Document Actions

IJEGE-11_BS-Schmidt-et-alii

background image
Italian Journal of Engineering Geology and Environment - Book www.ijege.uniroma1.it © 2011 Casa Editrice Università La Sapienza
583
DOI: 10.4408/IJEGE.2011-03.B-064
HYDROLOGIC CONDITIONS AND TERRESTRIAL LASER SCANNING
OF POST-FIREDEBRIS FLOWS IN THE SAN GABRIEL MOUNTAINS, CA,
U.S.A
k.m. sCHmidt
(*)
, M.N. HANSHAW
(*)
, J.F. HOWLE
(**)
, J.W. KEAN
(***)
, D.M. STALEY
(***)
,
J.D. STOCK
(*)
& w. bawdenG w.
(****)
(*)
U. S. Geological Survey, 345 Middlefield Rd., MS-973, Menlo Park, CA, 94025, U.S.A
(**)
U. S. Geological Survey, P.O. Box 1360, Carnelian Bay, CA, 96140, U.S.A.
(***)
U. S. Geological Survey, P.O. Box 25046, MS-966, Denver, CO, 80225, U.S.A
(****)
U. S. Geological Survey, 3020 State University Dr. E., Sacramento, CA 95819, U.S.A.
spread erosion of the valley axis with ground surface
lowering exceeding 1.5 m.
K
ey
words
: rainfall-runoff, debris flow, rainfall threshold,
terrestrial laser scanning, lidar, wildfire, warning system
INTRODUCTION
Wildfire in the chaparral ecosystem of southern
California, U.S.A. along the urban-wildland inter-
face has become a commonplace occurrence and is
often ignited by anthropogenic activities (k
eeley
et
alii, 1999). In the aftermath of wildfire, increased
sediment transport continues into the following win-
ter months and intense rainfall may trigger floods and
debris flows that threaten life and property of the com-
munities at the base of steep, burned watersheds. To
evaluate these hazards, the U. S. Geological Survey
(USGS) and the National Weather Service (NWS), a
division of the National Oceanic and Atmospheric Ad-
ministration (NOAA), have collaborated since 2005
on a demonstration flashflood and debris-flow early
warning system for recently burned areas in southern
California (NOAA-USGS Debris Flow Task Force,
2005). Goals of this warning system are to predict
the timing, location, and relative magnitude of rapid
sediment-laden flash floods and debris flows from re-
cently burned drainage basins. Comparisons between
forecast and observed precipitation with empirical
rainfall thresholds for the initiation of debris flows
(e.g., C
annon
et alii, 2008), as well as maps show-
ABSTRACT
To investigate rainfall-runoff conditions that
generate post-wildfire debris flows, we instrumented
and surveyed steep, small watersheds along the tec-
tonically active front of the San Gabriel Mountains,
California. Fortuitously, we recorded runoff-generat-
ed debris-flows triggered by one spatially restricted
convective event with 28 mm of rainfall falling over
62 minutes. Our rain gages, nested hillslope over-
land-flow sensors and soil-moisture probes, as well
as a time series of terrestrial laser scanning (TLS)
revealed the effects of the storm. Hillslope overland-
flow response, along two ~10-m long flow lines per-
pendicular to and originating from a drainage divide,
displayed only a 10 to 20 minute delay from the on-
set of rainfall with accumulated totals of merely 5-10
mm. Depth-stratified soil-moisture probes displayed a
greater time delay, roughly 20- 30 minutes, indicating
that initial overland flow was Hortonian. Furthermore,
a downstream channel-monitoring array recorded a
pronounced discharge peak generated by the passage
of a debris flow after 18 minutes of rainfall. At this
time, only four of the eleven hillslope overlandflow
sensors confirmed the presence of surface-water flow.
Repeat TLS and detailed field mapping using GPS
document how patterns of rainsplash, overland-flow
scour, and rilling contributed to the generation of me-
ter-scale debris flows. In response to a single small
storm, the debris flows deposited irregular levees and
lobate terminal snouts on hillslopes and caused wide-
background image
k.M. SCHMIDT, M.N. HANSHAw, J.F. HOwLE, kEAN, J.w., D.M. STALEY, J.D. STOCk & w. BAwDENG
584
5th International Conference on Debris-Flow Hazards Mitigation: Mechanics, Prediction and Assessment Padua, Italy - 14-17 June 2011
cate debris basin system was initiated and managed
by the Los Angeles County Dept. of Public Works
to capture sediment (http://dpw.lacounty.gov/wrd/
sediment/debris_basin_clean_out.cfm). However, the
intimate proximity of steep topography and numer-
ous debris basins adjacent to the urban environment
dictates frequent and timely maintenance to evacuate
trapped sediment. The Station fire burn area, just north
of the city Los Angeles, California, U.S.A. was the 10th
largest wildfire in California history since 1933, burn-
ing a total of 650 km2 of the San Gabriel Mountains
from 26 August to 16 October 2009 (http://cdfdata.fire.
ca.gov/incidents/). The fire consumed chaparral and
mixed coniferous forests within landslide-prone rug-
ged watersheds draining to steep, urbanized alluvial
fans. Advisories, watches, and warnings based upon
precipitation forecasts provided emergency-response
and public-safety agencies information necessary for
urgent response decision-making efforts (C
annon
et
alii, 2010). In addition, C
annon
et alii (2009) presented
an emergency assessment of potential debris-flow haz-
ards based upon statisticalempirical models for use by
civil authorities. As part of the USGS emergency re-
sponse, we instrumented sites in an Intensive Research
Area (IRA), within watersheds identified by C
annon
et
alii (2009) to have high debris-flow susceptibility. On
12 November 2009 a focused convective rain cell pro-
duced debris flows in our IRA. This paper documents
our observations, recorded rainfall-runoff data, and
changes in topography derived from repeat TLS
STUDY AREA
The San Gabriel Mountains are a prominent, tec-
tonically active range in a Mediterranean climate with
thoroughly dissected rugged topography, bordered to
the southwest by an urban fringe (Figure 1). The range
formed within a large restraining bend of a larger fault
system associated with the San Andreas fault, a major
continental transform fault. Our IRA field site is lo-
cated in topography generated by combined displace-
ment from the San Gabriel fault to the northeast, and
active thrusting of the Sierra Madre fault to the south-
west, with inferred vertical slip rates of 0.5-1.0 mm/yr
at the mountain front (l
indvall
& R
ubin
, 2003). The
field site is underlain by materially weak rock: late
Cretaceous hornblende diorite exposed to extensive
tectonic shear (s
mitH
, 1986) that is highly fractured,
friable, and chemically weathered. The soil and much
ing areas susceptible to debris flow (e.g., C
annon
et
alii, 2009) are used to advise the affected populace
and civil authorities. In an effort to better constrain
the rainfall-runoff conditions leading to initiation of
runoff-generated debris-flows and to explore the fea-
sibility of linking hillslope and channel monitoring
information with the warning system, we installed
ground-based monitoring instruments including rain
gages, soil-moisture probes, overland-flow sensors,
sediment traps, and erosion pins and conducted repeat
terrestrial laser scanning (TLS) in smaller watersheds
within six different fires since 2005.
Although the transport and depositional zones of
debris flows have been widely studied (Iverson et al.,
1997), the hydrologic conditions and topographic lo-
cations in which debris flows initiate within burned
landscapes remain poorly defined. In contrast to in
situ discrete hillslope-scale landslides mobilizing
downslope into debris flows (i
veRson
et alii, 1997),
runoff-generated debris flows occur typically in areas
of steep topography devoid of vegetation with spatial-
ly distributed transport of ample loose sediment (e.g.,
C
annon
et alii, 2003; Coe et alii, 2008). In burned
landscapes, overland flow is commonly observed dur-
ing relatively low-intensity rainfall events. In contrast,
undisturbed forest and chaparral environments with
intact organic duff layers above the mineral soil sur-
face have minimal overland flow during low intensity
rainfall. In burned areas, soil infiltration capacity is
reduced by a complex combination of factors includ-
ing the removal of organic material, fusing of soil
particles into aggregates, fire-induced hydrophobicity,
extreme drying of soil, and clogging of pore spaces by
ash (e.g., s
Hakesby
& d
oeRR
, 2006). This reduction of
infiltration capacity can amplify overland flow.
Early recognition of the connection between post-
fire conditions and rainfall generating debris flows
in the region can be traced to the devastating debris
flows and loss of life that occurred in the La Crescenta
- Montrose, CA communities in the 1934 New Year's
Day Storm (t
ayloR
, 1934; e
aton
, 1936) following
fires in 1933. Knowledge regarding debris flows at
the time was limited, and for the same event t
Roxell
& P
eteRson
(1937) described the event not as debris
flows, but as alluvial flows of traction transport with
“walls of water” up to 3-m high, although current in-
terpretations support the former. In response to large
sediment transporting events such as these, an intri-
background image
HYDROLOGIC CONDITIONS AND TERRESTRIAL LASER SCANNING OF POST-FIREDEBRIS FLOWS IN THE SAN GABRIEL
MOUNTAINS, CA, U.S.A
Italian Journal of Engineering Geology and Environment - Book www.ijege.uniroma1.it © 2011 Casa Editrice Università La Sapienza
585
of the rock exposed at the ground surface had similar
strength properties and grain size distributions, and
hence abundant unconsolidated sediment was avail-
able for transport. The colluvial soils, characteristi-
cally less than 0.5 m thick, locally expressed weak
pedogenic horizonation. Transport-limited conditions
dominated after the fire, prior to the onset of win-
ter rains. Pre-fire vegetation was composed of hard
chaparral species assemblages dominated by chamise,
ceanothus, mountain mahogany, manzanita, and yuc-
ca. Mean annual precipitation in this Mediterranean-
type climate ranges between ~700-800 mm/yr (l
avé
& b
uRbank
, 2004).
To examine the hydrologic conditions leading to
debris-flow generation, we selected a burned, domi-
nantly soil-mantled, unchannelized low-order valley
(Lukens IRA, Figg. 1 and 2) in the Station burn area
that was hydrologically unaffected by nearby roads.
The watershed had convergent steep slopes >30° with
Fig. 1 - Regional study area showing part of the area burned by the Station fire (partial burn perimeter shown as black
line) California, U.S.A and Lukens IRA watershed superimposed with November 12th, 2009, rainfall totals from (a)
NEXRAD radar data, and (b) cumulative tipping-bucket rain gage totals (mm) with height of bar scaling with rain
total [locations denoted by white triangles in (1a)]. Letters c-f indicate the location of the photos from this event: (c)
~1.2-m thick debris flow deposited across the Angeles Crest Highway, (d) U.S. Forest Service vehicle inundated by
sediment at the Angeles Crest Fire Station, (e) Mullally Canyon debris basin excavation after reaching full capacity
(~7190m3), and (f) debris-flow deposit composed of ample dark, burned material
Fig. 2 - Lukens IRA sub-watershed with instrument
locations on hillshade derived from airborne
lidar. Two surface-flow lines (Saddle and
Ridge) were monitored near the drainage di-
vide. Rain gage (Lukens IRA) also denotes lo-
cation of channel-monitoring array
background image
k.M. SCHMIDT, M.N. HANSHAw, J.F. HOwLE, kEAN, J.w., D.M. STALEY, J.D. STOCk & w. BAwDENG
586
5th International Conference on Debris-Flow Hazards Mitigation: Mechanics, Prediction and Assessment Padua, Italy - 14-17 June 2011
or to rainfall and data was retrieved the day following
the event. At this site, and within other catchments of
suspected high debris-flow susceptibility in the Sta-
tion burn area, instrumental records were recorded by
either non-telemetered, temporally synchronized data
loggers or in near realtime telemetered data streams
(e.g., http://landslides.usgs.gov/monitoring).
We also surveyed high-resolution (sub-cm-scale
laser point spacing) topography using TLS of the
entire Lukens IRA subwatershed on 12 and 13 No-
vember 2009, only hours prior to and immediately
following the storm, to record the temporal and spatial
distribution of surface erosion by overland flow and
sediment transport by debris flows. Bare-earth digital
elevation models (DEMs) were generated from scans
obtained from three laser instrument set ups on 12
November and two set ups on 13 November. The raw
data were processed in PolyWorks and TerraScan to
remove vegetation. Bare-earth DEMs, with a 0.02-m
cell spacing, were krigged in Surfer and imported into
ArcGIS to define the spatial projection, merged if til-
ing was necessary due to large file sizes, and georef-
erenced using differential, post-processed kinematic
GPS locations of four fixed location monuments vis-
ible in the scans. Topographic and hydrologic routing
analyses of TLS DEMs were performed using Arc-
GIS. Post-storm observations of runoff characteris-
tics were located using post-processed, differential
kinematic GPS. We identified evidence of debris-flow
generation by visually inspecting deposits for matrix
support, a lack of stratification, and an absence of tex-
tural sorting. We then walked upslope from identified
debris-flow deposits, tracing evidence to its farthest
upslope extent, and examined the ground surface for
indications of triggering mechanism.
RESULTS
Prior to the onset of 12 November 2009 rain, sedi-
ment transport in the IRA watershed was dominated
by post-fire dry ravel. Beginning on 6 October 2009,
we observed periodically active sediment transport
of dry-ravel material on the planar and convex por-
tions of the topography and accumulating into the
convergent valleys (Fig. 3a). The dry, principally un-
consolidated colluvium that had been stored upslope
of vegetation, which acted as local sediment storage
dams, began moving downslope from topographically
planar and divergent hillslopes toward convergent val-
ample post-fire ravel filling the valley axes (Figure 3a).
Burn severity at the field site was mapped as moderate
to high with almost all the ground and shrub vegeta-
tion consumed and lower (<10 m) tree canopies burned
above the ground. Most woody vegetation <5 mm di-
ameter was consumed by the fire. Relatively planar
topography below the convex drainage divide without
exposed bedrock, ideal for paired overland-flow and
soil-moisture monitoring in the upper reaches of the
watershed (Fig. 2). To constrain the timing and hy-
drologic conditions of coupled hillslope and channel
responses, downslope channel monitoring in the same
IRA (triangle in Fig. 2) is discussed by k
ean
& s
ta
-
ley
(this volume). Figure 2 illustrates the configuration
of our instrumented 12,680 m
2
south-facing, Lukens
IRA sub-watershed. The IRA sits within a larger, east-
facing 243,035 m
2
watershed with a ~1820 m3 debris
catch basin at the Angeles Crest Highway intersection.
METHODS
We measured rainfall-runoff relationships and
the geomorphic processes initiating sediment trans-
port, rilling, and debris flows on hillslopes, using
rainfall, overland-flow, and soil-moisture monitoring
equipment. In the upper watershed of our IRA, we in-
strumented 10-m long segment of two surface-water
flow lines (“Ridge” and “Saddle” in Figure 2), with
soil-moisture probes under the ground surface placed
near overland-flow sensors (OFS) to detect whether
surface-water flow generating mechanisms were satu-
ration driven and/or Hortonian. The soil-moisture
probes were installed >1.5 m to <6.1 m away from
the OFS placed on the ground surface. To minimize
disruption of surface-water flow paths during the dig-
ging of soil pits, probes were not installed directly un-
derneath the OFS. Soil-moisture probes were installed
by digging three pits to identify soil horizons, were
inserted roughly parallel to the ground surface, and
were vertically stratified. We installed a total of 6 OFS
(USGS-developed closed loop voltage conducting
circuits developed by John Moody, USGS, personal
communication, Figure 4) and 4 soil-moisture probes
used to infer volumetric water content (Decagon
ECH20 EC-5 sensor based on electrical conductivity)
on the Ridge flow line, and 5 OFS and 3 soil-moisture
probes on the Saddle flow line (Fig. 4). Rainfall was
recorded by an Onse RG3-M tipping-bucket rain
gage (0.2 mm/tip). All instruments were installed pri-
background image
HYDROLOGIC CONDITIONS AND TERRESTRIAL LASER SCANNING OF POST-FIREDEBRIS FLOWS IN THE SAN GABRIEL
MOUNTAINS, CA, U.S.A
Italian Journal of Engineering Geology and Environment - Book www.ijege.uniroma1.it © 2011 Casa Editrice Università La Sapienza
587
ing deposits ~0.6-1.2 m thick. Numerous flows also
blocked both lanes of the Angeles Crest Highway
(e.g., Fig. 1c), but some stopped abruptly upon exiting
steep, confined valleys (Figure 1f). Figures 3 and 4
illustrate how surface morphology was altered at the
site by runoff-induced surface flow in response to this
short, intense rainfall burst.
An empirical intensity (I) - duration (D) thresh-
old curve derived prior to the storm (C
annon
et alii,
2010) indicated that for durations of 30- and 60-min-
utes, precipitation greater than 15 and 13 mm/hr re-
spectively, would produce several debris flows over a
localized area. During the 12 November storm, how-
ever, the peak 30- and 60- minute intensities measured
by the rain gage in the IRA exceeded 60 and 30 mm/hr
respectively. Figures 1 and 5a depict how cumulative
rainfall from the storm dramatically varied across the
leys following vegetation removal by wildfire. Prior
to rainfall, all pre-fire deposits in the valley axis were
uniformly covered by recent post-fire dry-ravel depos-
its (Fig. 3a).
STORM RAINFALL AND SEDIMENT TRANS-
PORT
Only a few hours following site installation and
TLS surveying, at ~22:00 (Pacific Standard Time:
PST) 12 November 2009, a spatially restricted convec-
tive cell produced a total of 28.4 mm of precipitation
during a 62 minute period over our monitoring site
(Figg. 1 and 5). The resulting sediment transport and
debris flows filled catch basins at watershed outlets
(e.g., Fig. 1e). Some catch basins were filled beyond
capacity, causing debris to overtop structures and spill
over onto roads and into residential properties, leav-
Fig. 3 - Photographic view to North of upper portion of Lukens IRA sub-watershed before rainfall (a), and after (b);
times in upper right
Fig. 4 - Before (a) and after (b) photographs of the overland-flow sensors (Saddle flow line). Note the rainsplash-
induced pockmarks and cm-scale width semicontinuousrill network in (b) extending to within 0.5-2.0 m of
the diffuse convex drainage divide
background image
k.M. SCHMIDT, M.N. HANSHAw, J.F. HOwLE, kEAN, J.w., D.M. STALEY, J.D. STOCk & w. BAwDENG
588
5th International Conference on Debris-Flow Hazards Mitigation: Mechanics, Prediction and Assessment Padua, Italy - 14-17 June 2011
mountain front, with the highest totals recorded at the
Lukens IRA, and how the NEXRAD radar data under-
represented the rainfall totals. According to the Na-
tional Weather Service, precipitation models did not
accurately predict the location or total rainfall of the
storm, and satellite imagery did not adequately char-
acterize rainfall totals during the storm (NWS WFO
Memorandum, Eric Boldt, personal communication)
and as a result, no public advisory was issued.
TIMING OF RAINFALL, OVERLAND FLOw,
SOIL MOISTURE,AND DEBRIS FLOw
Figures 5 and 6 reveal a short response time be-
tween the onset of rainfall and the inception of meas-
urable overland water flow. Overland-flow response
along the Saddle and Ridge flow lines, originating
from the drainage divide, was delayed only by 10 to
20 minutes after the onset of rainfall; accumulated
rain totals during this time were merely 5-10 mm.
The Ridge line sensors reacted first, preceding the first
Saddle line response by more than 8 minutes, likely
a reflection of topographic location within a larger
contributing drainage area, higher gradient of con-
tributing area, and possibly differences in hydrologic
response of surficial materials. Analysis of our TLS-
derived topography showed that the farthest upslope
sensor on the Ridge line had a drainage area of ~1-2
m
2
, compared to <1 m2 on the Saddle line. Within an
individual flow line there was no discernable downs-
lope increase in recorded flow depth over the 10-m
long interval. Post-storm erosional paths were ob-
served to be irregular, concentrated within anastomos-
ing networks (Fig. 4) that possibly varied in position
and discharge throughout the storm.
Relative to the rapid responses of the Ridge line
overland-flow sensors, responses of depth-stratified
soil-moisture probes displayed a greater delay after
rainfall onset, as much as 20-30 minutes (Fig. 6).
Only one soil-moisture probe, the shallowest inserted
at 4-cm below ground surface in the Saddle line (Fig-
ure 6e, probe 3-A), displayed a response beginning at
about the same time as the first Saddle-line overland
flow sensor response. This soil-moisture probe did not
appreciably respond until 7 mm of rain had fallen,
18 minutes after the onset of rainfall. Initial surface
runoff, then, was likely Hortonian in behavior, and
occurred during dry soil-moisture conditions. Early
storm rain intensities likely surpassed the infiltration
Fig. 5 - Rainfall (a: cumulative, b: instantaneous in-
tensity, and c: 15-minute normalized intensity)
recorded at 5 gages located in Figure 1. Verti-
cal line denotes debris-flow discharge peak in
channel monitoring data at Lukens IRA outlet
Fig. 6 - Lukens IRA rainfall (a), overland-flow depth
(b: Ridge flow line, c: Saddle flow line), and
volumetric water content, VwC (d: Ridge flow
line, e: Saddle flow line). Lines in (b) and (c)
become darker and thicker with increasing di-
stance from ridge top. In (d) and (e) lines be-
come darker and thicker with increasing depth
below the ground surface. Numbers (1, 2, and
3) in (d) and (e) indicate different soil-moisture
pits, letters (A, B, and C) indicate the soil ho-
rizon where probes were situated and depth
below ground surface. Vertical line denotes
debris-flow discharge peak in channel monito-
ring data at Lukens IRA outlet
background image
HYDROLOGIC CONDITIONS AND TERRESTRIAL LASER SCANNING OF POST-FIREDEBRIS FLOWS IN THE SAN GABRIEL
MOUNTAINS, CA, U.S.A
Italian Journal of Engineering Geology and Environment - Book www.ijege.uniroma1.it © 2011 Casa Editrice Università La Sapienza
589
phology and to document magnitudes and spatial pat-
terns of erosion and deposition. These TLS surveys
revealed that the average gradient of the watershed
was 39°, the maximum gradient was 88° in an area of
exposed bedrock, and the relief from the drainage di-
vide at the overland-flow lines was ~112- 118 m, with
a maximum of 130 m. Pre-storm topography reflected
a dominantly soil-mantled landscape with smooth, pla-
nar hillslopes on the west flank of the north-northwest
trending, valley axis, an east flank with more incised
bedrock-controlled sub-basins, and an unchannelized
valley axis filled principally by dry-ravel accumulation
(Fig. 3a). Our post-storm survey depicted topography
with a much higher drainage density and a well-incised
valley axis (Fig. 3b). Almost the entire area expressed
evidence of surface runoff by the ubiquitous presence
of small rills, but the incision depth of many rills on
the hillslopes was generally <5 cm. Field observations
following the event indicated that the granular materi-
als below depths of ~10 cm principally remained dry.
We estimated ground-surface elevation change by dif-
ferencing bareearth DEMs generated from the 12 and
13 November surveys (Fig. 7). Although much of the
survey extent exhibited little to no vertical (within 5
cm) ground surface change in the upper third of the
Lukens IRA watershed, pronounced incision occurred
at a smaller number of larger rills and valley incision
exceeded 1.5 m the main previously unchannelized
valley axis. Recently exhumed materials on the valley
floor consisted of colluvium, post-fire dryravel deposits
exposed on valley walls, bedrock, and saprolite. Hence,
the sediment sources entrained in the debris flows were
a combination of material stored in the channel before
the fire, postfire dry-ravel deposits, and minor bedrock
excavation. Excavations deeper than 1 m into sedi-
ment stored in the valley axis appeared discontinuous
and probably were limited by structurally controlled
bedrock depressions. On the previously smooth, planar
western watershed flank, overland flow incised wide-
spread sub-parallel, levee-lined rills. Levee-bounded
tracks that were not incised into hillslope materials,
with base levels of the pre-existing surface, were also
observed. Numerous sub-meter wide and decimeter-
thick debris flows deposited over ravel deposits or white
ash and char horizons.Hillslope debris-flow deposits,
incised by later stage fluvial erosion, ranged up to 3-m
in length and were characterized by non-stratified, non-
imbricated, matrix-supported coarse and fine-gravel
capacity of the near-surface hydrophobic A-horizon
soil, a trait supported by infiltration experiments. In
addition, droplet tests at random locations to identify
the presence of hydrophobic soils revealed widespread
and somewhat laterally continuous hydrophobicity.
In soil-moisture soil pits 1 and 3, increased volu-
metric water content progressed downward, consistent
with top-down infiltration. Shallow probes responded
prior to those inserted at deeper depths within the
same soil pit (Fig. 6). The deepest probe (29-cm depth
in saprolite, pit 3) responded 36 minutes after rainfall
began with over 13 mm of accumulated rain. This pat-
tern was not observed at pit 2, where no response was
recorded, perhaps reflecting the effectiveness of hy-
drophobicity near the ground surface impeding local
vertical infiltration.
We constrained the timing of sediment transport us-
ing data from overland-flow sensors located on planar
hillslopes just downslope of the drainage divide, and
from the channel monitoring discharge array. Channel
monitoring instruments, located at the watershed outlet
(12,680 m
2
drainage area denoted by the Lukens IRA
rain gage in Fig. 2), recorded an abrupt, pronounced dis-
charge peak, presumably of debris-flow origin, after 16
minutes of rainfall with a total accumulation of merely
6.5 mm (vertical line on Figures 5 and 6) (k
ean
& s
ta
-
ley
, this volume). At this time, only four of the eleven
overland flow sensors recorded measurable surface
water and the soil-moisture probes had not yet dem-
onstrated significant infiltration leading to increase soil
moisture. Based on the stratigraphy of deposits at the
Lukens IRA watershed outlet, we suggest the following
sequence of transport events: an initial wave of debris-
flow transport that was over-printed by more spatially
widespread fluvial (traction, saltation, and suspension)
transport later in the storm in response to higher rainfall
intensities. Debrisflow deposits in the valley axis and
on the hillslopes were truncated and appeared incised
by later stage overland flow and fluvial transport. As
such, the majority of the response recorded after 22:30
by the overland-flow sensors likely represents runoff
contributing to the post-debris-flow traction transport.
TERRESTRIAL LASER SCANNING AND FIELD
OBSERVATIONS
We scanned the Lukens IRA topography with a
terrestrial laser, hours preceding and following the
rainfall-runoff event, to characterize watershed mor-
background image
k.M. SCHMIDT, M.N. HANSHAw, J.F. HOwLE, kEAN, J.w., D.M. STALEY, J.D. STOCk & w. BAwDENG
590
5th International Conference on Debris-Flow Hazards Mitigation: Mechanics, Prediction and Assessment Padua, Italy - 14-17 June 2011
Employing differential, post-processed kinematic
GPS surveys, we mapped and compared geomorphic
transport boundaries within the pre- and post-storm
TLS-derived DEMs. Rainsplash and unchannelized
sheetflow evidence was limited parallel to the ridge
top axis. Surface-water flow accumulations derived
from the 13 November TLS data indicate that upslope
heads of the levee-lined rills had contributing drainage
areas of <1 m2 whereas debris flows were identified in
>25-30 m
2
drainage areas.
From our TLS surveys and field observations, we
interpreted that surface erosion and debris flows were
generated by the following influences: i) sediment ac-
cumulation by post-fire, pre-rain dry-ravel processes
acting under the combined influence of gravity and
wind that loaded valley axes with loose sediment and
then ii) relatively small rainfall amounts promoted the
rapid development of shallow overland-flow drainage
networks into both colluvial soil and recently-trans-
clasts mantled by fine silt; low-density char and un-
burned organic material was uniformly disseminated
throughout the deposits. Hillslope debris-flow deposits
on the western flank of the watershed deposited ups-
lope of the primary valley axis. Topographic changes
to the western flank are interpreted to have progressed
in the following sequence: the levee-lined tracks not
incised into hillslope materials and decimeter-scale
debris-flow deposits were incised by later stage levee-
lined rills generated by overland surface flow. This
later stage incision exhumed dry material to depths of
0.6 m, far below the observed wetting front depths of
~10 cm, and hence their formation is consistent with
rill formation by fluvial transport. Although some of
the drainage flow paths generated by the storm were
sinuous and anastomosing near the ridge top (e.g., Fig.
4b), the hillslope-scale drainage pattern was generally
linear, sub-parallel paths within more planar topogra-
phy (e.g., Fig. 7).
Fig. 7 - Surface elevation change between terrestrial laser scans (11/13/09-11/12/09). Negative values indicate low-
ering, and positive values indicate aggradation. (a) Lukens IRA watershed, (b) view upslope from valley axis
[area defined by white dashed polygon in (a), with 5x vertical exaggeration] highlighting lobate terminal
snouts, shown in (c) photograph of debris-flow deposits on gradients of ~33°
background image
HYDROLOGIC CONDITIONS AND TERRESTRIAL LASER SCANNING OF POST-FIREDEBRIS FLOWS IN THE SAN GABRIEL
MOUNTAINS, CA, U.S.A
Italian Journal of Engineering Geology and Environment - Book www.ijege.uniroma1.it © 2011 Casa Editrice Università La Sapienza
591
volume (~70%) occurred by excavation of post-fire
dry-ravel accumulations and older colluvial soil in
the previously unchannelized valley axis with aver-
age ground surface lowering >0.5 m. Rainsplash and
sheetflow erosional processes were spatially restricted
along the gentler convex ridge top, whereas chan-
nelized flow processes that generated incisional levee-
lined rills and non-incisional levee9 bounded tracks
leading to debris-flow deposits were widespread.
Post-storm observations indicated that rills (~5-20 cm
wide) formed primarily where loose, granular mate-
rial, post-fire ash, and charcoal deposits remained and
extended to within 0.5-2.0 m of the drainage divide.
Based on our two monitored surface-flow lines, it ap-
pears that upslope contributing drainage area influ-
enced the location and timing of flow generation, with
the steeper, larger drainage area responding first.
Geomorphic sediment transport processes in our
study area varied spatially, as well as temporally dur-
ing a single storm. The disparity in timing and spa-
tial extent of both soil moisture and overland-flow
responses highlights the complications imposed by
rapid runoff from both colluvial soil expressing vary-
ing degrees of hydrophobicity and lower permeability
bedrock exposed at surface. Overland-flow sensors
and soil-moisture probes indicated that the colluvial
soil was unsaturated prior to and during the initial
pulses of overland flow and debris-flow transport. At
the time of recorded passage of a debris flow at the
Lukens IRA watershed outlet, only four of the eleven
overland flow sensors recorded measurable surface
water and the soil-moisture probes had not yet dem-
onstrated measurable increases. Within 15-20 minutes
after the onset of rainfall, there was rapid development
of surface runoff leading toincision and redistribution
of sediment. Although sediment transport rates should
be ultimately controlled by rainfall intensity, duration,
and subsequent available water discharge, the timing
of the debrisflow peak at the Lukens IRA watershed
outlet occurred during a lull in rainfall following two
initial higher intensity rainfall episodes totaling just
over 6 mm in sixteen minutes. This debris flow was
triggered by rainfall that exceeded previously estab-
lished empirical thresholds for post-fire debris-flow
initiation (C
annon
et alii, 2010). The highest recorded
rainfall intensities followed the occurrence of debris
flows. Field evidence indicates that initial debris-flow
deposits were over-printed by later fluvial incision, at
ported ravel deposits. We interpreted the remaining,
somewhat scant, evidence of debris-flow generation
by walking upslope from debris-flow deposits, tracing
evidence to the farthest upslope extent. Much of the
evidence indicates that the initial surface flow did not
appreciably incise, but rather flowed over a primarily
dry ground surface and entrained sediment by traction
and saltation with additional contributions from small,
cm-scale (<10 cm high scarps) landslides. Scarps con-
tained threadlike broken roots. These small Coulomb-
style slab failures were likely episodicand punctu-
ated sediment concentrations within the surface-water
flow. We suspect these onslope debris flows depicted
in Figure 7c were generated, at least in part, by a proc-
ess similar to that elaborated by w
ells
(1987) where
small Coulomb-style failures on the scale of centime-
tres to decimetres wide form above a water repellent
layer. These failures cause material to slide downslope
and mobilize into debris flows which in turn plows
a path through the granular sediment and deposits
steep-faced lobate snouts downslope of levee-bound-
ed tracks (Fig. 7c). Local bedrock exposures routing
water to newly formed drainage networks may have
also contributed to “fire-hose” mechanism generated
debris flows. Watershed-averaged surface elevation
change attributable to this individual storm, calculated
from the difference grid depicted in Figure 7a, was:
maximum 1.10 m, minimum -1.68, average -0.03, and
standard deviation ± 0.12 m. The associated volume
of material represented by this lowering is 336 m
3
.
Average lowering within the valley axis of -0.53 m
(standard deviation ± 0.29 m) resulted in an estimated
volume of 232 ± 127 m
3
, hence the majority (~70%)
of the sediment volume removed was derived from
post-fire dry-ravel accumulations and pre-fire sedi-
ment stored in the valley axis.
DISCUSSION AND SUMMARY
Using instrumental monitoring of rainfall and
runoff, bare-earth model DEMs developed from re-
peat laser scanning of topography, and field mapping
assisted by differential kinematic GPS, we document-
ed how spatial patterns of rainsplash, overland-flow
scour, and rilling changed in response to a localized
storm that generated meter-scale debris flows. These
flows deposited both irregular levees and lobate ter-
minal snouts on portions of the hillslope and at the
watershed confluence. The majority of the erosion by
background image
k.M. SCHMIDT, M.N. HANSHAw, J.F. HOwLE, kEAN, J.w., D.M. STALEY, J.D. STOCk & w. BAwDENG
592
5th International Conference on Debris-Flow Hazards Mitigation: Mechanics, Prediction and Assessment Padua, Italy - 14-17 June 2011
both the hillslope and higher-order valley scales.
During the height of the storm, when overland-
flow processes likely dominated, the greatest volumet-
ric water content in the colluvial soil was uniformly
less than 0.2 m
3
/m
3
, with one soilmoisture pit show-
ing no increase in volumetric water content during the
duration of the storm. Probe responses in the two pits
exhibiting moisture response were consistent with top-
down, infiltrationlimited water migration. Field obser-
vations following the event indicated that the granular
materials below depths of ~10 cm principally remained
dry in agreement with our soil-moisture probe respons-
es. The presence of incisional rills into the underlying
dry material at depths >10 cm is likely consistent with
rill formation by fluvial transport. These observations
suggest that instrumental monitoring could be used
to improve realtime warning systems, but lead times
for advisories after the onset of rainfall would remain
short in steep, small drainage area basins and thus the
operational aspect of issuing advisories with such short
lead times may be untenable.
Even with instrumental records, topographic sur-
veys, and firsthand observations immediately follow-
ing the storm, it remains unclear if the resulting drain-
age network formed in place or if rill heads migrated
upslope over time. Nor is it readily apparent why small
debris flows deposited on moderately steep gradients
of 30- 35°. A combination of downslope dewatering of
the granular material, slight decreases in the gradient
of the surface traversed, increased basal friction, and/
or a change in ground surface material from colluvial
soil to post-fire ravel deposits may have contributed
to deposition on such gradients. As the lobate debris-
flow snouts stratigraphically overlie widespread post-
fire ravel aprons located on lower reaches of hillslopes,
it is possible that a combination of slight decreases in
downslope gradient coupled with a more permeable
bed material may have contributed to deposition. Al-
though some debrisflows deposited on hillslopes, oth-
ers traversed the primary valley axis and left deposits
at the watershed outlet. These deposits were subse-
quently re-entrained by and incised into by traction-
dominated, surface-water transport later in the storm.
ACKNOWLEDGEMENTS
Land access was made possible by the courteous
and professional employees of the Cities of Pasadena
(l
isa
d
eRdeRian
& s
Hawn
k
wan
) and Glendale, CA
Parks, Recreation & Community Services (specifi-
cally J
eff
w
einstein
, d
avid
a
HeRn
, & R
uss
H
auCk
).
s
andRa
b
ond
(USGS) provided invaluable TLS data
processing support. The USGS Landslide Hazards
and National Cooperative Geologic Mapping Pro-
grams provided funding for this research. We thank
Sue Cannon and Mark Reid for thorough and insight-
ful reviews.
REFERENCES
C
annon
s.H., G
aRtneR
J.e., P
aRRett
C., & P
aRise
m. (2003) - wildfire-related debris-flow generation through episodic pro-
gressive sedimentbulking processes, western USA. In: R
iCkenmann
d. & C
Hen
C.l.,
eds
. Debris-Flow Hazards Mitigation:
Mechanics, Prediction, and Assessment: Davos, Switzerland, Millpress, 71-82.
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: 250-269.
C
annon
s.H., G
aRtneR
J.e., R
uPeRt
m.G., m
iCHael
J.a., s
taley
d.m. & w
oRstell
b.b. (2009) - 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. (http://pubs.usgs.gov/of/2009/1227/).
C
annon
s.H., b
oldt
e.m., k
ean
J.w., l
abeR
J.l. & s
taley
d.m. (2010) - 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. (http://pubs.usgs.gov/of/2010/1039/).
C
oe
J.a., k
inneR
d.a., & G
odt
J.w. (2008) - Initiation conditions for debris flows generated by runoff at Chalk Cliffs, central
Colorado. Geomorphology, 96: 270-297.
d
i
b
iase
R.a., w
HiPPle
k.x., H
eimsatH
a.m. & o
uimet
w.b. (2010) - Landscape form and millennial erosion rates in the San
GabrielMountains, CA. Earth and Planetary Science Letters, 289: 134-144.
e
aton
e.C. (1936) - Flood and erosion control problems and their solution. Proceedings of the American Society of Civil En-
gineers, 62(8): 1302–1362.
i
veRson
R.m., R
eid
m.e. & l
a
H
usen
R.G. (1997) - Debris flow mobilization from landslides. Annual Reviews of Earth and
background image
HYDROLOGIC CONDITIONS AND TERRESTRIAL LASER SCANNING OF POST-FIREDEBRIS FLOWS IN THE SAN GABRIEL
MOUNTAINS, CA, U.S.A
Italian Journal of Engineering Geology and Environment - Book www.ijege.uniroma1.it © 2011 Casa Editrice Università La Sapienza
593
Planetary Sciences, 25: 85-138.
k
ean
J.w. & s
taley
d.m. (this volume) - Direct measurements of post-fire debris-flows: examples from monitoring efforts in
southern California 1832.
l
avé
J. & b
uRbank
d. (2004) - Denudation processes and rates in the Transverse Ranges, southern California: Erosional re-
sponse of a transitional landscape to external and anthropogenic forcing. Journal of Geophysical Research, 109: F01006,
doi:10.1029/2003JF000023.
l
indvall
s.C. & R
ubin
C.m. (2003) - Slip rate studies along the Sierra Madre-Cucamonga fault system using geomorphic and
cosmogenic surface exposure age constraints. Collaborative research with Central Washington University and William
Lettis & Associates, Inc. U.S. Geological Survey final report NEHRP External Grant Award Number 03HQGR0084. (http://
earthquake.usgs.gov/research/external/reports/03HQGR0084.pdf)
NOAA-USGS d
ebRis
f
low
t
ask
f
oRCe
(2005) - NOAA-USGS debris-flow warning system - Final report. U.S. Geological Sur-
vey Circular 1283. (http://pubs.usgs.gov/circ/2005/1283/).
s
Hakesby
R.a. & d
oeRR
s.H. (2006) - wildfire as a hydrological and geomorphological agent. Earth-Science Reviews, 74:
269-307.
s
mitH
d.P. (1986) - Geology of the north half of the Pasadena quadrangle, Los Angeles County, California. Division of Mines
and Geology Open File Report, OFR 82-10LA, map scale 1:12,000.
s
Potilla
J.a., H
ouse
m.a., b
lytHe
a.e., n
iemi
n.a. & b
ank
G.C. (2002) - Controls on the erosion and geomorphic evolution of
the San Bernardino and San Gabriel Mountains, southern California. In: Barth, A., ed., Contributions to Crustal Evolution
of the Southwestern United States: Boulder, Colorado, The Geological Society of America Special Paper 365: 205–230.
t
ayloR
C.a. (1934) - Debris flow from canyons in Los Angeles County flood. Engineering News-Record, 112: 439-440.
t
Roxell
, H. C. & J. Q. P
eteRson
(1937) - Flood in La Cañada Valley California, January 1, 1934. U. S. Geological Survey
Water-Supply Paper 796-C: 98.
w
ells
w.G. ii (1987) - The effects of fire on the generation of debris flows in southern California. In: C
osta
J.e. & w
ieCzoRek
,
G.f.,
eds
., Debris flows/avalanches: process, recognition, and mitigation. Boulder, CO, The Geological Society of America,
Reviews in Engineering Geology, VII: 105-114.
Statistics