Document Actions

IJEGE-11_BS-Coe-et-alii

background image
Italian Journal of Engineering Geology and Environment - Book www.ijege.uniroma1.it © 2011 Casa Editrice Università La Sapienza
199
DOI: 10.4408/IJEGE.2011-03.B-024
ASSESSMENT OF TOPOGRAPHIC AND DRAINAGE NETWORK
CONTROLS ON DEBRIS-FLOW TRAVEL DISTANCE ALONG
THE WEST COAST OF THE UNITED STATES
J
effRey
A. COE
(*)
, m
aRk
E. REID
(**)
, d
ianne
L. BRIEN
(**)
& J
oHn
A. MICHAEL
(*)
(*)
U.S. Geological Survey, Denver Federal Center, MS 966, Denver, CO 80225 USA - Email:jcoe@usgs.gov)
(**)
U.S. Geological Survey, 345 Middlefield Rd., MS 910, Menlo Park, California 94025 USA
uting debris-flow initiation locations. Moreover, con-
tributing area and the proximity of the initiation loca-
tions to the drainage network both influenced travel
distances, but proximity to the drainage network was
the better predictor of travel distance. In both study
areas, flows that interacted with the drainage network
flowed significantly farther than those that did not. In
California, initiation sites within 60 m of the network
were likely to reach the network and generate long-
traveled flows; in Oregon, the threshold was 80 m.
K
ey
words
: rainfall, debris flow, travel distance, channel,
drainage, contributing area, curvature, initiation, entrain-
ment, LiDAR, California, Oregon
INTRODUCTION
Prediction of travel distance is a fundamental ele-
ment of debris-flow hazard assessments. Total volume
exerts a primary control over debris-flow travel dis-
tance and inundation area (e.g., i
veRson
et alii, 1998;
G
Riswold
& i
veRson
, 2007), and debris flows that
grow as they entrain sediment can generate exception-
ally large and destructive flows (H
unGR
et alii, 2005).
To help clarify controls on entrainment and travel
distance throughout the debris-flow prone west coast
of the U.S., we investigated debris flows triggered by
two major precipitation events in two separate geo-
graphic areas located about 700 km apart. The first area
is within the Coast Range of central Alameda County
in the San Francisco Bay region of northern Califor-
ABSTRACT
To better understand controls on debris-flow
entrainment and travel distance, we examined topo-
graphic and drainage network characteristics of ini-
tiation locations in two separate debris-flow prone
areas located 700 km apart along the west coast of
the U.S. One area was located in northern Califor-
nia, the other in southern Oregon. In both areas,
debris flows mobilized from slides during large
storms, but, when stratified by number of contrib-
uting initiation locations, median debris-flow travel
distances in Oregon were 5 to 8 times longer than
median distances in California. Debris flows in Ore-
gon readily entrained channel material; entrainment
in California was minimal. To elucidate this differ-
ence, we registered initiation locations to high-res-
olution airborne LiDAR, and then examined travel
distances with respect to values of slope, upslope
contributing area, planform curvature, distance from
initiation locations to the drainage network, and
number of initiation areas that contributed to flows.
Results show distinct differences in the topo-
graphic and drainage network characteristics of de-
bris-flow initiation locations between the two study
areas. Slope and planform curvature of initiation
locations (landslide headscarps), commonly used
to predict landslide-prone areas, were not useful for
predicting debris-flow travel distances. However, a
positive, power-law relation exists between median
debris-flow travel distance and the number of contrib-
background image
J.A. COE, M.E. REID, D.L. BRIEN & J.A. MICHAEL
200
5th International Conference on Debris-Flow Hazards Mitigation: Mechanics, Prediction and Assessment Padua, Italy - 14-17 June 2011
deRson
& s
itaR
, 1995; i
veRson
et alii, 2000; R
eid
et alii,
this volume; s
ilbeRt
et alii, 2002; m
C
C
oy
et alii, 2010).
However, for regional hazard evaluations, investigators
typically have little detailed data on spatially variable
material properties or expected flow dynamics, but they
often have reliable Digital Elevation Model (DEM) data,
as well as hydrologic drainage network data derived di-
rectly from DEM data. For this reason, forecasts based
on DEM data are appealing and often sought. For exam-
ple, shallow landslide initiation locations are commonly
predicted using topographic and drainage variables such
as local slope and upslope contributing area (m
ont
-
GomeRy
& d
ietRiCH
, 1994; t
aRboton
, 1997; d
ietRiCH
et
alii, 2001). This approach for identifying initiation sites
prompts the question: Do the topographic and drainage
network characteristics of initiation locations influence
the entrainment and travel distance potential of debris
flows generated from those sites?
In this paper, we investigate this question by exam-
ining the topographic and drainage network character-
istics of debris-flow initiation locations in the two study
areas using recently available, high resolution, Light
Detection And Ranging (LiDAR) topographic data.
Specifically, we examine whether values of slope, up-
slope contributing area, planform curvature, and prox-
imity to the drainage network at debris-flow initiation
locations influence resulting debris-flow travel distanc-
es. We also assess the effect that multiple numbers of
debris-flow initiation locations have on travel distance.
Based on the results of our assessment, we present two
empirical approaches that can be used to estimate travel
distance; one based on the number of contributing ini-
tiation locations, and the other based on the proximity
of initiation locations to the drainage network.
STUDY SITES
Both study sites are prone to precipitation triggered
debris flows. We analyzed topographic and drainage
network variables at the two study sites, but there are
also non-topographic variables, such as differing geo-
logic units and vegetation, that could potentially impact
travel distances. The following sections provide detailed
descriptions of the characteristics of each study area.
ALAMEDA COUNTY, NORTHERN CALIFORNIA
The Alameda County study area is about 125 km
2
in size and lies within the northwest-trending Coast
Ranges of northern California in the San Francisco
nia (Fig. 1). Here, we examined debris flows triggered
by February 1998 rainfall during the 1997/98 El Niño
Southern Oscillation, which caused widespread debris
flows and flooding throughout California. The second
area, located in the Coast Range of western Douglas
County in southern Oregon (Fig. 1), contains debris
flows that were triggered by rainfall during Novem-
ber 1996. These storms generated flooding and debris
flows throughout western Oregon (R
obison
et alii,
1999; w
iley
, 2000; H
ofmeisteR
, 2000).
In both areas, debris flows mobilized from slides,
but entrainment and travel distances of the flows were
very different. In California, nearly all debris flows
did not entrain material and had relatively short travel
distances. In Oregon, nearly every flow entrained ma-
terial and travel distances were relatively long.
The physical properties of surficial geologic materi-
als, topographic boundary conditions, drainage sinuosity,
confinement, and junction angles, and the properties and
dynamics of the flows themselves influence entrainment
and travel distances (e.g., b
enda
& C
undy
, 1990; a
n
-
Fig. 1 - A) Map showing study area locations and drainage
basins where we computed drainage densities (e.g.,
Eden Canyon, Harvey Creek, etc.). B) Part of the
debris-flow inventory map for Oregon study area.
C) Part of the debris-flow inventory map for Cali-
fornia study area. Inventory maps are from c
oe
et
alii (2011) and c
oe
et alii (2004), respectively
background image
ASSESSMENT OF TOPOGRAPHIC AND DRAINAGE NETWORK CONTROLS ON DEBRIS-FLOW TRAVEL DISTANCE ALONG
THE WEST COAST OF THE UNITED STATES
Italian Journal of Engineering Geology and Environment - Book www.ijege.uniroma1.it © 2011 Casa Editrice Università La Sapienza
201
hollows, R
eneau
& d
ietRiCH
, 1987). Debris flows in
February 1998 were triggered by about 90 mm of rain-
fall on February 2-3. Seasonal antecedent rainfall prior
to February 2 exceeded 350 mm and antecedent soil
moisture conditions were well above the previously
defined rainfall-threshold levels (Coe and Godt, 2001).
Field observations following the February debris-flow
event indicated that nearly all (>99%) of debris flows
did not erode or entrain hillslope and channel material.
DOUGLAS COUNTY, SOUTHERN OREGON
Our second study area, in Douglas County, Or-
egon, is about 94 km
2
in size, lies within the southern
Oregon Coast Range, and is transected by the Ump-
qua River (Fig. 1). Elevations in the map area range
from sea level to about 490 m. The area has a maritime
climate with wet winters and dry summers. Average
annual precipitation ranges from about 1,600 to 2,300
mm. The wet and relatively warm climate results in
a coastal rain forest that is dominated by Douglas fir,
western hemlock, red alder, and a variety of under-
story shrubs. Trees are harvested throughout the study
area, usually through commercial clear-cut operations.
The area is underlain by Tertiary marine sand-
stones and siltstones of the Tyee Formation (w
alkeR
& m
C
l
eod
, 1991; m
a
et alii, 2009). The rocks are
gently folded and have a slight westward dip (k
elsey
et alii, 1996). The drainage network is dense, with a
dendritic pattern (e.g., m
ay
, 2002) that has often been
characterized as highly dissected (e.g., b
enda
, 1990).
Hillslopes (Fig. 2) are short, steep (20-40º), and typi-
cally mantled by 0.5 to 3 m of colluvial soil (R
eneau
& d
ietRiCH
, 1991; m
ontGomeRy
et alii, 2002).
Debris flows typically mobilize from slides in con-
cave hillslope areas, and increase in volume by erosion
and entrainment of downslope channel sediment (Fig. 2)
before depositing material in higher order drainage chan-
nels and fans (b
enda
& C
undy
, 1990). Debris flows in
1996 were triggered by about 230 mm of rainfall between
November 17 and November 20. Even though debris
flows occurred relatively early in the fall/winter wet sea-
son, antecedent rainfall in the study area was well above
the threshold of 100-280 mm suggested by w
iley
(2000).
Rainfall in the area between Oct. 1 and November 17,
1996 was about 440 mm. Based on inventory mapping
of the November, 1996 debris flows, our best estimate is
that about 95% of debris flows in the study area entrained
sediment and wood from downslope channels.
Bay region (Fig. 1). Hillslopes in the area (Fig. 2) have
moderate to steep gradients (10°-30º) and are mantled
by colluvial soil up to a few meters in thickness. Veg-
etation is mostly grass but includes some shrubs and
deciduous trees. Land use is predominantly livestock
grazing, but some areas have been converted to residen-
tial use. Elevations reach a maximum of about 600 m.
Climate in the area is Mediterranean, with mean annual
precipitation of about 460 mm in valleys, and as much
as 610 mm along upper flanks of the prominent north-
west-trending ridges (R
antz
, 1971). The area is under-
lain by Cretaceous and Tertiary sedimentary rocks of
marine and non-marine origin that have been exten-
sively folded and faulted by multiple oblique-slip faults
(G
RaymeR
et alii, 1996). Most of these rocks release
clay as they weather (e
llen
& w
entwoRtH
, 1995).
Landslides occur primarily during the late fall
through early spring wet season. Debris flows typi-
cally mobilize from small, shallow slides (e.g., see
Fig. 2 and e
llen
& f
leminG
, 1987; w
ieCzoRek
et alii,
1988; a
ndeRson
& s
itaR
, 1995), often in topographi-
cally convergent, concave hillslope areas (also called
Fig. 2 - Photographs of hillslopes and debris flows in
our study areas. A) California study area. Dis-
tance from lower left to upper right edges of the
photograph is about 2000 m. B) Oregon study
area. Distance from left to right edges of the
photograph is about 650 m. Oregon photo modi-
fied from Stock and Dietrich (2006)
background image
J.A. COE, M.E. REID, D.L. BRIEN & J.A. MICHAEL
202
5th International Conference on Debris-Flow Hazards Mitigation: Mechanics, Prediction and Assessment Padua, Italy - 14-17 June 2011
To determine initiation locations for debris flows,
we selected single points located at the approximate
center of each slide headscarp. Horizontal travel dis-
tances for each flow were measured along the ap-
proximate centerline of the flow, progressing from the
initiation locations, to the downslope end of the flow.
Some debris flows had multiple initiation locations that
contributed material to the flow. In these cases, travel
distances were measured from each initiation loca-
tion until they intersected, at which point the remain-
ing downslope distance was measured along the single
combined flow path. We extracted slope, upslope con-
tributing area, and planform curvature from the LiDAR
DEM grid cell at each initiation location using functions
within ESRI's (Environmental Systems Research Insti-
tute) ArcMap GIS. Positive planform curvature values
indicate a convex (divergent) slope, whereas negative
curvature values indicate a concave (convergent) slope.
Determining distances from initiation locations to
the drainage network required defining the drainage
network for each area. We defined the upslope most
extent of the network (i.e., the approximate location
of channel heads) using a DEM analysis technique.
We first evaluated the drainage density of 2-3 large
basins (Figs. 1 and 3) in each DEM. Drainage density
is defined as the total length of drainages per unit area.
To determine drainage density, we utilized D8 flow di-
rection algorithms (J
enson
& d
ominGue
, 1988) imple-
mented in ArcMap to determine flow direction, flow
accumulation, and total length of drainages. The total
length of drainages was then summed and divided by
the drainage area for each of the large basins.
Using these drainage densities, we identified con-
tributing area values where densities rapidly increased,
or “feathered” indicating that the local DEM topography
was no longer convergent. Feathering occurred when
computed drainage paths extended onto planar or diver-
gent hillslopes. We found that contributing areas for this
feathering effect for California and Oregon were 2000
m
2
and 1500 m
2
, respectively (Fig. 3). These values are
similar to those determined by previous work in the San
Francisco Bay region (m
ontGomeRy
& f
oufoula
-G
eoR
-
Giou
, 1993) and the Oregon Coast Range (m
illeR
, 2008).
The contributing area upslope of actual channel
heads in a region can be quite variable and a func-
tion of both topographic, and non-topographic factors
(m
ontGomeRy
& d
ietRiCH
, 1992; w
ilson
& G
allant
,
2000; t
aRboton
& a
mes
, 2001). Precise channel head
METHODS
We used debris-flow inventories and bare-earth
LiDAR data to examine topographic and drainage net-
work characteristics of debris-flow initiation locations
in both study areas. In California, debris flows were
mapped as single polygons that included slide initia-
tion locations, travel paths, and deposition areas from
12:000-scale aerial photographs acquired in May, 1998
(C
oe
et alii, 2004) (Fig. 1). In Oregon, debris flows were
mapped from 12:000-scale aerial photographs taken in
May, 1997. Unlike California, mapped debris flows in
Oregon were divided into two types of polygons, source
areas (including slide initiation locations, travel paths,
and areas of erosion and entrainment, see C
oe
et alii,
2011 for details) and areas of deposition (Fig. 1). Debris
flows in both areas were mapped onto 1:24,000-scale
USGS quadrangle maps enlarged to 1:12,000-scale.
Airborne LiDAR data were acquired in California
in 2006 and in Oregon in 2008. In California, LiDAR
data had an average ground-point spacing of 1.4 m (0.5
points/m
2
) and were gridded to a 1.52 m (5 ft.) cell size.
In Oregon, LiDAR data had an average density of 0.6
points/m
2
for ground-classified points. Ground-classi-
fied data were gridded to a 0.91 m (3 ft.) cell size.
Our analysis consisted of the following steps:
1) transferring and registering inventory data to the
LiDAR topographic base, 2) identifying accurate de-
bris-flow initiation locations on the LiDAR base, 3)
measuring the travel distance of each flow, 4) determin-
ing local slope, upslope contributing area, and planform
curvature for each initiation location, 5) identifying the
distance from initiation locations to local drainage net-
work, and 6) analyzing the topographic and drainage
network characteristics of initiation locations. Below,
we discuss our methods for each of these steps.
The process of transferring and registering mapped
debris flows to LiDAR-based maps was difficult because
the topographic details visible in the LiDAR were often
not visible in, or were different from, the more general-
ized quadrangle-based topography. Nevertheless, when
transferring mapped debris flows to the LiDAR bases,
we attempted to maintain their original shapes and
sizes. Travel distance lengths were not affected during
the transfer process, and debris flow widths were mini-
mally affected. However, in many instances, we had to
adjust the locations of debris flows, particularly along
channels (sometimes by much as 25 m (about 2 mm at
1:12,000-scale)), to properly fit the LiDAR topography.
background image
ASSESSMENT OF TOPOGRAPHIC AND DRAINAGE NETWORK CONTROLS ON DEBRIS-FLOW TRAVEL DISTANCE ALONG
THE WEST COAST OF THE UNITED STATES
Italian Journal of Engineering Geology and Environment - Book www.ijege.uniroma1.it © 2011 Casa Editrice Università La Sapienza
203
locations in each individual basin cannot be deter-
mined without extensive fieldwork, which is imprac-
tical in large study areas. However, by using the den-
sity analysis approach, we were able to objectively
determine overall drainage networks.
Once the drainage networks were established, we
measured distances from debris-flow initiation loca-
tions to the network. These distances were determined
by progressively accumulating distances along flow
paths from initiation locations to the first intersection
with the drainage network using the flow direction
methodology of t
aRboton
(1997). A distance of zero
means that an initiation location is within the defined
drainage network. Throughout the text, we use the
phrase “interact with the drainage network” to describe
debris flows that either initiated within the network, or
entered the network as they flowed downslope.
We analysed topographic and network character-
istics using statistical techniques including histogram
analyses, descriptive statistics, regression analyses,
probability density analyses, and the Mann-Whitney U
test for non-parametric data (m
ann
& w
Hitney
, 1947).
The Mann-Whitney U-Test is a non-parametric test of
significance used to determine if differences between
two groups of samples are significant. Probability val-
ues (p values) from the Mann-Whitney test indicate
the probability that the two samples being compared
would be truly different from a single large popula-
tion. For example, a p value of 0.05 (5%) indicates
that there is a 95% chance that the two populations
are different. By convention, the typical critical value
of p used to determine significance is 0.05 (i.e., the
null hypothesis that the two groups are from the same
population is rejected for p values less than 0.05).
RESULTS
DEBRIS-FLOw TRAVEL DISTANCE
Visually, there is a large difference in typical de-
bris-flow travel distances between the two study areas
(Fig. 1). Our travel distance measurements, when strat-
ified by the number of contributing initiation locations,
show that median distances for debris flows in Oregon
were between 5 and 8 times longer than median dis-
tances for debris flows in northern California (Fig. 4).
In California, median travel distances were 27 m for
flows that originated from single initiation locations,
and 219 m for flows that initiated from four or more
initiation locations (Fig. 4). In Oregon, median travel
distances were 126 m for flows that originated from
single initiation locations, and 1748 m for flows that
initiated from four or more initiation locations (Fig. 4).
Fig. 3 - Diagram showing drainage density analysis for
the California and Oregon study areas. Drainage
areas where feathering occurs (approximate limit
of drainage network) are shown
Fig. 4 - Travel distances of debris flows as a function of
the number of initiation locations contributing
to the flow. Curves show power-law fits to the
median data. A) California. B) Oregon. Travel
distances of many flows in Oregon could not be
measured because they extended to areas outside
the coverage of the inventory map
background image
J.A. COE, M.E. REID, D.L. BRIEN & J.A. MICHAEL
204
5th International Conference on Debris-Flow Hazards Mitigation: Mechanics, Prediction and Assessment Padua, Italy - 14-17 June 2011
Because these three topographic variables, and the
lengths of travel distances, were distinctly different for
the two study areas, we tested each variable as a possi-
ble predictor of travel distance using regression analy-
ses. For these analyses, we only used debris flows with
single initiation locations. We did not find any signifi-
cant positive or negative correlation between travel dis-
tance and any of the topographic variables, commonly
used to identify initiation sites, in either study area.
We investigated the topographic variables further
by grouping travel distances of flows from single ini-
tiation locations according to important topographic
threshold values. For example, travel distances were
grouped according to slope values at initiation loca-
tions, one group from locations with slopes ≤ 20° and
one group from locations with slopes > 20°. Other slope
thresholds used were 25º, 30º and 35º. The threshold
value used for planform curvature was 0 (convergent
vs. divergent slopes) and the threshold values used for
contributing area were 2000 m
2
and 1500 m
2
for Cali-
fornia and Oregon, respectively. Mann-Whitney U tests
of these groupings showed that only contributing area
discriminated between different travel distances in
both study areas. Travel distances were significantly
(p=0.001 in California and 0.029 in Oregon) longer for
flows that initiated from locations with contributing ar-
eas greater than threshold values compared to those that
initiated from locations with contributing areas less than
the threshold values. In general, Mann-Whitney tests of
groups based on slope and planform curvature values
did not yield significantly different travel distances. One
exception was curvature in California. Here, convergent
initiation locations (curvature < 0) produced longer
In both study areas, there is a positive, power-law
relation between median debris flow travel distance and
the number of contributing debris-flow initiation loca-
tions (Fig. 4). In California, this relation is defined as
y=29.7x
1.6
, where y is median travel distance and x is
number of contributing initiation locations. In Oregon,
the relation is defined as y=125.6x
1.8
. This same relation
holds for maximum and minimum travel distances in
Oregon (Fig. 4A), and for minimum travel distances in
California (Fig. 4A). Maximum distances in California
appear to vary from this pattern (Fig. 4A), but the lack
of observed debris flows (<2% of the total number) with
more than 2 initiation locations may bias the results.
SLOPE, CONTRIBUTING AREA, AND CURVATURE
CHARACTERISTICS OF INITIATION LOCATIONS
There are distinct differences in the slope, con-
tributing area, and planform curvature values of de-
bris-flow initiation locations in California and Oregon
(Figs. 5 and 6). In general, Oregon initiation locations
have steeper and more convergent slopes with larger
upslope contributing areas than initiation locations in
California (Figs. 5 and 6). The median slope and ups-
lope contributing area of initiation locations in Oregon
are 38° and 1529 m
2
, respectively (Fig. 5). In Califor-
nia, the median slope and upslope contributing area of
initiation locations are 28° and 42 m
2
, respectively. In
Oregon, more than 41 percent of initiation locations
have planform curvature values less than -10, whereas
only about 1 percent of initiation locations in Califor-
nia have curvature values less than -10 (Fig. 6). The
median curvature values of initiation locations in Or-
egon and California are -6.81 and -0.29, respectively.
Fig. 5 - Diagram showing slope and contributing area
characteristics of initiation locations (land-
slide headscarps) in California and Oregon.
Number of initiation locations was 438 in Or-
egon and 2400 in California
Fig. 6 - Histogram showing planform curvature charac-
teristics of initiation locations in California and
Oregon. Number of initiation locations was 438 in
Oregon and 2400 in California
background image
ASSESSMENT OF TOPOGRAPHIC AND DRAINAGE NETWORK CONTROLS ON DEBRIS-FLOW TRAVEL DISTANCE ALONG
THE WEST COAST OF THE UNITED STATES
Italian Journal of Engineering Geology and Environment - Book www.ijege.uniroma1.it © 2011 Casa Editrice Università La Sapienza
205
PROXIMITY TO DRAINAGE NETwORk
The proximity of initiation locations to the drainage
network was distinctly different for the two study areas
(Fig. 7). In Oregon, 52% of debris flow initiation loca-
tions were within the drainage network, and the overall
median distance to the network was 0. In California, only
3% of initiation locations were within the drainage net-
work and the median distance to the network was 135 m
(Fig. 7). In Oregon, the distribution of initiation location
proximity is approximately log normal and decreases as
distance to the drainage network becomes larger. In Cali-
fornia, the distribution of initiation location proximity is
approximately normal about a mean of 156 m.
We investigated our distance to drainage network
data further by grouping travel distances according to
whether or not debris flows interacted with our defined
drainage networks. Using Mann-Whitney U tests we
found that the travel distances of flows that interacted
with the drainage networks were significantly (p <
1x10
-7
) longer than flows that did not (Fig. 8). Using
our proximity to drainage network data, we selected
a generalized threshold for forecasting whether or not
flows would reach the drainage network in each study
area (Fig. 9). These thresholds correspond to distance
(p=0.002) travel distances than planar or divergent to-
pography (curvature ≥ 0). None of the topographic vari-
ables were useful for forecasting whether or not flows
would interact with the drainage network.
Fig. 7 - Histogram showing distance from initiation loca-
tions to drainage network in each study area. Dis-
tance values of zero are from sites that initiated
within the drainage network. Number of initiation
locations was 438 in Oregon and 2400 in California
Fig. 8 - Box and whisker plots showing travel distance
data, grouped with respect to whether the flows
interacted with the drainage network, for debris
flows from single initiation areas. A) California.
179 flows interacted with the drainage network,
1829 did not. B) Oregon. 89 flows interacted with
the drainage network, 26 did not
Fig. 9 - Histograms showing distance from initiation lo-
cations to drainage network, grouped by whether
or not debris flows interacted with the drainage
networks. Distance values of zero are from initia-
tion locations within the drainage network. Only
flows with a single initiation location are shown.
A) California. 2008 flows. B) Oregon. 115 flows
background image
J.A. COE, M.E. REID, D.L. BRIEN & J.A. MICHAEL
206
5th International Conference on Debris-Flow Hazards Mitigation: Mechanics, Prediction and Assessment Padua, Italy - 14-17 June 2011
values where about 85% of flows that entered the
channel were less than the threshold, and 85% that
did not enter the channel were greater than the thresh-
old. The thresholds for entering the networks are 60
m in California and 80 m in Oregon.
Probability density functions (pdfs) of travel dis-
tances for flows that interacted with the drainage net-
works, and for those that did not, are shown in Fig. 10.
For both study areas, log normal pdfs provide good fits to
the observed data. In a predictive sense, these functions
could be used in combination with Fig. 9 to estimate the
distribution of travel distances in each study area.
DISCUSSION
Predictive maps and models of debris-flow initia-
tion locations, probability of occurrence, entrainment
potential, expected volumes, and travel distances are
critical to the efficient management of debris-flow haz-
ards (e.g., b
aum
et alii, this volume), yet high-quality
data sets from natural debris flows that are needed to
support these maps and models are relatively rare (e.g.,
m
C
C
oy
, et alii, this volume; s
taley
et alii, this vol-
ume). Our results, using high quality debris-flow in-
ventory maps and recently available LiDAR data, show
that there are distinct differences in the topographic
and drainage network characteristics of debris-flow
initiation locations of our two study areas. However,
only three variables, the number of contributing initia-
tion locations (Fig. 4), contributing area, and distance
from initiation locations to the drainage network (Figs.
8-10), were useful for estimating debris-flow travel
distances. Slope and planform curvature of initiation
locations could not consistently predict travel distance.
The positive correlation between number of initia-
tion locations and travel distance reinforces the concept
that flow volume is a primary factor controlling travel
distance, but it also indicates that the number of contrib-
uting initiation locations may be used as crude proxy for
volume. The regression equation for Oregon applies to
flows where volume increased from entrainment, where-
as the equation for California applies to flows with mini-
mal entrainment. These relations suggest that similar
equations could be developed for other geographic areas.
Results from our analysis of contributing area and
distance from initiation locations to the drainage net-
work indicate that distance from initiation locations
is the better predictor of travel distance (p ≥ 0.001 for
contributing area vs. p < 1x10
-7
for distance from ini-
tiation locations). Distance from initiation locations to
the drainage networks is also a good indicator of the
likelihood of debris flows interacting with the drain-
age network (Fig. 9). Our statistical analyses show that
flows which interact with the network generally travel
Fig. 10 - Probability density functions for debris-flow trav-
el distance data, grouped by whether or not debris
flows interacted with the drainage network. Log
normal equations are shown. A) California flows,
n= 1829. B) California flows, n= 179. C) Oregon
flows, n= 26. D) Oregon flows, n= 89
background image
ASSESSMENT OF TOPOGRAPHIC AND DRAINAGE NETWORK CONTROLS ON DEBRIS-FLOW TRAVEL DISTANCE ALONG
THE WEST COAST OF THE UNITED STATES
Italian Journal of Engineering Geology and Environment - Book www.ijege.uniroma1.it © 2011 Casa Editrice Università La Sapienza
207
ry flow entrained material and median travel distances
were 5 to 8 times longer than those in California.
Our results show that there are distinct differences
in the topographic and drainage network characteris-
tics of debris-flow initiation locations in the two study
areas, but that only three variables were useful for
estimating debris-flow travel distances: the number
of contributing initiation locations (Fig. 4), contribut-
ing area, and distance from initiation locations to the
drainage network (Figs. 8-10). A positive, power-law
relation exists between median debris flow travel dis-
tance and the number of contributing debris-flow initi-
ation locations. For the events we studied, this relation
is defined as y=29.7x
1.6
in California and y=125.6x
1.8
in Oregon, where y is median travel distance and x is
number of contributing initiation locations.
In both study areas, our analyses showed that flows
which interacted with the drainage network flowed
farther than flows that did not. We used distance to
drainage network data for specific triggering events to
establish a distance threshold to classify which flows
(that begin from single initiation locations) will enter
the network. The threshold is 60 m in California and
80 m in Oregon. Sources closer than these values will
likely enter the network, whereas flows from farther
away will not. Log normal probability density func-
tions of travel distances differ significantly for flows
that interact with the drainage network, compared to
flows that do not. These density functions could be
used to estimate travel distances of flows from single
initiation locations in the two study areas.
ACKNOWLEDGEMENTS
We thank Dennis Staley, Bill Burns, and two
anonymous reviewers for their constructive reviews
of this paper.
farther than flows that do not (Figs. 8 and 10). This ef-
fect is likely related to the availability of surface water
and readily erodible sediment in drainage networks. In
contrast, these effects are subdued on hillslopes. Based
on field work in Oregon, most small, steep drainage ba-
sins where slides initiate contain steep-walled channels
with perennial surface water flow and channel sediments
that are typically saturated, especially in the winter wet
season when debris flows are likely. Springs are com-
mon at and near channel heads throughout the year. In
our northern California study area, most debris flows
do not initiate in the drainage network and the flow of
surface water at initiation locations is rare, at least prior
to failure. Here, shallow slides are triggered by positive
pore-pressure pulses (J
oHnson
& s
itaR
, 1990) that may
or may not reflect the moisture level of downslope surfi-
cial sediments. Recent work at the USGS experimental
debris-flow flume in Oregon shows that the potential for
sediment entrainment by debris flows is positively corre-
lated with the moisture content of the bed sediment (Reid
et alii, this volume); in these experiments entrainment of
wetter sediment enhanced travel distance whereas en-
trainment of drier sediment retarded flow, and it follows
that flows that entrain sediment have larger volumes and
may travel greater distances than flows that do not.
CONCLUSIONS
We analyzed two separate, large-scale, debris-flow
data sets from the west coast of the U.S. using high res-
olution LiDAR data. Both data sets, one from northern
California and the other from southern Oregon, record
precipitation triggered debris flows. In both areas, de-
bris flows mobilized from slides, but entrainment and
travel distances of the flows differed markedly. In Cali-
fornia, debris flows did not entrain material and had
relatively short travel distances. In Oregon, nearly eve-
REFERENCES
a
ndeRson
s.a. & s
itaR
n. (1995) - Analysis of rainfall - induced debris flows. Journal of Geotechnical Engineering,
121: 544-552.
b
aum
R.l., G
odt
J.w. & C
oe
J.a. (this volume) - Assessing susceptibility and timing of shallow landslide and debris flow
initiation in the Oregon Coast Range, USA.
b
enda
L. (1990) - The influence of debris flows on channels and valley floors in the Oregon Coast Range, USA. Earth Surface
Processes and Landforms, 15: 457-466.
b
enda
l.e. & C
undy
t.w. (1990) - Predicting deposition of debris flows in mountain channels. Canadian Geotechnical Journal:
27: 409-417.
C
oe
J.a. & G
odt
J.w. (2001) - Debris flows triggered by the El Niño rainstorm of February 2-3, 1998, walpert Ridge and vicinity,
Alameda County, California. U.S. Geological Survey Miscellaneous Field Studies Map MF 2384, 3 plates, scale 1:24,000.
background image
J.A. COE, M.E. REID, D.L. BRIEN & J.A. MICHAEL
208
5th International Conference on Debris-Flow Hazards Mitigation: Mechanics, Prediction and Assessment Padua, Italy - 14-17 June 2011
C
oe
J.a., G
odt
J.w. & t
aCHkeR
P. (2004) - Map showing recent (1997-98 El Nino) and historical landslides, Crow Creek and vicinity,
Alameda and Contra Costa Counties, California. U.S. Geological Survey Scientific Investigations Map 2859, Reston, Virginia.
C
oe
J.a., m
iCHael
J.a. & b
uRGos
m.m. (2011) - Map showing debris flows caused by precipitation in 1996 in parts of the Reedsport
and Deer Head Point quadrangles, southern Coast Range, Oregon. U.S. Geological Survey Open File Report, Reston, Virginia.
d
ietRiCH
w.e., b
elluGi
d. & R
eal
de
a
sua
R. (2001)- Validation of the shallow landslide model, SHALSTAB, for forest management,
In w
iGmosta
m.s. & b
uRGes
s. J. (2001, Eds.), - Land Use and watersheds: Human influence on hydrology and geomorphology
in urban and forest areas. American Geophysical Union, Water Science and Application 2: 195-227, Washington, D.C.
e
llen
s.d. & f
leminG
R.w. (1987) - Mobilization of debris flows from soil slips, San Francisco Bay region, California, In:
C
osta
J.e. & w
ieCzoRek
G.F. (1987 - Eds.) - Debris flows/Avalanches: Process, Recognition, and Mitigation, Reviews in
Engineering Geology VII: 31-40, Geological Society of America, Boulder, Colorado.
e
llen
s.d. & w
entwoRtH
C.M. (1995) - Hillside materials and slopes of the San Francisco Bay region. U.S. Geological Survey
Professional Paper 1357, Reston, Virginia.
G
RaymeR
R.w., J
ones
d.l., & b
Rabb
e.e. (1996) - Preliminary geologic map emphasizing bedrock formations in Alameda County,
California - A digital database. U.S. Geological Survey Open-File Report 96-252, map scale 1:75,000, Reston, Virginia.
G
Riswold
J.P. & i
veRson
R.m. (2007) - Mobility statistics and automated hazard mapping for debris flows and rock avalanches.
U.S. Geological Survey Scientific Investigations Report 2007-5276, Reston, Virginia.
H
ofmeisteR
R.J. (2000) - Slope failures in Oregon, GIS inventory for three 1996/97 storm events. Department of Oregon Geology
and Mineral Industries Special Paper 34, Portland, Oregon.
H
unGR
o., m
C
d
ouGall
s., & b
ovis
m. (2005) - Entrainment of material by debris flows, In: J
akob
, m. & H
unGR
, O. (2005 -
Eds.) - Debris-flow Hazards and Related Phenomena, 135-158, Springer, Berlin Heidelberg,.
i
veRson
R.m., s
CHillinG
s.P., & v
allanCe
J.w. (1998) - Objective delineation of lahar-hazard inundation zones. Geological
Society of America Bulletin, 110: 972-984.
i
veRson
R.m., R
eid
m.e., i
veRson
n.R., l
a
H
usen
R.G., l
oGan
m., m
ann
J.e. & b
Rien
d.l. (2000) - Acute sensitivity of
landslide rates to initial soil porosity. Science, 290: 513-516.
J
enson
s.k. & d
ominGue
J.o. (1988) Extracting topographic structure from digital elevation data for geographic information
system analysis. Photogrammetric Engineering and Remote Sensing, 54: 1593-1600.
J
oHnson
, k.a. & s
itaR
, n. (1990) - Hydrologic conditions leading to debris flow initiation. Canadian Geotechnical Journal, 27: 789-801.
k
elsey
, H.m., t
iCknoR
, R.l., b
oCkHeim
, J.G., & m
itCHell
, C.e. (1996) - Quaternary upper plate deformation in coastal Oregon.
Geological Society of America Bulletin, 108: 843-860.
m
a
l., m
adin
i.P., o
lson
k.v., w
atziG
R.J., w
ells
R.e., n
iem
a.R. & P
Riest
G.R. (2009) - Oregon geologic data compilation
(OGDC), release 5. Oregon Department of Geology and Mineral Industries, Portland, Oregon.
m
ay
C.l. (2002) - Debris flows through different forest age classes in the central Oregon Coast Range. Journal of the American
Water Resources Association, 38: 1097-1113.
m
ann
H.b. & w
Hitney
d.R. (1947) - On a test of whether one of 2 random variables is stochastically larger than the other.
Annals of Mathematical Statistics, 18: 50-60.
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
C
C
oy
s.w., C
oe
J.a., k
ean
J.w., t
uCkeR
G.e., s
taley
d.m., & w
asklewiCz
t.a. (this volume) - 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
.
m
illeR
d. (2008) - Programs for DEM analysis. Earth Systems Institute, Mt Shasta, California.
m
ontGomeRy
d.R. & d
ietRiCH
w.e. (1992) - Channel initiation and the problem of landscape scale. Science, 255: 826-830.
m
ontGomeRy
d.R. & f
oufoula
-G
eoRGiou
e. (1993) - Channel network source representation using digital elevation models.
Water Resources Research, 29: 3925-3934.
m
ontGomeRy
d.R. & d
ietRiCH
w.e. (1994) - A physically-based model for the topographic control of shallow landsliding. Water
Resources Research, 30: 1153-1171.
m
ontGomeRy
d.R., d
ietRiCH
w.e. & H
effneR
J.t. (2002) - Piezometric response in shallow bedrock at CB1: Implications for
runoff generation and landsliding. Water Resources Research, 38: 1274, doi:10.1029/2002WR001429.
background image
ASSESSMENT OF TOPOGRAPHIC AND DRAINAGE NETWORK CONTROLS ON DEBRIS-FLOW TRAVEL DISTANCE ALONG
THE WEST COAST OF THE UNITED STATES
Italian Journal of Engineering Geology and Environment - Book www.ijege.uniroma1.it © 2011 Casa Editrice Università La Sapienza
209
R
antz
s.e. (1971) - Mean annual precipitation and precipitation depth-duration-frequency data for the San Francisco Bay
region, California. U.S. Geological Survey, San Francisco Bay Region Environment and Resources Planning Study Basic
Data Contribution 32: 1:500,000-scale map, Reston, Virginia.
R
eid
m.e., i
veRson
R.m., l
oGan
m., l
a
H
usen
R.G., G
odt
J.w. & G
Riswold
J.P. (this volume) - Entrainment of bed sediment by
debris flows: results from large-scale experiments.
R
eneau
s.l. & d
ietRiCH
w.e. (1987) - The importance of hollows in debris flow studies; Examples from Marin County,
California. In: C
osta
J.e. & w
ieCzoRek
G.f. (1987, Eds.) - Debris Flows/Avalanches: Process, Recognition, and Mitigation,
Reviews in Engineering Geology, VII: 165-180, Geological Society of America, Boulder, Colorado.
R
eneau
s.l. & d
ietRiCH
w.e. (1991) - Erosion rates in the southern Oregon Coast Range: Evidence for an equilibrium between
hillslope erosion and sediment yield. Earth Surface Processes and Landforms, 16: 307-322.
R
obison
e.G., m
ills
k., P
aul
J., d
ent
,l. & s
kauGset
a. (1999) - Storm impacts and landslides of 1996: Final Report. Oregon
Department of Forestry, Forestry Practices Technical Report number 4, Salem, Oregon.
s
ilbeRt
l.e., G
Rest
G.s., P
limPton
s.J. & l
evine
, d. (2002) - Boundary effects and self-organization in dense granular flows.
Physics of Fluids, 14: 2637-2646, doi: 10.1063/1.1487379.
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. (this volume) - Observations of debris flows at
Chalk Cliffs, Colorado, USA: Part 2, Changes in surface morphometry from terrestrial laser scanning in the summer of 2009.
s
toCk
J.d. & d
ietRiCH
w.e. (2006) - Erosion of steepland valleys by debris flows. Geological Society of America Bulletin, 118:
1125-1148.
t
aRboton
D.G. (1997) - A new method for the determination of flow directions and upslope areas in grid digital elevation
models. Water Resources Research, 33: 309-319.
t
aRboton
d.G. & a
mes
d.P. (2001) - Advances in the mapping of flow networks from digital elevation data: In World Water and
Environmental Resources Congress, Orlando, Florida.
w
alkeR
G.w. & m
aC
l
eod
n.s. (1991) - Geologic map of Oregon: U.S. Geological Survey, 1:500,000 scale, Reston, Virginia.
w
ieCzoRek
G.f., H
aRP
e.l., m
aRk
R.k. & b
HattaCHaRyya
a.k. (1988) - Debris flows and other landslides in San Mateo, Santa Cruz,
Contra Costa, Alameda, Napa, Solano, Sonoma, Lake, and Yolo Counties, and factors influencing debris-flow distribution. In:
e
llen
s.d. & w
ieCzoRek
G.F. (1987, Eds.) - Landslides, floods, and marine effects of the storm of January 3-5, 1982, in the San
Francisco Bay region, California. U.S. Geological Survey Professional Paper 1434: 133-162, Reston, Virginia.
w
ilson
J.P. & G
allant
J.C. (2000, Eds.) - Terrain Analysis, Principles and Applications: John Wiley & Sons, New York.
w
iley
t.J. (2000) - Relationship between rainfall and debris flows in western Oregon: Oregon Geology, 62: 27-43.
Statistics