Document Actions

IJEGE-11_BS-Kean-&-Staley

background image
Italian Journal of Engineering Geology and Environment - Book www.ijege.uniroma1.it © 2011 Casa Editrice Università La Sapienza
685
DOI: 10.4408/IJEGE.2011-03.B-075
DIRECT MEASUREMENTS OF THE HYDROLOGIC CONDITIONS
LEADING UP TO AND DURING POST-FIRE DEBRIS FLOW
IN SOUTHERN CALIFORNIA, USA
J
ason
W. KEAN
(*)
& d
ennis
M. STALEY
(*)
(*)
U.S. Geological Survey, Denver, Colorado, USA
INTRODUCTION
In the mountainous western United States, debris
flows can be a common hazard after wildfire. In the
first one to two years after a fire, most debris flows
are generated by surface-water runoff entraining loose
material stored on hillslopes and in channels (e.g.,
m
eyeR
& w
ells
, 1997; C
annon
et alii, 2001) and can
be triggered by much less rainfall than is required for
debris-flow initiation in unburned areas (C
annon
et
alii, 2001; 2008). There is a growing need to better
understand these hazards given the high fire frequency
associated with recent climate trends (w
esteRlinG
et
alii 2006) and continued residential development in
steep, fire-prone areas. Rain data, post-event field
measurements, and eyewitness accounts have been
used to develop both empirical rainfall intensity-du-
ration thresholds for post-fire debris flows (C
annon
et alii, 2008; C
annon
et alii, 2010a) and empirical
models for debris-flow probability and volume (C
an
-
non
et alii, 2010b). These tools have proven valuable
for assessing post-fire threats over large regions, and,
together with weather forecasts, are the foundation
for the current southern California debris-flow warn-
ing system operated jointly by the National Oceanic
and Atmospheric Association (NOAA) and the United
States Geological Survey (USGS) (Cannon et alii,
2010c; NOAA-USGS Debris Flow Task Force, 2005).
Yet further advancement in predictive capabili-
ties and development of new predictive tools, such
as physically based models, requires detailed data on
ABSTRACT
Steep, recently burned watersheds can be vulner-
able to debris flows. In southern California, USA, the
combination of mountainous terrain, dense population,
and high fire frequency put new areas at risk to debris
flows each year. In an effort to improve predictions
of the timing and magnitude of post-fire debris flows,
the U.S. Geological Survey (USGS) established five
debris-flow monitoring sites in different southern Cali-
fornia watersheds burned in 2009. These sites record-
ed, for the first time, detailed measurements of the hy-
drologic conditions leading up to and during post-fire
debris flows. Measurements included precipitation,
hillslope soil-water content, flow stage, and pore pres-
sure. Here, we present initial observations and com-
parisons of debris flows measured during four storms
at our smallest study site (0.01 km
2
) located in the Sta-
tion Fire burn area in the San Gabriel Mountains. The
monitored debris flows were generated by progressive
entrainment of sediment from hillslope rilling and
channel erosion, which occurred in response to short-
duration bursts of intense rainfall. The measurements
show a distinct change in the flow response over the
course of the winter storm season, beginning with a
debris-flow dominated response in the first part of the
season and followed by more watery flows in the latter
part of the season. The change in flow response is pre-
sumably related to a decrease in sediment availability.
K
ey
words
:debris flow, fire, monitoring
background image
J.w. kEAN & D.M. STALEY
686
5th International Conference on Debris-Flow Hazards Mitigation: Mechanics, Prediction and Assessment Padua, Italy - 14-17 June 2011
it difficult to quantify precisely the triggering condi-
tions and mechanics of post-fire debris flows.
In an attempt to bridge the data gap, we have
started a program to collect the basic measurements
necessary for development and rigorous testing of
predictive models of post-fire debris-flow initiation.
This type of data will also make it possible to make
detailed comparisons between post-fire debris flows
and non-fire-related debris flows. The approach in-
volves establishing post-fire debris-flow monitoring
stations having some of the measurement capabili-
ties found in long-term debris-flow monitoring sites
in unburned areas (e.g. b
eRti
et alii, 2000; m
aRCHi
et alii, 2002; m
C
a
Rdell
et alii, 2007; s
uwa
et alii,
2009; m
C
C
oy
et alii, 2010). Measurements include
precipitation, hillslope soil water-content, flow
stage, pore pressure, video, and, at some sites, high-
resolution topography from terrestrial and airborne
the hydrologic conditions leading up to and during
post-fire events (e.g. combined time-series of rain and
flow stage). Unfortunately, the existence of such data
is limited. Limits are imposed by both the destructive
nature of flash floods and debris flows, and the very
narrow time window available to collect such data. In
southern California, the end of the fire season gener-
ally corresponds to the beginning of the winter storm
season. Installation of a field monitoring network be-
tween the fire and the first rainstorm is technically and
logistically very difficult. For this reason, available
time-series of post-fire runoff is usually restricted to
floods, such as those measured at pre-existing stream-
flow gauging stations on high-order tributaries. Infor-
mation on the timing of debris flow is usually only
available from rare eyewitness accounts. The lack of
timing and other hydrologic information in the low-
order channels where debris flows originate has made
Fig. 1 - Study area. A) Map showing location of study area within the Station Fire burn area. B) Aerial photograph of main study
basin taken 11 Feb. 2010. The study area is almost completely denuded of vegetation except at the valley bottom, where
a partially burned canopy remains. C) Photograph of ridge station. The rugged terrain in the back ground is typical of
the burn area. D) Photograph of channel monitoring equipment near the outlet of the main study area. E) Map of main
study basin and drainage area above Highway 2. F) Photograph of valley rain gage and soil moisture probes
background image
DIRECT MEASUREMENTS OF THE HYDROLOGIC CONDITIONS LEADING UP TO
AND DURING POST-FIRE DEBRIS FLOW IN SOUTHERN CALIFORNIA, USA
Italian Journal of Engineering Geology and Environment - Book www.ijege.uniroma1.it © 2011 Casa Editrice Università La Sapienza
687
derived from granitic rocks (Y
eRkes
& C
amPbell
,
2005). Soil thickness ranges from 0.05 cm to about
50 cm. The climate is Mediterranean, and character-
ized by hot, dry summers and a winter with occasional
storms. The native vegetation is chaparral, which is
adapted for growing in fire-prone areas. Our instru-
ments are in the same basin as additional equipment
for monitoring hillslope hydrology, which is described
in Schmidt et alii (this volume).
The study site was burned in the 650 km
2
Station
Fire of August and September 2009 - the largest fire
in Los Angeles county history. Ninety-nine percent of
the study area was burned at moderate to high severity
(s
tation
f
iRe
baeR t
eam
, 2009). Several episodes
of damaging post-fire debris flows have occurred over
the last hundred years in the vicinity of the study site.
e
aton
(1936) documented post-fire debris flows in
the area between 1914 and 1935; the largest of these
killed 30 people and damaged 483 homes in the com-
munities of La Crescenta and Montrose (see also,
C
HawneR
, 1934). In 1969 and 1978, debris flows fol-
lowing wildfires earlier in the summer caused exten-
sive damage again in the area (s
Cott
, 1971; m
C
P
Hee
,
1989; s
HuRiman
& s
lossen
, 1992). Debris flows in
the winter months following the 2009 Station Fire
damaged or destroyed 41 homes along the mountain
front (k
im
et alii, 2010) and caused major damage to a
heavily travelled road below the monitoring site.
METHODS
Field reconnaissance for site selection began on
17 Sept. prior to containment on 16 Oct. 2009. The
Arroyo Seco site was the second of four sites to be
established in the Station Fire (A fifth was established
in the 2009 Jesusita Fire in Santa Barbara, Califor-
nia). Installation took place 7-10 Nov. by the authors.
The first debris flow occurred 12 Nov. - two days after
installation. This short 2-month timeline for site se-
lection and installation highlights one of the principal
challenges of post-fire debris-flow monitoring. Pho-
tos and specifications of the monitoring equipment
at the main station are shown in Fig. 1 and listed in
Tab. 2. Similar equipment and sensor configurations
were used at our other 2009 monitoring sites. The
equipment consists of sensors to measure stage (H),
pore pressure (P), rainfall (R), and soil-water con-
tent. Stage and pore pressure sensors were located at
a cross section 3-m upstream of the confluence with
laser scanning. Each year one to five sites are estab-
lished in new fires in southern California.
Southern California is an ideal region to conduct
this work because of the regular occurrence of large
fires (typically three or more each year), variety of
geologic settings, predictable winter rainy season;
and, more importantly, it is the region of the USA
most at risk to these destructive events. Sites are
established in small, steep drainage areas (0.01 to 1
km2) as soon as possible after the fire is extinguished
(typically late fall). The sites are maintained usually
for one or two rainy seasons and then the equipment
is moved to more vulnerable locations in new burned
areas. The monitoring sites are relocated because
rapid post-fire growth of vegetation quickly reduces
the debris-flow hazard after one or two years. Fol-
lowing this approach, it is possible over the course
of a few years to procure flash-flood and debris-flow
data sets in a number of basins of different scales and
geologic and topographic characteristics.
The purpose of this paper is to describe our post-
fire debris flow monitoring approach in general, present
initial results from the first (to our knowledge) direct
measurements of post-fire debris flows, and discuss
the implications of these results for debris-flow warn-
ing systems. Results are presented from the smallest of
five post-fire monitoring sites established in 2009. We
focus on a comparison of flows measured during a se-
ries of four debris-flow producing storms that occurred
in the first winter rainy season after the fire.
STUDY SITE
A map of the study site (Arroyo Seco) is shown
in Figure 1, and a summary of basin characteristics is
given in Tab. 1. The 0.01 km
2
study area is located on
the San Gabriel mountain front and drains to the heav-
ily travelled Angeles Crest Highway (Highway 2).
The study basin is very steep and composed of soils
Tab. 1 - Topographic characteristics of study area from a
2-m DEM
background image
J.w. kEAN & D.M. STALEY
688
5th International Conference on Debris-Flow Hazards Mitigation: Mechanics, Prediction and Assessment Padua, Italy - 14-17 June 2011
would decrease the amount of rain intercepted by the
gage. For this reason, only data from the lower (val-
ley) rain gage is used in this paper.
Near-real-time data telemetry was established
to check for problems with the sensors and provide a
rapid (albeit lower-resolution) assessment of flow con-
ditions. A cell-phone modem located at the ridge site
was used to transmit one-minute data from all of the
sensors. Higher resolution data from the laser and pore-
pressure sensor was downloaded manually due to the
large file sizes. Data from the valley datalogger, which
did not have cell phone reception, was first relayed to
the ridge site via radio and then transmitted to our of-
fice computer in Golden, Colorado. Data was collected
every 5 min., graphed, and then posted on a public
web page (see, for example http://landslides.usgs.gov/
monitoring/). The web-page data typically lagged the
current conditions by 5-10 min. The 10-Hz laser stage
data was processed using a simple median filter with a
window size of 5 data points (0.5 sec.). The filter re-
places each data point with the median of neighbouring
points contained in the window. This filter preserved
most of the complex structure of the time series, but
removed anomalously high and low readings, such as
those caused by missed laser returns and returns from
large splashes. At selected 20-min. segments of the
stage records, the local variability of the flow was as-
sessed by computing the standard deviation (σ) of the
stage data about a smoothed version of the time series
(H
smooth
). This was done by smoothing the median-fil-
tered signal using a 21-point moving (box-car) average
(2-second time window), subtracting it from the uns-
moothed series (H), and calculating the standard devia-
tion of the residuals. For a for a 20-min. time window, σ
is computed from 6000 residuals of H - H
smooth
.
Continuous time series of rainfall intensities for
different durations (D = 2, 5, 10, 15, 30, 60, 180, 360,
and 720 min.) were calculated from the rainfall data.
The rainfall intensity, I, for a given duration was cal-
culated at 1-min. intervals by first taking a backward
difference of the cumulative rainfall at the current time
and D minutes earlier; and then dividing this difference
by the duration. Results for the 10- and 60- min. intensi-
ties (I
10
and I
60
) are presented here. The 10-min. dura-
tion was chosen, because a correlation analysis of stage
and rain intensities from all five 2009 monitoring sites
showed I
10
to have the highest cross-correlation with
stage. The 60-min. duration was chosen for comparison
a higher order tributary. Stage was measured using
both laser and radar distance meters that were sus-
pended approximately 2-m above the channel on a
portable 6-m aluminium bridge. The laser distance
meter, which produces an analogue 4-20 milli-Amp
signal, was used as the primary stage sensor. The ra-
dar distance meter, which could not be sampled as fre-
quently, was used to provide a periodic check on the
laser measurement. The two measurements of stage
were found to be in reasonable agreement (within 5
cm) for both debris flows and water-dominated flows.
Discrepancies between the two sensors were due pri-
marily to different sampling footprints (~3-mm diam-
eter for the laser versus an 8 degree cone for the radar,
which, at 2-m above the bed, corresponds to a 28-cm
diameter footprint). The datum for the stage gages
was taken to be the level of a vented pore-pressure
sensor that was cemented into the bedrock beneath
the stage gages. After installation, the sensor was bur-
ied with 48 cm of sediment to the original bed level.
The stage and pore-pressure sensors, which were
located in the partially burned valley bottom, were
connected via cables to a data logger and power sup-
ply situated 50 m above the channel on the unveg-
etated hillslope (Fig. 1). A tipping-bucket rain gage
and two soil-moisture probes were installed at the
data logger site. The soil-moisture probes were bur-
ied 5-cm deep on a 20 degree section of hillslope.
The laser, pore pressure, and rain gages were sam-
pled at high rates only during rain events (Tab. 2).
The soil-moisture and radar sensors were sampled
continuously at 1-min. intervals. A second tipping-
bucket rain gage was installed near the ridge of
the study basin. The ridge rain gage often (but not
always) measured less rainfall than the valley rain
gage. This difference may have been related to sys-
tematically higher wind velocities at the ridge, which
Tab. 2 - Sensor specifications*
background image
DIRECT MEASUREMENTS OF THE HYDROLOGIC CONDITIONS LEADING UP TO
AND DURING POST-FIRE DEBRIS FLOW IN SOUTHERN CALIFORNIA, USA
Italian Journal of Engineering Geology and Environment - Book www.ijege.uniroma1.it © 2011 Casa Editrice Università La Sapienza
689
because it is used more commonly in the post-
fire literature. Field visits to the site were made
following each storm to download data, check
instruments, and document the geomorphic re-
sponse. The latter was done by hiking the chan-
nel between its intersection with Highway 2 and
the main study site. Evidence of debris-flow de-
posits were noted along the way. In many areas,
including at the instrumented cross-sections, de-
bris-flow deposits were not preserved because
the adjacent banks were too steep/confined
to support a deposit and/or the deposits were
washed away by subsequent more watery flows.
RESULTS AND DISCUSSION
Four storms produced debris flows at or
downstream of the monitoring site. The first
was a localized thunderstorm that occurred on
12 Nov. 2009. The hillslope response to this
storm is discussed in detail in s
CHmidt
et alii
(this volume). The last three were extensive
frontal storms that occurred on 12 Dec. 2009,
18 Jan. 2010, and 6 Feb. 2010. All storms
caused at least temporary closure of Highway
2. At the time of this writing (Jul. 2010) the
highway remains closed due to damage from
the 6 Feb. 2010 event. Time series plots of
data from each storm are shown in Fig. 2-5.
Each figure contains multiple panels with time
in Pacific Standard Time. The first two panels
(A and B) show long-duration time windows
of rain, stage, and pore pressure data. The last
panels (C-E) show details of stage data for se-
lected short-duration time windows together
with the value of σ during that time.
RAINFALL
Time series of 10- and 60-min. intensi-
ties for the two storms are shown in panel A
of Fig. 2-5. Cumulative rainfall from the start
of the storm is shown with the dotted lines in
the B panels. In almost all cases, distinct peri-
ods of flow, which are characterized by clusters
of peaks in flow stage or pore pressure, occur
within a few minutes of a local peak in the 10-
min. intensity. In contrast, the much broader
peaks in 60-min. intensities typically lag peaks
in flow activity by 10 min. or longer. The em-
Fig. 2 - Sensor data for Storm 1 (12 Nov. 2009). (A) I
10
and I
60
(solid
dark and light blue, respectively), and corresponding rainfall
intensity-duration thresholds from c
ANNoN
et alii (2008) (T
10
,
dotted dark blue line; T
60
, dashed light blue line). (B) Stage
(H), and storm cumulative rainfall (R, dotted blue) for 70 min.
The datum for stage and pressure head is the level of the pres-
sure transducer. Flow periods beneath brackets “C” and “D”
are shown in detail in the next panels. (C-D) Stage during
5-min. time windows. The stage scale is 0.3 m. The standard
deviation (σ ) of the data about a 2-sec. running average is
given for each 5-min. time window
Fig. 3 - Sensor data for Storm 2 (12 Dec. 2009). (A) See caption Fig. 2A.
(B) Stage (H), pore pressure head (P, orange line), and storm
cumulative rainfall (R, dotted blue) for 12 hrs. (C-D) Stage and
σ during 20-min. time windows. The stage scale is 0.4 m
background image
J.w. kEAN & D.M. STALEY
690
5th International Conference on Debris-Flow Hazards Mitigation: Mechanics, Prediction and Assessment Padua, Italy - 14-17 June 2011
pirical rainfall-intensity duration thresholds of
C
annon
et alii (2008) are shown in the A panels
of Fig. 2-5 for comparison. These thresholds,
which were developed for this area prior to this
study, are defined by the equation I = 7.2 D
-0.4
.
STAGE
The plots of stage shown in Fig. 2-5 show
a complex and evolving flow response. The
first storm on 12 Nov. 2009 lasted 1 hour (Fig.
2). Flow began abruptly at 22:28, a few min-
utes after a local peak in I
10
and only 15 min.
after the onset of rain. The flow front (Fig. 2C)
had the distinct asymmetric shape characteris-
tic of debris-flow surges. This shape is caused
by a longitudinal non-iformity in resisting
stresses, which is produced by the hinter-
granular friction of coarser grains at the front
followed by lowerore watery recessional flow
(H
unGR
, 2000). Although debris-flow deposits
were observed downstream of the station, no
evidence of debris-flow deposits was found at
the station. It is unlikely, however, that levees
could be preserved at that levees could be pre-
served at that location as steep banks confined
flow and precluded deposition. Also, intense
rain at the end of the storm could have washed
away evidence of thin mud deposits. Stage
and video data of surges recorded at a nearby
monitoring site during a different storm pro-
vide indirect support of our conclusion that
the surge shown in Fig. 2C is a small debris
flow. This data had surges of similar shape
and video confirmed the flows were debris
flows with large clasts in the front followed
by water-rich recessional flows. Towards the
end of the storm, a brief high intensity burst
of rain (I
10
= 60 mm/hr) produced a second
major pulse of flow (Fig. 2D). This flow peak
was more symmetrical than the initial surge. It
also had greater variability in stage: the value
of σ increased from 4.2 mm during the ini-
tial surge to 10.2 mm during the second pulse
(Fig. 2C and 2D). The increase in σ is likely
the result of an increase in the percentage of
water in the flow, which tends to produce a
more agitated flow surface with larger, high
frequency (> 0.5 Hz) stage fluctuations.
Fig. 4 - Sensor data for Storm 3 (18 Jan. 2010). Notation and time
scales same as Fig. 3. The stage scale in (C-E) is 0.5 m
Fig. 5 - Sensor data for Storm 4 (6 Feb. 2010). Notation and time
scales same as Fig. 3. The stage scale in (C-E) is 1 m
background image
DIRECT MEASUREMENTS OF THE HYDROLOGIC CONDITIONS LEADING UP TO
AND DURING POST-FIRE DEBRIS FLOW IN SOUTHERN CALIFORNIA, USA
Italian Journal of Engineering Geology and Environment - Book www.ijege.uniroma1.it © 2011 Casa Editrice Università La Sapienza
691
following this front, as well as the flows during the two
later bursts of rain (Figs. 5 d-e), have σ characteristics
similar to the more watery flows of the latter half of
the 18 Jan. 2010 storm. This similarity suggests that the
water content of the flows during the 6 Feb. 2010 storm
was relatively high and further support our hypothesis
that the transition in primary flow response midway
through the 18 Jan. 2010 storm is related to sediment
supply. However, despite the fact the three flow peri-
ods during the 6 Feb. 2010 storm have the rapid stage
fluctuations characteristic of high water content flows,
it is not possible from the stage records alone to identify
whether the flows after the initial 3:17 surge are water
floods, hyper-concentrated flows, or debris flows. Post-
event field observations of debris-flow deposits down-
stream of the site and major debris-flow damage to
Highway 2 suggest that the measured last two pulses of
flow (Fig 5d and 5e) either were debris flows or devel-
oped into debris flows downstream of the site. In fact,
the regional debris-flow response during this storm was
the largest of the season and damaged or destroyed 41
homes (k
im
et alii, 2010). The most common character-
istic shared by the plots in Fig. 2-5 is the close timing
between flow periods and local peaks in 10-min. rainfall
intensity. This connection provides strong evidence that
the debris flows in the main study area were initiated by
surface-water runoff rather than by landslide failure by
infiltration processes, which operate typically on longer
time scales. This conclusion is supported by post-event
field observations, which documented extensive rilling
on hillslopes (Fig. 6), substantial channel erosion (Fig.
7), and the absence of new landslide scars.
Stage measurements from the second storm on 12
Dec. 2009 are shown in Fig. 3. This storm contains
seven major periods of flow, the first of which begins
at 14:48. Within each period of flow there are 4 to 30
steep-fronted surges that range in height from 5 to 25
cm (Fig. 3 C, D, and E). Flow periods between those
highlighted in Fig. 3C and 3D, which are not shown
in detail for space considerations, have similar surge
frequency and σ characteristics as the C and D periods.
In contrast, the last flow period (Fig. 2E) has a distinct-
ly higher average surge frequency (35 seconds apart
versus 220 seconds apart). This set of surges comes
during the most intense rain of the storm. Possible rea-
sons for the increase in surge frequency near the end of
the storm include: existence of a more well-developed
drainage network, a potential increase in the number
of active overland flow pathways on the hillslopes as
a result of the higher rain intensity, and/or greater flow
velocities due to increased water content. Additional
insight into this phenomenon may be gained by analy-
sis of infiltration data and pre- and post-event terres-
trial LiDAR surveys, which is currently underway.
The third storm on 18 Jan. 2010 (Fig. 4) also con-
tained multiple periods of flow, three of which are
shown in detail in Fig. 4C, D, and E. The first surges
(Fig. 4C) began at 9:20 and had similar characteris-
tics to the ones on 12-13 Dec. 2009. At approximately
11:50 the character of the flow changes substantially
(Fig. 4D), and the value of σ increased an order of mag-
nitude from 2 mm to 25 mm (Fig. 4D). The flow in this
time period is no longer characterized by well-defined
surges, but rather by high-frequency fluctuations about
roughly symmetric, lower-frequency trends in stage.
Later in the storm (Fig. 4E), the flow has similar high-
frequency fluctuations about a near steady flow. The
change in flow characteristics is likely due to a transi-
tion from debris flow to more water-dominated flow, as
a result of a change in hillslope and channel sediment
supply. This conclusion is supported by field observa-
tions after the 18 Jan. 2010 storm, which showed that
large portions of the main channel and many rills up-
stream of the site were scoured to bedrock.
Stage data from the fourth storm on 6 Feb. 2010,
is shown in Fig. 5. This storm had the highest recorded
rain intensities of the season and produced three major
periods of flow, which are shown in detail in Figs. 5c-e.
The first flow period began at 3:17 with a distinct 0.8 m
high debris-flow front (Fig. 5c). The flow immediately
Fig. 6 - Photograph of hillslope erosion after the 12 Nov.
2009 storm (Storm 1). The largest burned shrub in
the photo is about 3-m high
background image
J.w. kEAN & D.M. STALEY
692
5th International Conference on Debris-Flow Hazards Mitigation: Mechanics, Prediction and Assessment Padua, Italy - 14-17 June 2011
PORE PRESSURE
Measurements of pore pressure head (m of water
depth) are shown in panel B of Fig. 3-5. No pore pres-
sure changes were detected during the first short storm
on 12 Nov. 2009. Throughout the entire second storm
on 12 Dec. 2009 the pore pressure sensor is buried
beneath at least 20 cm of sediment. At the time of the
first surge at 14:48 the pore pressure sensor indicates
the water table is about 3 cm below the bed surface.
After the first surge, pore pressure generally tracked
changes in flow stage; however, the pressure response
was muted throughout the entire event, because the
bed was not completely saturated. m
C
C
oy
et alii
(2010; this volume) observed similar responses dur-
ing debris flows measured at the Chalk Cliffs monitor-
ing site. Unsaturated bed conditions also were present
during the third storm on 18 Jan. 2010 up until about
11:50 (Fig. 3b). The bed nearly reached saturation at
8:00 after about 25 mm of rainfall, as can be seen by
the close agreement between the stage and pressure
head. The muted signal of the pore pressure during
the first several sets of surges (9:20 to 11:50) indicates
the bed did not reach complete saturation until about
11:50. At this time there is a rapid increase in pressure
that eventually becomes greater than hydrostatic.
The transition of the pressure signal around 12:00
on 18 Jan. 2010 from less than hydrostatic to greater
than hydrostatic is probably associated with erosion
of the bed material above the sensor, which at least
removed the unsaturated layer that previously damped
the signal. After about 12:10, the pore pressure head
remains consistently above the stage and has occa-
sional spikes that are several times hydrostatic. Ex-
cess pore pressures have been observed during both
field (m
C
a
Rdell
et alii, 2007; m
C
C
oy
et alii, 2010)
and large-scale experimental debris flows (i
veRson
et
alii, 2010). However, given the stage characteristics of
the flow during this time (e.g. Fig. 4E), which are not
characteristic of debris flow, it is more likely that the
non-hydrostatic pressures measured here are caused
by accelerating water-dominated flow. Non-hydro-
static conditions in water flows are common in steep
irregular channels, such as the one at the station (d
en
-
linGeR
& o’C
onnell
, 2008). The fact that the spikes
in pressure often coincide with local peaks in I
10
sug-
gests the discharge during these times was very high;
the fact that the stage also was low implies the flow
velocity was very fast. Similar high pore pressures
were observed during the 6 Feb. 2010 event (Fig. 5).
IMPLICATIONS FOR WARNING AND
MODELLING
The rapid response to rainfall in the 0.01 km
2
ba-
sins show that real-time data from post-fire debris-flow
monitoring stations cannot provide sufficient warning
of impending debris flows and floods for communi-
ties downstream. Instead, warnings with practical lead
times of several hours must come from a combination
of weather forecasts, rainfall measurements of ap-
proaching storms, and debris-flow triggering thresh-
olds, such as a rainfall-intensity duration threshold
used in the current southern California debris-flow
warning system. In addition, the transition of the domi-
nant flow from debris flow to flood, which occurred
during the 18 Jan. 2010 storm under similar rain con-
ditions, demonstrates the difficulty of differentiating
warnings for debris flows from those for floods.
A comparison of rainfall thresholds developed by
C
annon
et alii (2008) with the data in Fig. 2-5, show
that measured rainfall intensities exceeded the thresh-
olds during most of major flow periods. Although
rain data collected at the 2009 monitoring sites was
used to slightly refine these thresholds (adjusting
the thresholds upwards by 6 percent, C
annon
et alii,
2010a), it is unlikely availability of additional data
Fig. 7 - Photograph of channel erosion after the 18 Jan.
2010 storm (Storm 3). Geologist for scale in
center of photo. Much of the channel bed in the
foreground is bedrock
background image
DIRECT MEASUREMENTS OF THE HYDROLOGIC CONDITIONS LEADING UP TO
AND DURING POST-FIRE DEBRIS FLOW IN SOUTHERN CALIFORNIA, USA
Italian Journal of Engineering Geology and Environment - Book www.ijege.uniroma1.it © 2011 Casa Editrice Università La Sapienza
693
SUMMARY AND CONCLUSIONS
This paper has presented initial results from the first
post-fire debris flows to be recorded by in-situ monitor-
ing equipment. The flows were generated by surface-
water runoff, and the timing of flows corresponded most
strongly with short duration (~10-min.) bursts of rain.
High frequency (10 Hz) stage data from a laser distance
meter with a small beam footprint was used to provide
a qualitative measure of the water content of the flows.
The addition of normal stress sensors and video cam-
eras at future monitoring sites may permit determination
of flow-density time series that could better distinguish
the flow response (debris flow, hyper-concentrated flow,
water flood). Approximately halfway through the third
winter storm of the season, the primary flow response in
the small, 0.01 km
2
basin transitioned abruptly from de-
bris flow to more water-dominated flow, and this transi-
tion is likely related to a decrease in the sediment availa-
bility in the hillslopes and channels. Continued post-fire
debris-flow monitoring of this type should provide de-
tailed documentation of the varied flow responses that
can occur in post-fire settings, and this, in turn, will help
improve predictions of post-fire debris flows.
ACKNOWLEDGEMENTS
Land access was granted by the City of Pasade-
na. Robert Leeper provided field assistance. Helpful
reviews by Sue Cannon, Scott McCoy, and Kevin
Schmidt greatly improved the manuscript.
would produce major revisions to the current thresh-
olds. Despite the fact that the current regional rain
thresholds are about as good as they can be, addition-
al data, such as that presented here, may help define
alternative basin-specific debris-flow triggering con-
ditions (e.g. extent and magnitude of surface water
runoff), which could be implemented in a warning
system. In the meantime, this data, together with the
data from the other four monitoring sites, shows that
closer attention should be paid to short duration rain
intensities (e.g. 10 min.), which are connected most
closely to the events, but have received less attention
in the post-fire literature than longer durations (see
review in C
annon
et alii, 2008). This also increases
the importance of rainfall monitoring “upwind” of
the burn area, which can verify high intensity cells in
advancing storms before they are over the burn area.
In regards to modelling, the high correlation be-
tween short-duration high-intensity rainfall and debris
flows suggests that a simple rainfall runoff model may
be a useful tool to predict the timing of post-fire flash
flood and debris-flow from low-order basins. Howev-
er, without the addition of physically based models for
initiation (i.e. how surface-water flow becomes debris
flow), entrainment, and routing, such models will not
reproduce the complex stage changes of the debris-
flow surges shown in Fig. 2-5, nor predict accurately
the peak stage of debris flows, which are likely to be
much higher than flood peaks (H
unGR
, 2000).
REFERENCES
b
eRti
m., G
enovois
R., l
a
H
usen
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: 707-715,
doi: 10.1016/S1464-1909(00)00090-3.
C
annon
s.H., k
iRkHam
R.m., & P
aRise
m. (2001) - wildfire-related debris-flow initiation processes, Storm king Mountain,
Colorado, Geomorphology, 39: 171-188.
C
annon
s.H., G
aRtneR
J.e., w
ilson
R.C., b
oweRs
J.C., & l
abeR
J.l. (2008) -Storm rainfall conditions for floods and debris
flows from recently burned areas in southwestern Colorado and southern California, Geomorphology, 96: 3-4, 250-269.
C
annon
s.H., b
oldt
e.m., k
ean
J.w., l
abeR
J.l., & s
taley
d.m. (2010a) - Relations between rainfall and postfire debris-flow
and flood magnitudes for emergency-response planning, San Gabriel Mountains, southern California: U.S. Geological
Survey Open File Report 2010-1039, 31 pp.
C
annon
s.H., G
aRtneR
J.e., R
uPeRt
m.G., m
iCHael
J.a., R
ea
a. H., & P
aRRett
C. (2010b) -Predicting the probability and volume of
postwildfire debris flows in the intermountain western United States, GSA Bulletin, 122: 1/2,127-144, doi: 10.1130/B26459.1.
C
annon
s.H., G
aRtneR
J.e., R
uPeRt
m.G., m
iCHael
J.a., s
taley
d.m., & w
oRstell
b.b. (2010c) - Emergency assessment of
postfire debris-flow hazards for the 2009 Station fire, San Gabriel Mountains, southern California: U.S. Geological Survey
Open-File Report 2009-1227, 27 pp.
C
HawneR
w.d. (1934) - The Montrose-La Crescenta (California) flood of January 1, 1934, and its sedimentary aspects,
California Institute of Technology, Master’s thesis, 78 pp.
background image
J.w. kEAN & D.M. STALEY
694
5th International Conference on Debris-Flow Hazards Mitigation: Mechanics, Prediction and Assessment Padua, Italy - 14-17 June 2011
d
enlinGeR
R.P. & o’C
onnell
d.R.H (2008) - Computing nonhydrostatic shallow-water flow over steep terrain, Journal of
Hydraulic Engineering, 134: 11, 1590-1602.
e
aton
E.C. (1936) - Flood and erosion control problems and their solution, Proceedings of the American Society of Civil
Engineers, 62: 8, 1302-1362.
H
unGR
O. (2000) - Analysis of debris flow surges using the theory of uniformly progressive flow, Earth Surface Processes and
Landforms, 25: 483-495.
i
veRson
R. m., l
oGan
m., l
a
H
usen
R.G., & b
eRti
m. (2010) - The perfect debris flow? Aggregated results from 28 large-scale
experiments, Journal of Geophysical Research, 115: F03005, doi:10.1029/2009JF001514.
k
im
v., v
ives
R., R
onG
-G
onG
l., & G
ottlieb
J. (2010) - Mandatory evacuations ordered for 500 homes in La Canada Flintridge,
La Crescenta and Acton, Los Angeles Times, 6 February 2010.
m
aRCHi
l., a
Rattano
m., & d
eGanutti
a. (2002) - Ten years of debris-flow monitoring in the Moscardaro Torrent (Italian Alps),
Geomorphology, 46: 1-17, doi: 10.1016/S0169-555X(01)00162-3.
m
C
a
Rdell
b.w., b
aRtelt
P., & k
owalski
J. (2007) - Field observations of basal forces and fluid pore pressure in debris flow,
Geophysical Research Letters, 34: L07406, doi: 10.1023/A:1008064220727.
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, 38-8: 735-738, doi: 10.1130/G30928.1.
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
C
P
Hee
, J.a. (1989) - The Control of Nature, Farrar, Straus and Giroux, New York, 272 pp.
m
eyeR
G.a., & w
ells
s.G. (1997) - Fire-related sedimentation events on alluvial fans, Yellowstone National Park, U.S.A.,
Journal of Sedimentary Research, 67(5): 776-791.
NOAA-USGS Debris Flow Task Force (2005) - NOAA-USGS debris-flow warning system - Final report: U.S. Geological Survey
Circular 1283: 47 pp.
s
CHmidt
, k.m., H
ansHaw
m.n., H
owle
J.f., k
ean
J.w., s
taley
d.m., s
toCk
J.d., b
awden
G.w. (This volume) - Hydrologic
conditions and terrestrial laser scanning of post-fire debris flows in the San Gabriel Mountains, CA, USA.
s
Cott
, k.m. (1971) - Origin and sedimentology of 1969 debris flows near Glendora, California, U.S. Geological Survey
Professional Paper 750-C: C242-C247.
s
HuRiman
G., & s
lossen
J.e. (1992) - Forensic engineering - Environmental case histories for civil engineers and geologists,
chap 8, Academic Press, Harcourt Brace Javanovich, San Diego, p. 168-209.
Station Fire BAER Team (2009) - Station Fire Burned Area Emergency Response (BAER) Report, FS-2500-8, U.S. Forest
Service, 22 pp.
s
uwa
H., o
kano
k., & k
anno
t. (2009) - Behavior of debris flows monitored on test slopes of kamikamihorizawa Creek, Mount
Yakedake, Japan, International Journal of Erosion Control Engineering, 2-2: 33-45.
w
esteRlinG
a.l., H
idalGo
H.G., C
ayan
d.R. & s
wetnam
t.w. (2006) - warming and earlier spring increase western U.S. forest
wildfire activity, Science, 313: 940-943., doi: 10.1126/science.1128834.
y
eRkes
R.f., & C
amPbell
R.H. (2005) - Preliminary geologic map of the Los Angelse 30’x60’ Quadrangle, Southern California,
U.S. Geological Survey Open-file Report 2005-1019.
Statistics