively. Pre-flow sediment depths at the up-
per, middle and lower stations were 1.1 m, 1.0 m,
and 0.22 m, respectively. Net changes in sediment
depth at the upper, middle and lower stations were
-1.1 m, -0.28 m, and +0.14 m, respectively C):
20-minute time slice from the upper station
background image
OBSERVATIONS OF DEBRIS FLOWS AT CHALK CLIFFS, COLORADO, USA: PART 1, IN SITU MEASUREMENTS OF FLOW DYNAMICS,
TRACER PARTICLE MOVEMENT AND VIDEO IMAGERY FROM THE SUMMER OF 2009
Italian Journal of Engineering Geology and Environment www.ijege.uniroma1.it © 2011 Casa Editrice Università La Sapienza
723
station had a well-defined h
1/2
scaling, and were best fit
by ū
=
3.6
h
1/2
. We measured maximum front velocities
of 3.7 m/s during the 15 September event.
STAGE
Stage data from the upper, middle and lower sta-
tions are shown in Figs. 3-6 for all four events. Nei-
ther the 2 June event nor the 26 July event reached the
lower station. Both of these events had maximum flow
depths < 0.35 m at the upper and middle stations and
peak discharges at the upper station less than 2 m
3
/s.
The two events that travelled past the lower station,
6 September and 15 September, had maximum flow
depths >= 0.45 m at the upper and middle stations and
peak discharges at the upper station > 5 m
3
/s. Peak
flow depth of the 15 September event was not consist-
ently larger than the 6 September event (Tab. 2).
Sediment deposition occurred at the upper and
middle station during the short-runout flows of 2 June
and 26 July (Tab. 2). In contrast, the 6 September event
eroded the bed slightly (<= 0.08 m) at the upper and
middle stations. The 15 September event eroded to bed-
rock at the upper station (1.1 m) and eroded significant-
ly at the middle (0.34 m) and lower stations (0.11 m).
The volume (both sediment and water) of each
flow measured at the upper station varied significantly
between the long-transport 15 September event and the
other shorter-transport events. The 15 September event
had an event volume at the upper station of ~820 m
3
,
which was ~4 to 23 times larger than the other events
(Table 2). Note that these volumes are not totals for each
event because material from the other main tributary
(east channel) joins the west channel down slope of the
accumulation area and measured rainfall). This ratio
was not constant (range of 6% to 57%) with the large
15 September event having a ratio of 54% and the
smallest (2 June event) having 6% (Tab. 2). During
small flow events, a larger part of the water potentially
available for overland flow and debris transport was
not measured. In larger events, a greater part of availa-
ble water passed the station as surface flow of measur-
able depths, often as debris flows or water-dominated
flow vigorously transporting sediment.
VIDEO OBSERVATIONS
Although each debris-flow event varied in size and
duration, all events consisted of multiple surges with
distinct longitudinal sorting of sediment grainsize. Each
surge began with a steep and deep granular front, com-
posed of boulders and other coarse-grained material
(often without visible interstitial fluid), which was fol-
lowed by a shallower, water-rich tail of relatively fine-
grained material. The finer-grained tail rapidly changed
from a mud-rich slurry of intermediate thickness with a
moderate concentration of visible coarse-surface clasts
directly behind the granular front, to a shallow, water-
dominated, turbulent flow in which intense bed load
transport was evident. The duration and depth of the wa-
ter-dominated tail flows varied greatly. In small events
like that of 2 June, the water- dominated flow between
surge fronts was a trickle, compared to the 15 Septem-
ber event in which the water-dominated flow vigorously
transported sediment between surge fronts and contin-
ued transporting and eroding sediment for tens of min-
utes after the last well-developed granular surge front
had passed. Flow front velocities measured at the upper
Tab. 2 - Summary of rainstorm characteristics, flow properties, and tracer travel distances for the four debris-flow events
in 2009. Front velocities measured from video at the upper station are denoted with Vus. Front velocities calculated
from travel time between the middle station and the upper station are denoted with Vtt. US, MS, LS are upper station,
middle station and lower station, respectively
background image
S.w. M
c
COY, J.A. COE, J.w. kEAN, G.E. t
ucKer
, D.M. S
tAley
& T.A. w
ASKlewicZ
724
5th International Conference on Debris-Flow Hazards Mitigation: Mechanics, Prediction and Assessment Padua, Italy - 14-17 June 2011
upper station. We did not calculate volumes at the middle
station because the poor spatial referencing of video im-
agery there precluded accurate velocity measurements.
The total length of time that the measured flow
depth was greater than 0.1 m, 0.2 m, or 0.3 m over the
coarse of each debris-flow event varied significantly
between large and small transport events (Tab. 2). The
total time elapsed while the measured flow depth was
greater than 0.1 m, 0.2 m, or 0.3 m was largest for
the 15 September event. Although the 6 September
event had surges of comparable depth to the 15 Sep-
tember event, there were fewer surges and the shallow
flow between surge fronts only lasted for a short time.
The short duration of shallow flow observed during
the 6 September event is in strong contrast to the 15
September event where there were sustained periods
of flow greater than 0.1 m between the deeper surge
fronts. This indicates that total event volume can be
strongly influenced not only by the largest surges, but
also by sustained periods of moderate flow.
For the two larger events, September 6 and Sep-
tember 15, we observed changes in surge character-
istics as the flows moved downstream past the three
stations (Fig. 5 and 6). The coalescence of small surges
into larger, longer-duration surges is particularly pro-
nounced in the 6 September event (Fig. 5C). During
the 15 September event, the opposite behavior oc-
curred after 17:50. At this time, flows measured at the
upper and middle stations became water-dominated
and lacked the deep surge fronts measured during the
previous 10 minutes. However, by the time these mod-
erate-depth, water-dominated flows reached the lower
station, deep surges had developed (Fig. 6). Although
there was a range in surge depth (0.31 m to 1.1 m at
the upper station) almost all surges had the asymmetric
shape characteristic of debris flows, with a steep and
deep front followed by a more gently decreasing tail.
TOTAL NORMAL BASAL STRESS
Total normal basal stress varied in phase with
stage, with no consistent time lag observed between
changes in stage and changes in stress. Assuming a
one-dimensional static stress state, we calculated that
the mean wet bulk densities of the flows ranged be-
tween 1400 and 2200 kg/m
3
(see m
C
C
oy
et alii, 2010
for an explicit density time series). We measured dens-
er mixtures near the surge front (1800-2200 kg/m
3
),
and less dense mixtures in the water-dominated tails
(1400-1800 kg/m
3
). A similar range of densities have
been measured in other field and large-scale laboratory
measurements of passing debris flow surges (P
ieRson
,
1986; m
C
a
Rdell
et alii, 2007; i
veRson
et alii, 2010)
PORE-FLUID PRESSURE AND SOIL MOISTURE
Over the four different debris-flow events, the
pore-fluid pressure measured at the sediment-bed-
rock interface was variable and depended largely on
the thickness of overlying bed sediment and the state
of saturation of the bed sediment. For the majority
of the events, no pressure response due to the pres-
sure head of the overriding flow was measured (pre-
sumably because the static bed sediment between the
overriding debris flows and the pressure transducer
at the bedrock-sediment interface was partially dry).
When there was no bed sediment covering the pres-
sure transducer and the flow interacted directly with
the transducer (such as during the end of the event
on the 15 September) we measured large spikes in
pore-fluid pressure. Only during the 2 June event did
we measure a significant pressure response beneath
some thickness of bed sediment. During this event,
positive pore-fluid pressures were developed in the
static bed sediment at the upper and middle stations
prior to the first surge front arriving. At the upper
station, we measured an attenuated pressure signal
that generally followed the trend of flow stage, but
sharp increases in pore-fluid pressure lagged sharp
increases in stage and normal stress by 4-8 s (Fig.
3B). We presume the pressure signal was muted be-
cause the bed sediment may not have been complete
saturated. From video observations, it was evident
that the measured lag between increases in stage and
increases in pore-fluid pressure was equal to the time
the coarse-grained, visually unsaturated front spent
traveling over the sensors. Similar lags between the
rise in stage and the rise in pore-fluid pressure have
been repeatedly measured in large-scale laboratory
debris flows and related to the passage of the dilated,
coarse grain granular front (i
veRson
et alii, 2010).
At the middle station during the 2 June event, we
measured a less attenuated pressure signal and peak
measured pressures were up to two times hydrostatic
(m
C
C
oy
et alii, 2010, Fig. 3C).
Volumetric water content (i.e. soil moisture)
measured at depths of ~5 cm and ~45 cm in the chan-
nel sediment ~50 m downstream of the middle sta-
background image
OBSERVATIONS OF DEBRIS FLOWS AT CHALK CLIFFS, COLORADO, USA: PART 1, IN SITU MEASUREMENTS OF FLOW DYNAMICS,
TRACER PARTICLE MOVEMENT AND VIDEO IMAGERY FROM THE SUMMER OF 2009
Italian Journal of Engineering Geology and Environment www.ijege.uniroma1.it © 2011 Casa Editrice Università La Sapienza
725
and maximum distances were generally tens of meters
and the tracers only travelled a small fraction of the
debris-flow dominated catchment length. In the con-
trasting long-transport mode, the population of tracer
particles had mean and maximum travel distances on
the order of thousands of meters and the majority of
the population travelled the entire length of the catch-
ment during a single flow event.
TLS surveys (s
taley
et alii, 2011) taken before
and after flow events showed little change in reach-
scale topographic parameters over the course of the
summer. Therefore, the large variation in observed
transport and entrainment response must be controlled
by individual storm characteristics and flow proper-
ties, rather than by changes in flow-path topography.
Large-entrainment, long-transport flows were best
differentiated from small-entrainment, short-transport
flows by peak 10-minute rain intensity, total elapsed
time with flow of significant depth in the channel, and
to a lesser extent, peak surge depth and velocity. Larger
events had higher 10-minute rain intensities and during
these higher intensities (20-40 mm/hr) a larger fraction
of the precipitation exited the basin as measurable
debris-flow surges and overland flow. The 15 Septem-
ber event had the longest elapsed time during which
flow depths were measurable (>0.1 m), and the long-
est elapsed time during which flow depths were deep
(>0.3 m), in addition to having the most rapidly mov-
ing surge fronts. With a growing dataset of events in
which both the effects of the debris-flow event and the
triggering rainstorm and flow properties are quantified
it will be possible to test the strength of these correla-
tions and gain additional insight into the controlling
mechanisms of initiation, entrainment, and transport
by debris flows in natural, uncontrolled settings.
ACKNOWLEDGEMENTS
This research was supported by the National Sci-
ence Foundation (NSF) Graduate Fellowship, NSF
grants EAR-0643240 and 0934131, NSF CAREER
grant 0239749, and the U.S. Geological Survey Land-
slide Hazards Program. ALSM data was provided by
NCALM. We thank Joe Gartner, for field assistance
and Nate Bradley for advice and technical help with
the RFID survey equipment. Careful reviews on an
earlier version of the manuscript by Mark Reid and
Nate Bradley improved the clarity of presentation.
tion, where total sediment depths were greater than 2
m, remained unchanged from the pre-storm values of
~10% and 13%, respectively, during the short dura-
tion storms of 26 July, 6 and 15 September (Tab. 2).
During the longer duration 2 June event soil moisture
increased from 15% to 18% at 5 cm, and hours after
the end of the flow event we measured soil moistures
> 30% at 45 cm depth. Our measurements indicate
that relatively dry bed sediments are typical of storms
in which the duration is short relative to the infiltra-
tion time. Only during longer duration storms and
only locally where the bed sediment was thin and sur-
rounded by large areas of bedrock (as was character-
istic of the upper and middle station) did sediments
approach saturation. The June 2 event also indicates
that merely having a saturated or nearly saturated bed
(as was locally present at both the upper and middle
station) does not ensure entrainment. The June 2 event
was depositional at both stations where the shallow
bed sediments were locally saturated.
SUMMARY AND CONCLUSIONS
Using a combination of PIT tagged tracer rocks,
automated monitoring, and terrestrial laser scanning,
we measured the entrainment and transport response
of the four debris-flow events that occurred at Chalk
Cliffs in the summer of 2009. We also constrained
the rainstorm characteristics, topographic parameters,
and flow properties that were responsible for the ob-
served responses. All the observed debris flows were
trigged by storms capable of generating surface-water
runoff. These storms generally had durations of less
than two hours, but variable intensities (10-40 mm/
hr). The arrival of debris-flow surges generally fol-
lowed the onset of rain by less than one hour. All
flow events consisted of multiple surges, with each
asymmetric surge composed of a steep granular front
and a more watery tail. Despite these similarities, the
four events in the summer of 2009 were surprisingly
variable in terms total volume and, especially, runout
length, and volume of entrained bed sediment. The 15
September event was conspicuously larger by these
measures than the preceding three flows.
Using PIT tagged tracer rocks we quantified the
large variability in particle transport distance and
found a strongly bimodal transport response. In the
short-transport mode, the population of tracer parti-
cles had transport distances in which both the mean
background image
S.w. M
c
COY, J.A. COE, J.w. kEAN, G.E. t
ucKer
, D.M. S
tAley
& T.A. w
ASKlewicZ
726
5th International Conference on Debris-Flow Hazards Mitigation: Mechanics, Prediction and Assessment Padua, Italy - 14-17 June 2011
REFERENCES
a
Rattano
m. & m
aRCHi
l. (2000) - Video-derived velocity distribution along a debris flow surge. Physics and Chemistry of the
Earth Part B, 25 (9): 781-784.
b
eRti
m. & s
imoni
a. (2005) - Experimental evidences and numerical modeling of debris flow initiated by channel runoff.
Landslides, 2 (3): 171-182.
b
eRti
m., G
enevois
R., l
aHusen
R., s
imoni
a. & t
eCCa
P. R. (2000) - Debris flow monitoring in the Acquabona watershed on the
Dolomites (Italian Alps). Physics and Chemistry of the Earth Part B-Hydrology Oceans and Atmosphere, 25 (9): 707-715.
b
Radley
d.n. (2010) - Modeling Sediment Transport with Random walks. PhD thesis, University of Colorado, Boulder,
Colorado, USA.
C
annon
s
H
., k
iRkHam
R
m
. & P
aRise
m. (2001) - wildfire-related debris-flow initiation processes, Storm king Mountain, Colorado.
Geomorphology, 39 (3-4): 171-188.
C
oe
J.a., k
inneR
d.a. & G
odt
J.w. (2008) - Initiation conditions for debris flows generated by runoff at Chalk Cliffs, Central Colorado.
Geomorphology, 96 (3-4): 270-297.
C
oe
J.a., k
ean
J.w., m
C
C
oy
s.w., s
taley
d.m. & w
asklewiCz
t.a. (2010) - Chalk Creek Valley: Colorado’s natural debris-flow
laboratory. In: m
oRGan
l.a. & Q
uane
s. (eds), Through the Generations: Geologic and Anthropogenic Field Excursions in the
Rocky Mountains from Modern to Ancient: Geological Society of America Field Guide 18. The Geological Society of America.
f
Ryxell
f.m. & H
oRbeRG
C.l. (1943) - Alpine mud flows in Grand Teton National Park, wyoming. Bulletin of the Geological
Society of America, 54 (3): 457.
i
veRson
R.m., l
oGan
m., l
aHusen
R.G. & b
eRti
m. (2010) - The perfect debris flow? Aggregated results from 28 large-scale
experiments. Journal of Geophysical Research, 115 (F3): F03005.
l
aRsen
i.J., P
edeRson
J.l. & s
CHmidt
J.C. (2006) - Geologic versus wildfire controls on hillslope processes and debris flow
initiation in the Green River canyons of Dinosaur National Monument. Geomorphology, 81 (1-2): 114-127.
m
C
a
Rdell
b.w., b
aRtelt
P. & k
owalski
J. (2007) - Field observations of basal forces and fluid pore pressure in a debris flow.
Geophysical Research Letters, 34 (7).
m
C
C
oy
s.w., k
ean
J.w., C
oe
J.a., s
taley
d.m., w
asklewiCz
t.a. & t
uCkeR
G.e. (2010) - Evolution of a natural debris flow:
In situ measurements of flow dynamics, video imagery, and terrestrial laser scanning. Geology, 38: 735-738.
m
eyeR
G.a., w
ells
s.G., & t
imotHy
J
ull
a
J
. (1995) - Fire and alluvial chronology in Yellowstone National Park: Climatic and
intrinsic controls on Holocene geomorphic processes. Geological Society of America Bulletin, 107 (10): 1211.
m
illeR
m.G. (1999) - Active breaching of a geometric segment boundary in the Sawatch Range normal fault, Colorado, USA.
Journal of Structural Geology, 21 (7): 769-776.
P
ieRson
t.C. (1986) - Flow behavior of channelized debris flows, Mount St. Helens, washington, In a
bRaHms
A.D. (ed.), Hillslope
Processes, Allen and Unwin, Boston, 269-296.
s
taley
d.m., w
asklewiCz
t.a., C
oe
J.a., k
ean
J.w., m
C
C
oy
s.w. & t
uCkeR
G.e. (2011) - Observations of debris flows at
Chalk Cliffs, Colorado, USA: Part 2, changes in surface morphometry from terrestrial laser scanning in the summer of
2009.
In: G
enevois
R. H
amilton
d.
l
. & P
Restininzi
a. (
eds
.). Debris-Flow Hazards Mitigation: Mechanics, Prediction and
Assessment. Italian Journal of Engineering Geology and Environment Book, 3.
s
uwa
H., o
kunisHi
k. & s
akai
M. (1993) - Motion, debris size and scale of debris flows in a valley on Mount Yakedake, Japan,
In Sediment Problems: Strategies for Monitoring, Prediction and Control, IAHS Publ., 217: 239-248.
t
uCkeR
G.e. & b
Radley
d.n. (2010) - Trouble with diffusion: Reassessing hillslope erosion laws with a particle-based model.
Journal of Geophysical Research, 115: F00A10.
v
andine
d
f
. (1985) - Debris flows and debris torrents in the southern Canadian Cordillera. Canadian Geotechnical Journal, 22 (1).
w
oHl
e.e. & P
eaRtHRee
P.P. (1991) - Debris flows as geomorphic agents in the Huachuca Mountains of southeastern Arizona.
Geomorphology, 4 (3-4): 273-292.
Statistics