Professional Documents
Culture Documents
Any use of trade, product, or rm names in this publication is for descriptive purposes only
and does not imply endorsement by the U.S. Government
Reston, Virginia
Abstract .............. 1
Introductionwhy study debris ows? ................................................................................................ 1
Debris ows in Southwest China .................................................................................................. 4
The great rivers of China and erosion of the landscape ................................................................. 6
What can we learn from the video?..................................................................................................... 6
Flows of water and sediment comprise a spectrum of increasing sediment content from
oods, to hyperconcentrated ows, to debris ows ................................................................... 6
Debris ows can move huge boulders ........................................................................................... 8
Forming rivers of rock................................................................................................................... 8
Debris ows are faster than ood waves ........................................................................................ 8
Flow geometrydepth and widthand channel slope ................................................................. 10
How much sediment is moved by individual surges and by complete series? ................................. 10
Why debris ows can crush man-made structurespowerful impact forces .................................... 10
Like wet cement............................................................................................................................ 10
Why multiple surges? .................................................................................................................... 11
Ballistic clast ejectionestimating ow velocity from runup height ................................................ 11
Estimating cross sections of debris ows ....................................................................................... 11
Knickpoints .. 12
How and why we seek to identify deposits of old debris ows ....................................................... 12
Can people be warned that a debris ow is coming? ...................................................................... 14
Mitigation and countermeasures ................................................................................................... 17
Acknowledgments .............................................................................................................................. 17
References cited, and for further study ............................................................................................... 18
FIGURES
1. Index map showing location of Jiangjia Ravine ........................................................................ 2
2. Photo showing video site in Jiangjia Ravine ............................................................................. 3
3. Photo of surge front on June 15, 1990 ..................................................................................... 5
4. Photo of debris ow deposits in Jiangjia Ravine, showing debris cones on valley-side
slopes formed by mass ow ................................................................................................... 9
5. Photo of debris ow in channel, incorporating mass of material. .............................................. 9
6. Photo of longitudinal roll waves and concave surface of right-to-left ow in bend
on June 15, 1990 .................................................................................................................. 12
iv Debris FlowsGeologic Process and HazardSurge Sequence at Jiangjia Ravine, Yunnan, China
TABLES
1. Annual sediment yields of debris ows in Jiangjia Ravine ......................................................... 4
2. Data for surges seen in video/DVD .......................................................................................... 7
APPENDIX
Narration ...... 23
CONVERSION FACTORS
Multiply By Obtain
1
Institute of Mountain Hazards and Environment, Chengdu, China
2
o o o o
o 75 90 105 120
100 o
105
o 45o
110
M in
BEIJING
Dadu Rive
r
Qinghai-Tibet
River
Jial i n g Ri
SICHUAN
ver
Fu R i v er
Plateau River
NT o
30
e
O Chengdu
Wuhan
gtz
FR
Chengdu
an
o
U
Y
o 30
30 Kunming
Tu o
EA
Min
R Guangzhou
0 500 Miles Nanning ea S
iver
Ri
0 500 Kilometers h i na
PLAT
S outh C
ve
r o
15
JIANGJIA RAVINE
oo
J i nsh
103 10'
a
r
ive
Xia
Rive r
Mekong
ojian g
J insha R
o
26 20'
River
R iv
e
Xia ojiang River
o
r Observation site
25 Kunming
0 2 Kilometers
o Jiangia
26 10'
Ravine
YUNNAN
0 500 Miles
0 500 Kilometers
Debris FlowsGeologic Process and HazardSurge Sequence at Jiangjia Ravine, Yunnan, China
Figure 2. Photo showing video site (arrow) in Jiangjia Ravine and location of the Dongchuan Debris Flow Research and
Observation Station (DDFROS). Note rice eld development on recent debris ow deposits, upper left. Photo by R.J.
Janda.
4 Debris FlowsGeologic Process and HazardSurge Sequence at Jiangjia Ravine, Yunnan, China
Table 1. Annual sediment yields of debris ows in Jiangjia Ravine. Data from Dongchuan Debris Flow
Observation and Research Station. Total sediment discharges estimated for 1965 to 1981 range from 820,000 in
1971 to 3,530,000 m3 in 1965 (Wang and others, 2000).
Debris Flows in Southwest China their deposits remain to reduce ood conveyance
Debris ows are a widespread means of in channels and the water-storage capacity in
sediment transport in the mountainous and the regions reservoirs. More than 360 small-to-
semiarid parts of Southwest China (Du and medium-scale hydroelectric stations and more
others, 1995; Ou, 1999). In recent years, the than 50 reservoirs were destroyed by 1991, and at
annual economic loss from debris ows has least 1,000 reservoirs had lost signicant capacity
exceeded 2 billion yuan (US$ 235 million), and primarily as a result of debris ows [IMHE
the annual loss of life from debris ows and Institute of Mountain Disasters (renamed
landslides, many of which are actually debris Hazards) and Environment, 1991].
ows, is estimated at between 500 and 1,000. Jiangjia Ravine is tributary to the Xiaojiang
Facilities periodically damaged by debris ows River (g. 1) which has 106 other tributaries
include 150 cities, 1,200 towns and villages, that yield debris ows. The watershed of the
36 railway lines, 50,000 km (31,000 mi) of Xiaojiang River is known in China as the
highways, 120,000 ha (300,000 ac) of farmland, Worlds Natural Museum of Debris Flows (Cui
and large numbers of mines, secondary roads, and others, 1999a). That description is not an
and hydroelectric facilities. In addition to the overstatement. Debris ows would be common
death and destruction from moving debris ows, in the Xiaojiang watershed under natural
Introduction Why Study Debris Flows? 5
conditions, but they are larger and more frequent as he designed it, utilizing the natural ux of
in part because of progressive deforestation that coarse sediment toward the inside of a large bend
began during the Tang Dynasty (A.D. 618-907) to separate it from ow with low concentrations
and intensied during the last 300-400 years of sediment toward the outside of the bend.
for the smelting of copper from local mines Relatively clear water is then decanted from
(Wieczorek and others, 1987). the outside of the bend through a bedrock weir
The Xiaojiang is the main southern tributary (named Neck of Precious Bottle) into low-
of the Jinsha River (the fabled River of the gradient canals that irrigate the Sichuan Basin,
Golden Sands in Chinese literature), which joins the so-called Granary or Rice Basket of China.
the Min River to form the upper Yangtze River,
here known as the Chang Jiang (Long River).
The Great Rivers of China and Erosion of the
Northwest of the provincial capital of Chengdu
(of Sichuan, north of Yunnan; g. 1), the Min
Landscape
River descends from the Qinghai-Tibet Plateau, The two large rivers of Chinathe Yellow
an area of high relief and frequent debris ows, to River of the north and the Yangtze of the
disgorge sediment as coarse as boulders (particles southare strikingly different. The Yellow
with diameters > 256 mm, or about 10 in). At River transports approximately 3 times as much
this point, famed engineer Li Bing designed sediment as the Yangtze on average, whereas
the rst large-scale, hydraulically sophisticated the Yangtze discharges approximately 20 times
system for ood control and irrigation in the 3rd as much water as the Yellow River (Jordaan,
century BC. The system still functions exactly 1992). Sediment yields and erosion are intense
Figure 3. Photo of surge front on June 15, 1990. Note spray of ejected particles from surge front, especially at right
margin (left bank). Width of surge is 20 m (66 ft). Photo by K.M. Scott.
6 Debris FlowsGeologic Process and HazardSurge Sequence at Jiangjia Ravine, Yunnan, China
throughout most of the Yellow River watershed, The characteristics and dynamics of the
due in signicant part to widespread deposits of surges seen in the video were recorded by the
loess, or windblown silt. Slopes covered by or staff of the Dongchuan Debris Flow Observation
cut into loess readily fail, and the disaggregated and Research Station (DDFORS), part of IMHE
material may move as a debris ow. In contrast, (Zhang and Xiong, 1997). Flow episodes consist
high sediment yields and erosion rates are mainly of as many as 100 or more surges, each separated
conned to the rugged headwaters of the Yangtze by less than one to tens of minutes. Each surge
watershed. There, the active mountain-building is a separate wave of moving rock fragments and
processes that have elevated the Qinghai-Tibet water. Table 2 includes data for all the surges in
Plateau have formed a steep and unstable the video, to and including surge 29, the second
landscape where debris ows are common. For of 2 surges in the series that were sampled for
example, the annual sediment yield (volume of sediment content. Complete data for the 77-
sediment only) from Jiangjia Ravine is typically surge series appear by surge number in DDFORS
from 14 million m3 (35-140 million ft3). records (Zhang and Xiong, 1997, p. 105-106).
Table 1 summarizes the annual sediment yields Figure 2 shows the DDFROS facilities and
and the numbers of ow episodes at Jiangjia the location at which the video was made. A
Ravine from 1982 to 2001. The maximum complete narration of the video, for reference by
recorded annual yield is 6.6 million m3 educators and keyed by surge number, is included
(230 million ft3), which corresponds to an in the appendix.
astounding denudation rateequivalent to
lowering the entire area drained by Jiangjia
Ravine in a single year, by over 13.6 cm (more Flows of Water and Rock Particles Comprise
than 5 in)! a Spectrum of Increasing Sediment Content
from Floods, to Hyperconcentrated Flows, to
Debris Flows
WHAT CAN WE LEARN FROM THE VIDEO? Floods consist of water containing less
The video illustrates the features of debris than 20 percent sediment by volumene
ows that make them an important geologic sediment that moves in suspension and coarser
process and hazardtheir content of rock particles that roll and saltate along the bed
particles (sediment), high velocities, ability to (bedload). The properties of debris ows, in
move huge boulders, and other unique attributes which a slurry of sediment moves a subordinate
described under the following headings. We amount of entrained water, are less well known
recorded most of the rst 24 in a sequence of 77 but are the object of intense experimental study
rainfall-triggered surges lasting over 3 hours and (cf., Iverson, 1997). Debris ows typically
45 minutes in Jiangjia Ravine on July 9, 1990. contain about 65 percent sediment by volume,
The site is renown for similar ow episodes, and may be dened as containing at least
lasting between 30 minutes and over 10 hours, 60 percent sediment by volume (cf., Pierson
that have occurred here with summer rainstorms and Costa, 1987). Least well known are the
between 5 and 28 times annually and a total of properties of the ows between these two end
427 times from 1965 to 2001. Segments of this membershyperconcentrated ows, which
video appear in many television documentaries contain sediment in the range of 20 to 60 percent
on natural hazards, including those dealing with by volume, or 40 to 80 percent by weight. The
lahars. Except for their lack of coarse boulder occurrence and deposits of hyperconcentrated
snouts or fronts, explained below, the ows in ow are rarely recognized, but may actually
Jiangjia Ravine are typical, in appearance and be common in the steeplands of volcanic and
characteristics, of debris ows as they occur in
steep terrain around the world.
What Can We Learn from the Video? 7
Table 2. Data for surges seen and sequential with those seen in the video. Data for all 77 surges in series from
Zhang and Xiong, (1997, p. 105-106). Surges 13 and 29 were sampled (data in gure 8).
semiarid terrains (cf., Pierson and Scott, 1985; low that access and direct observations are not
Scott, 1988). prudent (and were not permitted in 1990).
Surge 1 in the video is a hyperconcentrated Nevertheless, our observations during clear
ow (data in table 2). Note that the high weather suggest that the source of the ows is the
sediment content causes it to resemble dirty, gradual mobilization or ushing of the reservoir
turbulent motor oil in appearance. The body of of deposits, during storm runoff, that have
surge 2 is a debris ow, but note that it is pushing accumulated in the channel network since the last
hyperconcentrated ow that has remained in the ow episode. The surges generally do not form
channel ahead of the surge peak. Some of the by progressive upstream slope failures, as observed
ows in the video by Costa and Williams (1984) in some areas of recurrent debris ow.
have the same appearance and behavior and are
probably also hyperconcentrated ows.
Debris Flows are Faster than Flood Waves
At similar discharges, slurries of debris
Debris Flows Can Move Huge Boulders ow travel about twice as fast as water ows.
Debris ows are widely cited as capable of Velocities of the surges typically range from
moving huge boulders. In fact, most of the several m/s to 10 m/s, about 10 to 33 ft/s or 7
debris ows in the western United States and to over 20 mphthe range of those we see in the
other parts of the world have a characteristic video. Velocities of most debris ows fall in this
snout or front of boulders, the concentration of range. The velocity of the fastest surge in the
which may act like a moving dam. Examples of video (surge 8) is 9.3 m/s (table 2), equivalent
ows with boulder-dominated fronts are recorded to 30.5 ft/s or 20.8 mph. The maximum surge
in the video by Costa and Williams (1984). This velocity recorded at the site is 18.18 m/s (July 9,
common feature is not present in the debris 1998)over 40 mph! The values are based on the
ows in our video, only because little sediment as travel times, measured by stopwatch, of the surge
coarse as boulders is present at Jiangjia Ravine fronts in the straight, 200-m reach upstream from
the bedrock readily disaggregates to small the observation point (g. 2). In the calculations
fragments. The boulder seen rolling in the ow of ow and sediment discharge, discussed below,
between surges 6 and 7 is near the maximum it is assumed that the debris ow surges are
size recorded at the site, over 1 m (about 4 ft) in translatory waves of moving material rather than
maximum dimension. of formthat is, that the particles in the wave are
moving at the speed of the wave, as opposed to
oscillatory surges in which the wave form moves
Forming Rivers of Rock faster than the material within it [see the video
The Jiangia Ravine watershed is underlain of Costa and Williams (1984) for an example of
by highly fractured, metasedimentary a wave form moving through nearly stationary
and volcaniclastic siltstone and shale that sediment]. The dominantly translatory nature
disaggregates directly to pebble-size clasts. The of the surges at Jiangjia Ravine, where sediment
smaller particles are predominantly elongate, moves with the ows, is clear in the video.
or prismoidal, in shape. This material is We designate surge velocity in the video as c,
continuously supplied to the main channels from celeritythe velocity of a wave form independent
hill- and valley-side slopes, by small landslides, of sediment movement.
dry sliding, and in the summer by wet creep at The channel retains a signicant volume of
a visibly obvious rate (g. 4). Bank erosion of ow between the surges, and this material may
previous debris ow deposits (g. 5) adds to the cease to ow, as seen in the video, or it may ow
surge volumes (ow bulking). During the storms continuously. Note the stationary material in
that trigger the ows, slope stability throughout the channel as it is overrun by all of the surges
the steep source areas of Jiangjia Ravine is so following surge 3.
What Can We Learn from the Video? 9
Figure 4. Photo of tilled debris ow deposits in Jiangjia Ravine (note texture in rice-eld levee at left)
and debris cones of sediment owing from the steep valley-side slopes. Note patterns on cone surfaces,
indicating mass ow, in some instances at visibly detectable rates during summer dry periods. Photo by
K.M. Scott
Figure 5. Photo of debris ow in channel, incorporating mass of material derived from bank failure
(arrow). Note texture of deposit surface, in cutbank and on surface at lower left. Photo by K.M. Scott.
10 Debris FlowsGeologic Process and HazardSurge Sequence at Jiangjia Ravine, Yunnan, China
Flow GeometryDepth and Widthand as dense as pure water (as well as owing at twice
Channel Slope the velocity, as noted above). The unit weight
The heights of individual surges typically of the solid sediment at Jiangjia Ravine is taken
range from 12 m (3.36.6 ft). The largest surge uniformly as 2.65 t/m3 (a density of 2.65).
in the video (surge 8) is 2.83 m (9.3 ft) high.
The highest peak recorded at the site through
Like Wet Cement
2001 was 5.5 m (18 ft). The ows range in
width from 20.0 m (66 ft), the width of the Debris ow behavior is strongly inuenced
channel bottom, to 27.0 m (89 ft) for surge 8. by a highly viscous, nearly incompressible pore
Slope of the reach at the site is 0.06. uid composed of water with suspended ne
sediment (Iverson, 1997). Although the property
of viscosity does not strictly apply to granular
How Much Flow and Sediment are Moved by mixtures, like most debris ows, viscosity has
the Surges at Jiangjia Ravine? been measured or estimated for some ows. The
The largest peak discharge of any surge in the viscosities of debris ows themselves are reported
video (table 2; no. 8) is 710.6 m3/s (25,000 cfs). at over 1,000 times that of pure water at 20
The largest peak discharge measured for any surge (Pierson, 1980). Viscosity of 2030 poises is
at Jiangjia Ravine through 2001 is 2,913.5 m3/s, reported by Li and others (1983) for the ows
or 103,000 cfs (July 9, 1998). Total sediment at Jiangjia Ravine. That range compares with
yield (the volume of sediment discharged from viscosities reported for mayonnaise, 6.3 poises,
the watershed) of the 77 surges, including the and wet cement, 24 poises (see table 2 of Costa,
hyperconcentrated ows, is 243,000 m3 (over 1984).
8 million ft3 ). The most sediment discharged by
any single ow series is 2 million m3, or about 70 Why Multiple Surges?
million ft3 (Ou, 1999, p. 37). The sediment-yield
volumes are derived from the total discharges Prolonged episodes of debris ow commonly
of the surges and their measured or estimated occur as a series of separate surges, both at
sediment concentrations. Jiangjia Ravine and elsewhere (see the video of
Costa and Williams, 1984). Explanations include
Maximum sediment transport rates for ow instability, which causes uniform ow to
each surge are not listed in Table 2, but values evolve into a series of waves (Davies, 1986; Liu
range between 29 t/s (surge 1) and 1,370 t/s and Mei, 1990). Based on research at Jiangjia
(surge 8)the largest transport rate for any of the Ravine in the month following the surges in our
77 surges in the series. video, Davies and others (1991, 1992) observed
that surges form with the continuous inow of
Why Debris Flows Destroy Manmade material from upstream into downstream, wider,
StructuresPowerful Impact Forces and less-steep reaches that already contain a large
volume of static debris (such as the 200-m reach
The ability of debris ows to crush many upstream from the video site; g. 2). Discharge
structures results from their high impact forces of a surge from those reaches is triggered
they are literally rivers of owing rock fragments periodically at some critical point, temporarily
with interstitial water. The maximum unit depleting them, and setting the stage for another
weight of a surge in the video is 2.26 t/m3 (surges cycle of continuous upstream inow that will
13 to 22), compared to 1.0 for pure water, only again trigger episodic outow. Between surges,
slightly less than the maximum recorded value at ow commonly ceases entirely, as we see in the
the site of 2.37 t/m3, corresponding to 81 percent video.
sediment by volume (Wang, 1989, p. 237; Ou,
1999, p. 37). Thus the ows are more than twice
What Can We Learn from the Video? 11
Ballistic Clast EjectionEstimating Flow ows were obtained from bends, when compared
Velocities from Runup Heights with corresponding values for the same ow in
The fronts of the surges contain clasts as straight reaches. Compared to both upstream
coarse as small boulders, but the dominant size and downstream sites, the cross-sectional areas
class is the pebble (4-64 mm) fraction. Grain- at bends were 1.3 to 3.6 times larger (Webb
to-grain collisions are intense, and pebbles and others, 1989). In each location, the ow
are constantly ejected from the moving front, surface and thus the cross-sectional area was
at times as a spray of material (g. 3). The reconstructed by connecting the peak ow levels
spray appears to be mud, but it is mainly if not as indicated by deposits on both sides of the
entirely mud-coated particles, based on the direct channel with a straight line. This large difference
observations of the authors during this ow in discharges could not be explained, but one
series. Ejection is intense when ow impacts suggestion was the presence of static material in
the channel side or a barrier at a high angle, and curves. The video indicates the probable cause
cataclasis (breaking and crushing) of the pebbles of the discrepancy in discharge valuesthe ow
occurs. The ballistically ejected, mud-coated surface in bends is markedly concave (g. 6).
clasts yield mud-splatter deposits that extend This conclusion corresponds with observations of
above those that are formed as the ows run up bend-surface concavity during ume experiments
an obstacle or superelevate against the outside (Iverson and others, 1994). Figure 7 shows the
of a channel bend. Thus, based on deposits interpretations (assuming straight ow surfaces)
after the fact, applications of the energy-balance of the cross sectional area in a case where the
equation (assuming conversion of kinetic to discharge obtained from a bend is twice that of a
potential energy) at this site could yield velocities discharge from a straight reach.
that are too large. The distance (h) a ow runs
up an obstacle is related to velocity (c) by the Knickpoints
energy-balance relation, c = [2gh]1/2, where g = Knickpoints are vertical, transverse falls in
gravitational acceleration (cf., Pierson, 1985). channel elevation that are scoured by erosion and
However, in a large-scale ume, Iverson and which commonly migrate upstream. Knickpoints
others (1994) found that the measured velocities as much as 11.2 m in height form during
of moving debris ow exceeded those obtained the surge series at Jiangjia Ravine. They were
from application of the energy-balance equation observed but not photographed during the late
to the runup height (where not affected by mud stages of the ows of July 9, 1990. As described
splatter) by about 30 percent. by Davies and others (1991 and 1992), they
are probably associated with channel incision.
Estimating Cross Sections of Debris Flows However, during the series in the video and a
previous series on June 15, 1990 (g. 3), they
Instantaneous debris ow discharges are
were transient features developed in deposits
commonly estimated by obtaining velocities
only temporarily in the channel, and they did
from runup as described above, as well as by the
not result in any notable change in mean bed
superelevation of the ow surface (h) in a bend
elevation.
due to centrifugal force (cf., Pierson, 1985).
Velocity (c) is related to h by the relation,
c = [ghrc/b]1/2, where rc = centerline radius of Why and How Do We Seek to Identify the
curvature, and b = channel width. The velocity Deposits of Old Debris ows?
estimate is then multiplied by the cross sectional Debris ow occurrence is rarely as frequent
area to yield the rate or discharge of ow. and predictable as at Jiangjia Ravine. Elsewhere,
Webb and others (1989) found that potential ow hazards may be evident only
unrealistically high discharge values of debris from the deposits of previous ows. How can
12 Debris FlowsGeologic Process and HazardSurge Sequence at Jiangjia Ravine, Yunnan, China
Figure 6. Photo of longitudinal roll waves and concave surface of right-to-left ow in bend on June 15, 1990. For
scale, this bend is visible in the videofor example during surge 10. Observer in left foreground does not indicate
scale. Photo by Kang Zhicheng
we recognize these deposits, and thus their past In addition, the ne fractions of the two
as well as the possible or even probable future samples were not analyzed. Otherwise, the
occurrence of debris ows? distributions in B and C would be similar to
1. They have a ne-grained, commonly muddy that of A.
matrix. Debris ows are commonly called The amount of silt- and clay-size material is
mudows because of their content of ne useful for determining debris ow origin, and
sediment [mud = the total of silt-size sediment samples D and E of gure 8 are deposits that
(0.0040.0625 mm) and clay-size sediment are cohesive (or muddy), and noncohesive
(< 0.004 mm; < 0.005 mm in Chinese data)]. (or granular), respectively. These two
The ows and deposits at Jiangjia Ravine textural subpopulations help us distinguish
typically contain 2 to 5, and as much as 10 lahars of landslide and meltwater origins,
percent clay-size sediment by weight (Kang, respectively, at volcanoes (Scott and others,
1989). Figure 8 shows histograms of sample 2001). However, facies of lahars with mainly
A, a complete analysis of all size fractions in cohesive deposits may have very little ne
a ow in Jiangjia Ravine; B and C are the sediment, as in the boulder-rich bars of the
partial analyses of surges 13 and 29. The channel faces of the 1980 lahar in the North
samples of moving ows were taken with a Fork Toutle River (Scott, 1988, g.15).
suspended sampler with a restricted orice, R.M. Iverson (written commun., 2002) has
thereby excluding the coarsest fractions. noted formation of a nes-free levee facies at
What Can We Learn from the Video? 13
A C
A' C'
B'
PLAN VIEW OF CHANNEL BED
B1 B1'
A A' C C'
Cross-sectional
area = 20 m2
Cross-sectional
area = 10 m2
Figure 7. Diagrammatic cross-sections of ow illustrating the differences in cross-sectional area (and thus,
discharge) as they depend on reconstruction of the ow surface from deposits. Case B1B1: assumption of
straight line connecting highest deposits on each side of channel; case B2B2: observed concave surface at Jiangjia
Ravine, yielding cross-sectional area in accord with those from upstream and downstream cross-sections AA and
CC.
14 Debris FlowsGeologic Process and HazardSurge Sequence at Jiangjia Ravine, Yunnan, China
the large-scale ume noted above. Typical 4. They are commonly massive (without
alpine debris ows that move with boulder- stratication). The multiple surges at
rich fronts, as noted above in contrast to the Jiangjia Ravine, where they spread over a
ows at Jiangjia Ravine, also leave lateral downstream ood plain (g. 11), yield single
levees dominated by boulder-size clasts. massive, matrix-supported units, analogous
2. They are commonly bimodal (two peaks in to the successive surges of material from a
histograms of sediment size). For example, large debris ow ume (Major, 1997). The
sample D in gure 8 has a primary mode series of texturally and compositionally
or peak in a cobble-size fraction of the size homogeneous surges yield massive composite
distributionbetween 128 and 256 mm, and units without obvious stratigraphic
a secondary mode in a sand fractionbetween breaks, indicating a need for caution when
0.125 and 0.25 mm. The coarse mode interpreting ow magnitudes and frequencies
represents particles in the dispersed phase; from stratigraphy, particularly in alluvial fan
the ner mode is sediment in the muddy environments.
matrix or continuous phase. The coarse Because the stratigraphic units are commonly
mode is like the plums dispersed throughout a of composite construction, and therefore
plum pudding. With the other features listed relatively uniform, graded bedding is poorly
here, we can identify debris ow deposits developed. Wang and others (1999) note the
from this bimodal, plum-pudding texture. occurrence of normal grading (upward ning) in
3. They contain a large range of sediment sizes; some units.
that is, they are poorly sorted (high values of
G; g. 9). In gure 9, several of the samples
Can People Be Warned That a Debris Flow is
in gure 8 are plotted as cumulative curves
Coming?
on a graph with a probability ordinate, on
which normal distributions (a bell-shaped Warning times of 20-45 minutes at the
distribution in a histogram) plot as straight observation site are based on threshold values of
lines. Note that, excluding the primary mode, rainfall intensity and antecedent precipitation
sample D is close to a normal distribution, as (Ou, 1999). For predictive purposes, rainfall data
seen both in the histogram in gure 8 and the are telemetered from recording stations high in
nearly straight line in gure 9. the watershed. The ows in the video occurred
with little precipitation at the observation site.
The fewer the fractions in a sediment
sample, the more steeply the line representing Various devices for event warnings (ow
their cumulative distribution is sloped and the is occurring) have been tested at this location
better the sorting (low values of G; g. 9). (e.g., Kang, 1990). They include Acoustic
Note the better sorting of ood deposits Flow Monitors (AFMs)solar-powered,
represented by samples G and H. Sample microprocessor-based eld computers linked
G is a deposit of coarse gravel resulting from to exploration-model geophones that detect
bedload transport in ood ow, and sample the specic frequency of moving debris ows
H is the deposit of a sand layer deposited (LaHusen, 1998). Arrays of AFMs are widely
by a ood. For comparison, with both the used for detection of lahars. In addition, lives
poorer sorting produced by debris ows and can be saved by educating the residents of
the better sorting of ood deposits, sample vulnerable areaslike the many towns in the
F is the deposit of a hyperconcentrated ow. valleys of mountainous southern Chinato
Figure 10 shows the deposit of a debris ow recognize ground vibrations as the possible signal
overlying that of a ood or hyperconcentrated of an approaching debris ow, and to go to high
ow. ground immediately. With this knowledge
Education for Self Warning and Evacuation
What Can We Learn from the Video? 15
30
20
PERCENT
20
10
finer than
10
0 0.625 mm
40 0
B. Surge 13
30
PERCENT
20 FLOOD DEPOSITS
finer than 0.15 mm
Mount St. Helens Toutle River
10 40
G. Flood gravel
0 30
PERCENT
40
20
C. Surge 29
30
PERCENT
10
finer than 0.15 mm
20
0
10 50
H. Flood sand
0 40
30
Mount St. Helens Toutle River
PERCENT
40
20
D. Cohesive deposit
30 North Fork lahar 10
PERCENT
20
0
1024 512 1 0.031 0.001 mm
10
-10 -5 0 5 10
0
40
E. Noncohesive deposit = log2 mm
30 South Fork lahar
PERCENT
20
10
0
1024 512 1 0.031 0.001 mm
-10 -5 0 5 10
Figure 8. Texture (particle size distribution) of debris ow deposits, shown as histograms. A, Debris ow at
Jiangjia Ravine (Li and others, 1983); B, Surge 13; C. Surge 29; D, Cohesive (muddy) debris ow deposit in North
Fork Toutle Riverpeak ow facies of 1980 lahar from Mount St. Helens (g. 17, Scott, 1988); E, Noncohesive
(granular) debris ow deposit in South Fork Toutle River1980 lahar 35.6 km downstream from Mount St.
Helens (g. 20, Scott, 1988); F, Lahar-runout ow, 1982 in Toutle River (g. 41, Scott, 1988); G, Fluvial gravel
representing bedload in Toutle River (pre-lahar alluvium, g. 22, Scott, 1988); H, Fluvial sand in 1980-81 ood
berm of Toutle River (g. 41, Scott, 1988).
16 Debris FlowsGeologic Process and HazardSurge Sequence at Jiangjia Ravine, Yunnan, China
0.1
0.2
0.5 EXPLANATION
1
2 Debris flow
5 Hyperconcentrated flow
Flood flow
CUMULATIVE PERCENT
10
20
30
40
50 E
60 G = 4.40
70
80
90
95 D
A
98 G = 4.85 G = 3.70
99 F H
G = 1.15 G = 0.75
99.8
99.9
-6 -5 -4 -3 -2 -1 0 1 2 3 4 5 6 7 8 9 phi
64 32 16 8 4 2 1 0.5 0.25 0.125 0.0625 0.031 0.016 0.008 0.004 0.002 mm
SEDIMENT SIZE
Figure 9. Selected data in gure 8 plotted as cumulative curves of the size distributions.
Values of sorting () as the graphic standard deviation of Folk (1980),
G = (84 - 16)/2. values dened as in Figure 8.
Figure 11. Photo of downstream reach of Jiangjia Ravine where debris ow surges spread and deposit lateral
levees. Photo by J.J. Major.
18 Debris FlowsGeologic Process and HazardSurge Sequence at Jiangjia Ravine, Yunnan, China
REFERENCES CITED, AND FOR FUTHER Du, R., Li, H., Tang, B., and Zhang, Z., 1995, Debris ow studies in
STUDY(*) China in the last 30 years: Journal of Chinese Geography, v. 5,
no. 4, p. 90-102.
(For Chinese authors, surnames are followed by given
Folk, R.L., 1980, Petrology of sedimentary rocks: Austin, Texas,
names) Hemphill Publishing Company, 182 p.
Institute of Mountain Disasters and Environment, 1991, A manual
Chen, J., 1989, A study on debris ow warning in
of the distribution of debris ow disaster and its dangerous
China: Proceedings of Japan-China Symposium on
regionalization map in China: 1:6,000,000 scale map and
Landslides and Debris Flows, Tokyo, p. 177-181.
accompanying report, 8 p. [presently, Institute of Mountain
*Costa, J.E., 1984, Physical geomorphology of debris ows, Hazards and Environment]
in Costa, J.E. and Fleisher, P.J., eds., Development and
*Iverson, R.M., 1997, Physics of debris ows: Reviews of
Applications of Geomorphology: Berlin, Springer-Verlag,
Geophysics, v. 35, p. 245-296.
p. 268-317.
Iverson, R.M., LaHusen, R.G., Major, J.J., and Zimmerman, C.L.,
*Costa, J.E., and Williams, G.R., 1984, Debris ow dynamics: U.S.
1994, Debris ow against obstacles and bends: Dynamics and
Geological Survey Open-File Report 84-606 (a video).
deposits (abs.): EOS, v. 75, no. 44, p. 274.
Cui, P., 1992, Study on conditions and mechanics of debris ow
*Iverson, R.M.., Reid, M.E., and LaHusen, R.G., 1997, Debris-ow
initiaion by means of experiment: Chinese Science Bulletin, v.
mobilization from landslides: Annual Review of Earth and
37, no . 9, p. 759-763.
Planetary sciences, v. 25, p. 85-138.
*Cui, P., 1999, Impact of debris ow on river channels in the
Jan, C.D., Wang, Y.Y., and Han, W.L., 2000, Resistance reduction
upper reaches of the Yangtze River: International Journal of
of debris-ow due to air entrainment [at Jiangjia Ravine]:
Sediment Research, v. 14, p. 201-203.
in Wieczorek, G.F., Naeser, N.D., eds., Debris-Flow Hazards
Cui, P., Thornes, J., and Guan, J., 1996, International Mitigation: Mechanics, Prediction, and Assessment,
Symposium Interpravent 1996Garnisch-Partenkirchen, Proceedings of the Second International Conference on
Tagungspublikation, Bd. 3, p. 1-3-112. Debris-Flow Hazards Mitigation, Taipei, Taiwan, p. 369-372.
*Cui, P., Wei, F., and Li, Y., 1999a, Sediment transported by debris Jordaan, J.M., 1992, Railway debris ow and ood disaster
ows to the lower Jinsha River: International Journal of protection works in Mainland China: Die Siviele Engenieur in
Sediment Research, v. 14, p. 67-71. Suid Afrika, November, Technical Note, p. 367-377.
Cui, P., Liu, S., Tang, W., and Chen, Y., 1999b, Debris ow Kang, Z., 1989, A mechanical analysis on the occurrence of debris
monitoring and forecasting in China, progress and prospect: ow: Proceedings of Japan-China Symposium on Landslides
1999 Annual Report Dongchuan Debris Flow Observation and and Debris Flows, Tokyo, p. 206-208.
Research Station, p. 17-23.
Kang, Z., 1990, Motion characteristics of debris ow at Jinagjia
Cui, P., Liu, S., and Tan, W., 2000, Progress of debris ow forecast Gulley, Yunnan Province, China: Beijing, International
in China (in Chinese): Journal of Natural Disasters, v. 9, no. 2, Research and Training Center on Erosion and Sedimentation,
unpaginated. 38 p.
*Davies, T.R.H., 1986, Large debris owsa macroviscous *LaHusen, R.G., 1998, Detecting debris ows using ground
phenomenon: Acta Mechanica, v. 63, p. 161-178. vibrations: U.S. Geological Survey Fact-Sheet FS-236-96, 2 p.
Davies, T.R.H., Phillips, C.J., Pearce, A.J., and Zhang, X.B., 1991, Li, J., Yuan, J., Bi, C., and Luo, D., 1983, The main features of the
New aspects of debris ow behavior: Proceedings, Japan-U.S. mudow in Jiang-Jia Ravine: Zeitschrift fur Geomorphologie,
Symposium on Snow Avalanches, Landslides and Debris Flow N.F., Bd. 27, p. 325-341.
Prediction and Control.
Liu, K.F., and Mei, C.C., 1990, Waves in a uid mud layer owing
Davies, T.R.H., Phillips, C.J., Pearce, A.J., and Zhang, X.B., 1992, down an incline: Proceedings, International Symposium on
Debris ow behavioran integrated overview: Proceedings of the Hydraulics/Hydrology of Arid Lands, Hydraulics Division
Symposium on Erosion, Debris Flows, and Environment in of the American Society of Civil Engineers, San Diego, p. 403-
Mountain Regions, IASH Pub. 209, p. 217-225. 409.
Dongchuan Institute of Debris Flow Control, 1987, Debris ow of *Major, J.J., 1997, Depositional processes in large-scale debris-ow
Xiaojiang River basin in photographs: Sichuan Publishing experiments: Journal of Geology, v. 105, p. 345-366.
House of Science and Technology, 37 p.
Ou, G., 1999, Sediment hazards and counter measures in China:
Du R., 1989, Debris ows and their countermeasures in China: 1999 Annual Report, DDFROS, p. 35-39.
Proceedings of Japan-China Symposium on Landslides and
Debris Flows, Tokyo, p. 191-194.
References 19
Pierson, T.C., 1980, Erosion and deposition by debris ows at Wang, Y., 1989b, An approach to rheological models for debris
Mt. Thomas, North Canterbury, New Zealand: Earth Surface ow research: Proceedings, Fourth International Symposium
Processes, v. 5, p. 227-247. on River Sedimentation in Beijing, p. 714-721.
Pierson, T.C., 1985, Iniation and behavior of the 1980 Pine Creek Wang, Z.Y, Wai, O.W.H., and Cui, P., 1999, Field investigation into
and Muddy River lahars, Mount St. Helens, Washington: debris ow mechanism: International Journal of Sediment
Geological Society of America Bulletin, v. 96, p. 1056-1069. Research, v. 14, p.10-22.
*Pierson, T.C., and Scott, K.M., 1985, Downstream dilution of Wang, Y., Jan, C, and Yan, B., 2000, Debris ow structure and
a lahar: transition from debris ow to hyperconcentrated rheology: National Natural Science Foundation of China, 216
streamow: Water Resources Research, v. 21, p. 1511-1524. p. In Chinese.
*Scott, K.M., 1988, Origins, behavior, and sedimentology of lahars *Webb, R.H., Pringle, P.T., and Rink, G.R., 1989, Debris ows
and lahar-runout ows in the Toutle-Cowlitz River System: from tributaries of the Colorado River, Grand Canyon National
U.S. Geological Survey Professional Paper 1447-A, 74 p. Park, Arizona: U.S. Geological Survey Professional Paper 1492,
*Scott, K.M., Macas, J.L., Naranjo, J.A., Rodriguez, S., and 39 p.
McGeehin, J.P., 2001, Catastrophic debris ows transformed Wieczorek, G.F., Wu, J., and Li, T., 1987, Deforestation and
from landslides in volcanic terrainsMobility, hazard landslides in Yunnan, China: International Erosion Control
assessment, and mitigation strategies: U.S. Geological Survey Association, Proceedings of Conference 18, Reno, Nevada;
Professional Paper 1630, 59 p. February, 1987, p. 183-196.
*Takahashi, T., 1991, Debris ows: Balkema, Rotterdam, 165 p. Zhang, J., and Xiong, G., 1997, Data collection of kinematic
Wang, Y., 1989a, The main static characteristics of debris ow observation of debris ows in Jiangjia Ravine, Dongchuan,
in the southwestern mountain areas of China: Proceedings Yunnan (1987-1994): Beijing, Science Press, 258 p. In
of Japan-China Symposium on Landslides and Debris Flows, Chinese.
Tokyo, p. 237-239.
APPENDIX
Appendix 23
APPENDIX
Narration. A. During introductory black-and-white lm segment showing ows and sampling at Jiangjia
Ravine.
B. During remainder of video/DVD.
A. Debris ows are slurries of rock and water that ow like wet concrete. They are an important way that
sediment is transported in mountainous areas. Debris ows are also a signicant geologic hazard.
Each summer, in arid and mountainous SW China, debris ows are triggered by rainstorms. Each
episode may consist of more than 100 separate waves, or surges, lasting over several hours. In the 1960s,
sampling procedures were primitive (narrated as technician runs from surge front).
B.
SURGE NUMBER
AND VEL0CITY (c)
IN M/S NARRATION WITH SURGES, BY NUMBER
2 The second surge, beginning as debris ow, pushes and mixes with the
hyperconcentrated ow already in the channel.
c = 6.78
3 As surge 3 leaves the mountain front, note the emerald green areas. These are
elds of rice and sweet potatoes cultivated on the deposits
c = 7.63 of larger debris ows during the 1980s.
The debris ow fronts are the deepest part of each surge.
The fronts contain the coarsest clasts in the ow.
At this site, surges occur at regular intervals, from one to several minutes. A
single series of surges may continue for over 10 hours.
B. (Continued)
SURGE NUMBER
AND VEL0CITY (c)
IN M/S NARRATION WITH SURGES, BY NUMBER
5 Note how fast the surges travelabout twice as fast as the same size of water wave.
We are seeing the ows at actual speed.
c = 8.16 Debris ows are more than twice as dense as water without sediment.
These two factorsvery dense ows moving very fastcreate the high impact
forces that enable debris ows to destroy many manmade structures.
7 [not photographed]
8 This is the surface of the largest and fastest surge in the entire series.
Velocity of the front is over 9 m/s, or 20 mi/h.
c = 9.30 Debris ow sediment, seen here between surges, consists of two phasescoarse
particles, mainly pebbles in these ows, that are dispersed and suspended in a
matrix of sand, silt and clay.
B. (Continued)
SURGE NUMBER
AND VEL0CITY (c)
IN M/S NARRATION WITH SURGES, BY NUMBER
11 In the lm from the 1960s at the beginning of this video, this channel had not
been cut, and ows spread across the entirely valley. That
c = 8.00 lm was made at this location.
12 Mud, the term for particles of silt and clay size, causes debris ows to be widely
known as mudows.
c = 8.00 Scientists prefer the term debris ow because mud is generally less than 25% of
most debris ows, and some have only a few percent mud.
The streaks in the ow, parallel to the banks, are lines of shear as ow approaches
the channel boundary.
13 Surge 13 has as high a unit weight as any in the series2.26 t/m3 vs. 1.0 for pure
water.
c = 8.89 That value is only slightly less than the maximum value recorded for any debris
ow at this site2.38 t/m3.
14 In contrast to the dangerous techniques of the 1960s, samples are now taken by
lowering the torpedo-like cylinder into the ow with a steel cable.
c = 8.89
B. (Continued)
SURGE NUMBER
AND VEL0CITY (c)
IN M/S NARRATION WITH SURGES, BY NUMBER
22 Debris ows have high impact forces that can destroy many structures and crush
most buildings.
c = 5.88 This is because they are more than twice as dense as water, and nearly twice as
fast as a surge of water the same size.