Channel scour and fill by debris flows and bedload transport

Geomorphology 243 (2015) 92–105

Contents lists available at ScienceDirect

Geomorphology journal homepage: www.elsevier.com/locate/geomorph

Channel scour and fill by debris flows and bedload transport J.I. Theule a, F. Liébault a,⁎, D. Laigle a, A. Loye b, M. Jaboyedoff b a b

Irstea Grenoble, Unité de Recherche ETNA (Erosion Torrentielle, Neige et Avalanches), Université Grenoble Alpes, Saint-Martin-d'Hères, France Institut des sciences de la Terre (ISTE), Faculty of Geosciences and Environment, University of Lausanne, Lausanne, Switzerland

a r t i c l e

i n f o

Article history: Received 10 October 2014 Received in revised form 7 May 2015 Accepted 8 May 2015 Available online 16 May 2015 Keywords: Debris flow Bedload transport Slope effect Roughness Scour and fill

a b s t r a c t In steep channels, debris flows are known to dramatically increase in volume under the effect of channel erosion. However, the critical factors controlling channel erosion by debris flows are not well documented by field studies. This is particularly true for the effect of slope on the depth at which erodible beds are scoured during debris flows and during bedload transport. This topic has been addressed by intensive cross section resurveys (54 cross sections) of debris flows (n = 5) and flow events (n = 9) that occurred in two torrents of the French Prealps, the Manival and Réal torrents, between 2009 and 2012. This study provided evidence that debris-flow scouring increases with slope, whereas this is not the case for bedload transport (no slope effect detected during floods). A functional relationship defined from a piecewise regression model is proposed as an empirical fit for the prediction of channel erosion by debris flows with a critical slope threshold at 0.19 (95% confidence interval: 0.17–0.21). This slope threshold is interpreted as the transition between the transport-limited and supplylimited regimes, associated with the upstream decreasing erodible bed thickness. The erodible bed was also characterized by quantifying erosion, deposition, and surface roughness with multidate terrestrial laser scans (TLSs) in a short reach of high sensitivity of the Manival torrent. Debris-flow erosion occurred preferentially on smooth surfaces corresponding to the unconsolidated gravel deposits from bedload transport. A 20-cm resolution roughness profile from an airborne laser scan (ALS) and a slope profile of the whole channel were used to detect the unconsolidated sediment deposits that can potentially feed future debris flows. © 2015 Elsevier B.V. All rights reserved.

1. Introduction Although steep-slope channels represent a considerable length of upland stream networks, our understanding of geomorphic laws that govern these channels is less advanced than for lowland fluvial systems. The variety of sediment transport processes encountered in this environment (debris flows and bedload transport) and the strong influence of hillslopes on sediment supply conditions interact to produce a specific geomorphic behavior in headwaters. This situation limits our ability to propose reliable tools for assessing the transfer of sediment from mountains to piedmont rivers, whereas at the same time, stakes related to the protection against natural hazards in mountains, and to the preservation of sediment continuity in aquatic ecosystems, are becoming more and more important. Debris flows are surges of saturated nonplastic debris traveling in steep channels at a range of 2–20 m s−1 with sediment concentrations generally higher than 50% by volume (Hungr et al., 2001; Jakob and Jordan, 2001). In contrast, bedload transport occurs during flow events with low sediment concentrations (b 20% by volume) and consists of the stochastic motion of individual particles

⁎ Corresponding author. E-mail address: [email protected] (F. Liébault).

http://dx.doi.org/10.1016/j.geomorph.2015.05.003 0169-555X/© 2015 Elsevier B.V. All rights reserved.

that are traveling in a series of step and rest periods at a velocity that is much lower than the fluid velocity. The geomorphic work of debris flows has been well documented over the past decades with several studies focusing on the quantification of magnitude and frequency and rainfall thresholds of debris flows (e.g., Innes, 1983; Kotarba, 1992; Jonasson and Nyberg, 1999; Vedin et al., 1999; Beylich and Sandberg, 2005; Decaulne et al., 2005). The sediment recharge of the channel from active erosion on hillslopes was found to be a critical factor of debris-flow magnitude and frequency (Bovis and Jakob, 1999; Marchi and D'Agostino, 2004; Jakob et al., 2005) as channel erosion is generally the most important contribution to the debris-flow volume (Hungr et al., 1984, 2005). Our study focuses on runoff-generated debris flows that are triggered by high intensity rainfalls; these surges progressively entrain the channel where erodible sediments are present, which in turn increases the sediment concentration (Rickenmann et al., 2003; Berti and Simoni, 2005; Godt and Coe, 2007; Coe et al., 2008; Gregoretti and Fontana, 2008). One of the most striking characters of debris flows is their ability to dramatically cut and fill the granular bed during their propagation. These channel deformations can commonly reach several meters (Jakob et al., 2005; Remaître et al., 2005; Imaizumi et al., 2006). A classic way to predict erosion and deposition for a given reach is to use the conservation of mass (Exner equation) (Ashmore and Church, 1998). The rate of change of channel elevation per unit time is a function of the

J.I. Theule et al. / Geomorphology 243 (2015) 92–105

rate of change of sediment transport per unit length of stream. The sediment transport rate is controlled by the forces exerted by the fluid on the bed (shear stress) and also by sediment supply conditions (in terms of frequency, volume, and grain-size distribution), which are determined by channel conditions and also by external delivery from hillslopes. These drivers of channel deformations are extremely difficult to quantify in a deterministic way, which limits our understanding of the spatial variability of erosion and deposition in debris-flow channels. Even though in-channel sediment deposits are known to be an important source of material for debris flows, our understanding of their entrainment is still limited. Recently, studies gave insights on channel conditions controlling the release of sediment from a steep bed during a debris flow, notably by looking at the effect of pore water pressure on channel scouring. This effect has been demonstrated in large-scale flume experiments, which show that a small increase of the bed sediment water content before the flow release had a considerable effect on the rate at which the bed is entrained by the flow (Iverson et al., 2011). These experimental findings were recently confirmed by field observations at the Chalk Cliffs debris-flow observatory, where it has been shown that the time-averaged bed sediment entrainment rate was faster when saturation of the bed prevails when the flow occurs (McCoy et al., 2012). Another study using erosion sensor column data has observed the timing of bed scouring during debris flows and showed that the majority of scouring occurs at the debris-flow front, during the shear stress peak (Berger et al., 2011); this was also observed at the Chemolgan field laboratory site (Rickenmann et al., 2003). Although these field and laboratory experiments provided insights into the instantaneous scour processes, they were not designed to address the spatial variation of scouring and notably the effect of slope and material properties on channel erosion by debris flows and bedload transport. An experimental study has examined thoroughly the role of erodible sloping beds on granular flow dynamics with a tilting apparatus (Mangeney et al., 2010). It was found that the growth of granular flow momentum from scouring became insignificant when the slope is below half the repose angle of the bed material. This study also found that the presence of an erodible bed increases granular flow travel distance by 40%. Few field studies have attempted to look at the effect of slope. Post-event field measurements of maximum scouring for the 1987 debris flows in Switzerland revealed a broad linear relationship with slope (Rickenmann and Zimmermann, 1993). A database compilation of 174 debris flows and debris avalanches in the Queen Charlotte Islands (British Columbia) was used to look at the relation between slope and debris-flow yield rates, without any clear trend in the data (Hungr et al., 2005). This result has been partly attributed to the poorly constrained nature of erosion observations, which are not based on high-resolution topographic resurveys (Hungr et al., 2008). This is not the case of a recent study in Iceland, based on airborne laser scanning (ALS) and differential GPS surveys which showed a clear effect of slope on erosion and deposition patterns of first-order gullies entrenched into steep talus slopes (Conway et al., 2010). Despite the many efforts for understanding debris-flow erosion, there is very little field observations using reliable topographic surveys across long reaches with high periodicity. More field data are needed to address the effect of slope on channel erosion by debris flows and by bedload transport. This is particularly true for steep-slope channels located on debris-flow fans because this is a critical zone where debris flows can dramatically grow in volume due to the presence of loose debris. The focus of this paper is to investigate the different scour and fill patterns for debris flows and bedload transport along steep-slope channels and to better define the controls on debris-flow erosion by assessing the effect of channel slope and channel conditions (roughness). Two study sites are investigated: the Manival and Réal debrisflow torrents in the French Alps, which frequently experience both debris flows and bedload transport.

93

2. Study sites 2.1. Manival torrent Debris flows frequently occur in the Manival torrent (Peteuil et al., 2008), situated near Grenoble in the Chartreuse Mountains of the northern French Prealps, located at 45°17′ N, 5°49.75′ E (Fig. 1C). This torrent has been previously studied to quantify debris input and channel storage by using photogrammetry with historical aerial photographs (Veyrat-Charvillon and Memier, 2006). The torrent flows intermittently into a large sediment trap (25,000 m3 capacity) that protects the urbanized fan from debris flows. The 3.6-km2 catchment above the sediment trap has 1130 m of relief with a mean catchment slope of 0.81 (39°) (Table 1). The 1.8-km study reach extends from the apex of the debris-flow fan to the sediment trap (long profile shown in Fig. 1B). The Manival area has a mean annual precipitation of 1450 mm and a 10-year daily rainfall of 88 mm (from monthly rainfall data of the Saint-Hilaire-du-Touvet Météo France station since 1964). The catchment typically experiences convective storms during the spring and summer (May to September) which trigger debris flows. During autumn (September to December), steady and long duration rainfall generates bedload transport. The upper catchment is typically covered by snow in the winter (January to March), thereby having a dormant channel. The geology of the Manival catchment corresponds to the Mesozoic cover of the external alpine crystalline belt. The bedrock is composed of highly fractured, alternating sequences of marls and limestones from the Upper Jurassic to early Cretaceous with a bedding thickness ranging from decimeters to meters (Loye et al., 2012). Geomorphic processes are typical of upland prealpine catchments. Shallow landslides, hillslope debris flows, and snow avalanches form thick colluvial deposits below rockwalls and hillsides. Limestone rock faces are known to produce frequent rockfall that supplies debris to talus slopes (Fig. 2A). Upstream from the sediment trap the mean channel slope is 0.16 (9°) over 1.8 km to the apex of the debris-flow fan. The channel has a mean active width of ~15 m (range: 10–20 m) with a typical morphology of a debris-flow scoured channel consisting of levees, boulder fronts, and coarse lags (Fig. 2B). It is formed within the wide debrisflow fan (40 to 250 m wide, increasing downstream). Along the main channel, bedload transport macroforms can be observed (Fig. 2C). They are defined as gravel deposits with well-sorted grain-size distributions that can fill the U-shaped debris-flow channel. These macroforms are an important part of the sediment input for debris-flows mass balance (Theule et al., 2012). 2.2. Réal torrent The Réal torrent is a very active debris-flow torrent located in the upper Var River catchment of the Southern French Prealps, located 44°07′ N, 06°54.5′ E (Fig. 1A). It flows intermittently into the Tuébi River, a tributary to the Var River, near the small village of Péone. Debris flows occur 2–3 times every year and interact with bedload transport. The 2.3-km2 catchment has 800 m of relief with a mean catchment slope of 0.58 (30°) (Table 1; Fig. 1B). The Réal area has a drier climate, with a mean annual rainfall of 1050 mm, but it also experiences large storms with a 10-year daily rainfall of 102 mm (from monthly rainfall data of the Péone Météo France station since 1951). Convective storms occur in the spring and summer (May to September). During autumn (September to December), steady and long duration rainfall generates bedload transport. The catchment is also covered by snow in the winter (January to March). Bedrock geology is composed of Paleogene sandstones and alternating sequences of Cretaceous and Jurassic marls and limestones. Quaternary

94

J.I. Theule et al. / Geomorphology 243 (2015) 92–105

(A)

(B)

Geneva

Manival Grenoble

Italy Réal Nice

Marseille

(C)

(D)

fan apex cross section sediment trap active scarps upper-catchment active channel

The Big Ravine

terraces landslide deposit whole catchment

fan apex cross section

Tuébi River

test reach

active scarps active landslide active channel terraces whole catchment

Fig. 1. (A) Location of the two study sites in the French Alps; (B) long profiles and slopes of the main channels of the two torrents in reference to the debris-flow fan apex. Map of the main geomorphic processes and landforms of the Manival (C) and the Réal (D) showing cross section locations, active hillslope erosion, and the active channel. The yellow zone in the Manival and the ‘Big Ravine’ in the Réal are the most important sediment sources of the catchments. The test reach is indicated in C where the TLS scans have taken place.

deposits cover ~70% of the catchment. Most of the sediment transported during flow events comes from active gullying in the spectacular alluvial fills related to the obstruction of the valley by a glacier during the Last

Table 1 Topographic characteristics and surveys of the Manival and Réal catchments.

2

Drainage area (km ) Minimum elevation (m asl) Maximum elevation (m asl) Mean catchment slope (m/m; °) Length of the study reach (km) Mean slope of the study reach (m/m; °) Mean active channel width (m) Monitoring period Number of cross sections Number of topographic surveys Number of check-dams along the study reach

Manival

Réal

3.6 570 1738 0.81; 39° 1.8 0.16; 9° 15 07/2009–12/2010 39 9 19

2.3 1254 2048 0.58; 30° 1.8 0.16; 9° 25 04/2010–09/2011 15 7 8

Glacial Maximum. These 100-m-thick, unconsolidated alluvial constructions are prone to intense gullying and landslides, which provide an unlimited sediment supply to the torrent. An active deep-seated landslide affecting Jurassic black marls since the 1920s located on the right bank of the main channel may also contribute to the sediment recharge. However, field observations reveal that the most important sediment source comes from a very active gully entrenched into the fluvioglacial deposits (Chambon and Richard, 2004); we call this gully the ‘Big Ravine’ (Fig. 3A). The 1.8-km study reach extends from the confluence of the Tuébi to the proximal limit of the debris-flow fan. The mean channel slope is 0.16 (9°) and the mean active width is ~25 m (range: 15–55 m). The channel morphology is a complex assemblage of erosional and depositional forms resulting from debris flow and from bedload transport (Fig. 3B, C). In the upstream part of the study reach, the active channel is wider and occupies the entire valley floor. An isolated vegetated alluvial terrace is observed in the middle part of the study reach, on the left bank of the torrent. This terrace tread is continuous along the last 800 m of the study reach and creates a buffer zone between the active channel

J.I. Theule et al. / Geomorphology 243 (2015) 92–105

95

(A)

(A)

(B) (B)

(C) (C) gravel deposit

gravel deposit

debris flow levee

debris flow levee

Fig. 2. Pictures of the Manival torrent showing (A) the sediment source area, (B) a debrisflow scoured reach, and (C) gravel deposits from bedload transport with an older debrisflow levee on the bank.

and the hillslopes. Between the exit of the hillslope-confined valley and the confluence with the Tuébi, the Réal flows along a short 250-m reach entrenched into the recent terraces of the Tuébi.

3. Methods 3.1. Scour-and-fill surveys Scour and fill in the Manival and Réal torrents were derived from topographic resurveys of permanent cross sections. They were measured with a total station (Leica Flexline TS02) with an electronic distance measurement precision of 1.5 mm ± 2 ppm, and an angular resolution of 7″ or 3.4 mm of precision at a distance of 100 m. Along the 1.8-km study reach of the Manival, 39 cross sections are distributed with a mean spacing of 40 m (three times the mean active channel width). For practical reasons, it was not possible to reproduce the same intensive surveying for the Réal; therefore only 15 cross sections were

Fig. 3. Pictures of the Réal Torrent showing (A) the sediment source area in the ‘Big Ravine’, (B) a debris-flow-scoured channel, and (C) gravel wedges formed from bedload transport with older debris-flow levees on the banks.

deployed along the 1.8-km study reach, giving a mean cross section spacing of 120 m (five times the mean active channel width). Eight post-event surveys were completed in the Manival between 2009 and 2012, with two being carried out after debris flows of moderate magnitude (1900 and 3300 m3; Table 2). The monitoring period in the Réal was from spring 2010 to autumn 2011; six post-flow surveys were completed, with three after debris flows of moderate intensity (4100, 4500, and 7400 m3; Table 2). Further details on debris-flow monitoring and volume assessments for these events are found in Theule et al. (2012) and Theule (2012). The level-of-detection for significant elevation change between pre- and post-event surveys was determined from the D84 of the bed surface grain-size distribution, providing a good indicator of the grain roughness of the bed, which for the Manival and Réal are ~5 cm. It is important to mention that debris flows in these torrents are often in the form of multiple surges. The topographic surveys capture the time-integrated elevation change of the torrent and not the individual surges. An eyewitness reported four surges for an event in the Manival, and in the Réal high-frequency monitoring stations detected up to four surges during recorded debris flows (Navratil

96

J.I. Theule et al. / Geomorphology 243 (2015) 92–105

Table 2 Sediment volume and rainfall characteristics (from five-min time intervals) of flow events in the Manival and the Réal. DF: debris flows; BL: bedload transport. Modified from Theule (2012). Torrent

Process

Cross section survey time period

Sediment output (m3)

Total rainfall (mm)

Mean storm intensity (mm h−1)

Maximum intensity (mm h−1)

TLS survey period in the test reach

Manival

DF BL BL BL DF BL BL BL DF DF BL BL DF BL

06/07/2009–28/08/2009 30/08/2009–07/10/2009 08/10/2009–12/11/2009 13/11/2009–01/06/2010 02/06/2010–08/06/2010 09/06/2010–08/10/2010 14/10/2010–25/11/2010 25/11/2010–10/12/2010 15/04/2010–18/05/2010 18/05/2010–01/07/2010 01/07/2010–21/10/2010 21/10/2010–21/06/2011 21/06/2011–13/07/2011 13/07/2011–05/10/2011

1900 0 300 600 3300 800 200 500 4100 4500 3400 500 7400 300

93 24 101 239 26 174 150 33 100 57 130 499 41 143

3.2 2.8 1.9 2.1 6.2 3.0 2.2 2.0 3.0 1.3 1.1 2.2 1.9 1.8

49 17 7 10 25 79 17 7 21 31 21 34 79 60

01/08/2009–29/08/2009 29/08/2009–12/11/2009

Réal

et al., 2013). In many cases for individual surges, the debris-flow front scours the bed and the tail reworks the bed and may induce deposition (e.g., Rickenmann et al., 2003; Berger et al., 2011). The tail deposit of the debris flow is measured in the post-event topographic survey; this gives a lower-bound estimate of scour depth. We used the maximum scour depth (dmax) for analysis to have the closest true measurement of debris-flow scour. The maximum scour depth corresponds to the maximum negative elevation change within the active channel of the resurveyed cross section. Bank erosion is not included as channel scouring because it is not directly controlled by the shear stress of the debris flow acting vertical to the channel surface. To remove the effect of flow confinement on channel scouring, we normalize the maximum scour by the channel scour width (Ws). This width was measured from the top edge of each bank where the debris flow occurred, excluding levees that are depositional landforms (Fig. 4). Frequency distributions of scour and fill depths for debris flows and water runoff events with bedload transport were computed by sampling scour and fill depths at a 2-m interval along the active width (Wa) of each cross section. The active width was defined as the portion of the channel characterized by significant elevation change between two surveys (elevation change above the level-of-detection). In order to remove the bias toward wider cross sections in our sample of scour and fill depths, the three maximum values of scour and fill depths of each cross section were finally used to process the frequency distributions (Fig. 4).

167

cross section at t0 cross section at t1

Elevation (m)

166 165 164 163 2-m sampling

162

Max 3 scour depths Max 3 fill depths

161 160 0

5

10

15

20

Distance from right bank (m) Fig. 4. Cross section view before and after a debris-flow event. Scour width (Ws) and maximum scour depth (dmax) were derived from cross section overlay. The 2-m sampling of scour-and-fill depths are found within the total active width (Wa). The maximum three scour samples (red dash) and maximum three fill samples (blue dash) are used for frequency distribution analysis.

12/11/2009–09/07/2010

An upstream slope (S) was used for analysis, which is measured along a length equal to six times the active channel width (upstream from the cross section). This was determined empirically by looking at the best fit between different slope measurements and channel scour (Theule, 2012). 3.2. Laser scanning for the characterization of erodible material A 200-m reach of the Manival (Fig. 1), referred as the test reach, was scanned before and after three events (debris flow–bedload transport– debris flow; Table 2) with an ILRIS-3D (Optech Inc.) terrestrial laser scanner (TLS). Details on instrumentation and measuring techniques are found in Theule et al. (2012). Several scanning positions from different viewing angles were always used for obtaining the maximum point coverage, thereby minimizing the most important source of error (Schürch et al., 2011). The multiple scans for each survey campaign were merged together using Polyworks™, which uses the Automatic Iterative Closest Point algorithm (Besl and McKay, 1992). The root mean square error (RMSE) obtained from this technique ranges from 0.6 to 1.0 cm. The final point density after data processing for each campaign ranged from 22 to 55 points per 20-cm grid cell. Merging the multidate scans together by selecting permanent features as reference points produced a RMSE range from 1 to 3 cm. Digital elevation models (DEMs) with a 20-cm resolution were computed by taking the mean elevation of points within each 20-cm cell. Cells with at least five points were used for analysis; this limits any noisy data that does not accurately represent the surface. Interpolation methods were not used for filling in areas of no data. Elevation differences were directly calculated from the post- and pre-event 20-cm grids with the differences covering between 52 and 59% of the study reach area. For characterizing the nature of the channel sediment, we applied a method proposed by Cavalli et al. (2008) that identifies channel features using roughness from airborne LiDAR. Step-pools and riffle-pools were distinguished with the standard deviation of residual elevations with a 0.5-m grid and a 2.5-m search window (1:5 ratio). The residual elevations are the cell-by-cell difference of the LiDAR data and the mean elevation of the 2.5-m search window. With the TLS data in the test reach, we applied this method at a higher resolution, using a 20-cm grid and a 1-m diameter search window. The kernel-cell in the search window is kept for analysis only if there are at least five cells in the window. This avoids misrepresentative surfaces with little data coverage, usually found on the vegetated banks of the channel. This method of channel roughness characterization was also tested on airborne laser scan (ALS) data (June 2009) of the Manival using the same grid size of 20 cm. The ground point density of the ALS within the channel averaged 16 points m−2. The channel bed was manually

J.I. Theule et al. / Geomorphology 243 (2015) 92–105

higher for scour depths highlighting the erosive characteristic of the debris flows in the studied reaches.

delineated for the entire catchment. Along the profile, the average roughness was calculated from a 14-m-diameter moving window (the average channel width). This gives a general characterization of the different reaches within the catchment.

4.2. Slope effect on channel scour during debris flows Scour-and-fill data have been grouped into classes of homogeneous channel slope to look at the effect of slope. This has been done separately for each site and for each flow process (Figs. 6 and 7). A slopedependence or any scour-fill pattern cannot be seen for bedload transport. However, fill gently increases with slope in the Réal, but this pattern can be explained by the deposition of small debris flows coming from small lateral gully channels and should not be strictly attributed to bedload transport. For debris flows, a strong slope-dependence is observed with an increasing scour with slope. Important scouring can be seen above a slope of 0.16 (9°) for the Manival and 0.15 (8.5°) for the Réal. There is also a gentle decrease of fill with slope in the Manival; however, not for the Réal (data are less clear in the Réal owing to the smaller number of samples). The investigated range of slopes is too steep to capture any important fill patterns. Outliers in Figs. 6 and 7 are most likely influenced by transported boulders or scouring on the edges of older debris-flow deposits. These observations lead us to investigate the relation between normalized channel scour and channel slope for debris flows (Fig. 8). A database of debris-flow scour (dmax), scour width (Ws), and slope (S) was compiled with 127 cross section measurements from the Manival torrent, the Réal torrent, and the Manival upper catchment (Theule et al., 2012). Slopes (S) range from 0.12 (6.8°) (distal reaches) to 0.63 (32°)

4. Results 4.1. Scour and fill associated with debris flows and bedload transport According to the three maximum samples of scour and fill at each cross section (described in Fig. 4), scour exceeds fill during debris flows (Fig. 5). This is observed for the two sites, but the difference between debris-flow-induced scour-and-fill depths is more important for the Manival (U test p-value is 0.003 for the Manival, and 0.023 for the Réal). Scour depths are generally twice as large as the fill depths (the scour/fill ratio is 2.58 for Manival, 1.57 for the Réal). During bedload transport, scour and fill compensate each other. There is a small difference of scour-and-fill depths (b1 cm) observed for bedload transport in the Manival (U test p-value of 0.001), and no difference could be found in the Réal because of fewer samples (p-value of 0.339). In general, little difference is observed for both sites with a scour/ fill ratio of 1.17 for the Manival and 1.05 for the Réal. When comparisons are made between debris flows and bedload transport, channel scour-and-fill are shown to be systematically higher during debris flows. The debris flow/bedload ratio for scour depths is 4.42 for the Manival and 2.2 for the Réal. The same ratios for fill depths are 2.0 for the Manival and 1.47 for the Réal. These ratios are especially

bedload

debris flow

(A) 2

n=227

n=227

2

1.5

1

1

Depth (m)

1.5

0.5

97

n=538

n=538

0.5 (0.31) (0.12)

0

0 Scour

Scour

(0.06)

Fill

box plot legend

(B)

Depth (m)

Fill

(0.07)

n=129

n=129

n=128

2

2

1.5

1.5

n=128

within 1.5 times the inter quartile range

75th percentile median

1

0.5

1

25th percentile

0.5

(0.44) (0.28)

0

(0.20)

(0.19)

0 Scour

95% confidence interval

Fill

Scour

Fill

within 1.5 times the inter quartile range greater than 1.5 times the inter quartile range

Fig. 5. Notched box plots of the three maximum samples of scour and fill for each cross section (described in Fig. 4) of the Manival (A) and (B) Réal.

98

J.I. Theule et al. / Geomorphology 243 (2015) 92–105

debris flow n=27

24

54

42

17

2 12

12

30

3

bedload n=68

59

120

96

42

30

28

0.14

0.15

0.16

0.17

0.18

0.19

0.20

59

120

96

42

30

28

72

0.14

0.15

0.16

0.17

0.18

0.19

0.20

72

1.8

Scour (m)

1.6 2.5

1.4

2

1.2 1

1.5

0.8 0.6

1

0.4 0.5

0.2 0

0 0.13

0.14

0.15

0.16

0.17

0.18

0.19

0.20

0.13 2

n=27

24

54

42

17

12

12

30

3

1.8

n=68

1.6

Fill (m)

2.5

1.4 1.2

2

1 1.5

0.8 0.6

1

0.4 0.5

0.2 0

0 0.13

0.14

0.15

0.16

0.17

0.18

0.19

0.20

S (m/m)

0.13

S (m/m)

Fig. 6. Notched box plots of the three maximum samples of scour and fill for each cross section (described in Fig. 4) in the Manival related with upstream slope (S) for debris flows and bedload transport. Debris-flow scour and fill relates to slope, whereas bedload transport shows no relation.

(first-order gully) where the steepest reaches are within the range of the repose angle of talus slopes (0.61–0.73; 31–36°). The scatterplot of the normalized channel scour (dmax/Ws) as a function of slope clearly shows a positive covariation with a change of regime above a given slope value (Fig. 8). A piecewise regression model was used to automatically detect this critical slope threshold (or breakpoint) where the regime shift occurs. This kind of statistical analysis was already used in bedload transport studies to objectively determine the discharge at which a regime shift of bedload rating curves occur (Ryan et al., 2002, 2005). We performed the piecewise regression analysis using the so-called segmented R-script (Muggeo, 2008) for the log-transformed data set of normalized scour (dmax/Ws) and slope (S) with the function written as follows:  ln  ln

dmax Ws dmax Ws

 ¼ a þ b1 ln ðSÞ for ln ðSÞ ≤ c

ð1Þ

¼ a þ b1 ln ðSÞ þ b2 ½ ln ðSÞ−c for ln ðSÞ N c

ð2Þ



where the critical slope threshold (breakpoint) is c and linear coefficients are a, b1, and b2. The results of the model are indicated in Table 3, which has an adjusted R2 of 0.51 and a residual standard error of 0.59. The piecewise regression detects a critical slope threshold for normalized scouring at 0.19 (11°) with a 95% confidence level comprised between 0.17 and 0.21 (9.5° to 12°) (Fig. 8). The application of the test of Davies (1987) showed a significant difference of slope for the regimes above and below the threshold (p-value b 0.001).

4.3. Test reach topographic description and sequence of events High-resolution channel changes were characterized in a very active reach of the Manival using TLS (Fig. 9). The TLS results for the multidate surveying in the test reach includes an event sequence of debris flow, bedload transport (flooding), and debris flow (Fig. 10). Initially (1 August 2009) the reach was full of unconsolidated, sorted gravels (gravel wedges) corresponding to an aggradation state visible in the long profile (Fig. 10C). These gravel wedges are characterized by low roughness (D50 = 0.039 m), which makes the reach the smoothest for the entire monitoring period (Fig. 10B). On 25 August 2009, a debris flow passed through the reach. Scans with the TLS were taken several days after showing significant scouring (Fig. 10A). However, the long profile shows that the debris flow did not remove all the gravel deposits (Fig. 10C). The mean elevation change of the reach was −0.41 m. Net erosion reached 786 m3, corresponding to 41% of the debris-flow volume measured at the sediment retention basin. The post-surface becomes rougher (Fig. 10B) with a D50 of 0.042 m but still contains smooth surfaces (gravel wedges deposited at the lower end of the reach). From the end of August to November 2009, bedload transport occurred during low intensity, long duration rainfalls. Several flow events induced the formation of new gravel wedges along the test reach (Fig. 10A). The mean elevation change of the reach was 0.12 m with a normal distribution. Some aggradation can be seen in the long profile, notably in the upper section of the reach (Fig. 10C). The reach gained 226 m3 of material that corresponds to 29% of the volume scoured during the previous debris flow. The test reach becomes smoother where deposition took place (D50 = 0.039 m).

J.I. Theule et al. / Geomorphology 243 (2015) 92–105

18

27

34

9

4

n = 38

26

18

35

9

5

4

3

Scour (m)

hillslope debris flow

bedload

debris flow n = 42

99

3

2 2

1 1

0

0

0.11-0.13 0.13-0.15 0.15-0.17 0.19-0.21 0.25-0.27 n = 42

27

18

34

9

4

n = 39

26

18

35

9

5

4

3

Fill (m)

0.11-0.13 0.13-0.15 0.15-0.17 0.19-0.21 0.25-0.27

3

2 2

1 1

0

0

0.11-0.13 0.13-0.15 0.15-0.17 0.19-0.21 0.25-0.27

0.11-0.13 0.13-0.15 0.15-0.17 0.19-0.21 0.25-0.27

S (m/m)

S (m/m)

Fig. 7. Notched box plots of the three maximum samples of scour and fill for each cross section (described in Fig. 4) in the Réal related with upstream slope (S) for debris flows and bedload transport. Debris-flow scour and fill relates to slope whereas bedload transport shows no relation. During bedload transport events, the steeper reaches experience deposition from small debris flows in the sediment source area. Notches sometimes exceed the quartiles because of small sample sizes and skewness.

The last debris flow scoured down to what was referred as the channel bottom, corresponding to highly consolidated old debris-flow deposits. The scour in the highly consolidated material was insignificant compared to the large scour depths in the unconsolidated gravels. The minimum long flat profile after the event shows that it is now at channel bottom (Fig. 10C). The elevation change again has an asymmetric distribution (mean of − 0.46 m). The reach lost 1013 m3 of material

Slope (m/m) Slope (º)

corresponding to 31% of the debris-flow deposition in the sediment retention basin. After this event, the test reach had the highest roughness for the entire monitoring period (D50 = 0.049 m). There were smooth deposits in the lower end of the reach which were likely formed during the recession limb, but they quickly eroded away later by bedload transport. Since this event, there has not been any recharge from the source area for the rest of the year. Without unconsolidated gravels in this reach, no channel changes were observed, even during high rainfall intensities observed from June–July 2010 (maximum 79 mm h−1).

dmax /Ws

ln(dmax /Ws)

4.4. Classifying erodible material with roughness In the maps of Fig. 10A and B, it can be seen that debris flows preferentially scour smooth surfaces. This can be quantified by grouping cells that have been scoured and not scoured by debris flows and comparing pre- and post-event roughness distributions of these groups, with 100 cells randomly sampled from each group (Fig. 11). Scoured and nonscoured cells have a median pre-event roughness of 0.032 and

n=127 Table 3 Coefficient results and standard errors of piecewise regression model.

ln(S) Fig. 8. Scatterplot of the log-transformed normalized channel erosion (dmax/Ws) as a function of slope (S) for the Manival and Réal torrents. The 95% confidence interval of the detected slope threshold of regime shift from piecewise regression is indicated by the two vertical lines. The best-fit obtained from the piecewise regression is also indicated.

Coefficients

Estimate

Standard error

a b1 b2 c (break point)

3.768 3.412 −2.915 −1.662 (0.19 m/m; 11°)

0.861 0.449 0.527 0.063

100

J.I. Theule et al. / Geomorphology 243 (2015) 92–105

(A)

(B)

(C)

(D)

Fig. 9. View of the Manival test reach looking downstream during the survey periods: (A) before 25 August 2009 debris flow; (B) after 25 August 2009 debris flow; (C) after 26 September 2009 flood; and (D) after 6 June 2010 debris flow. The channel reach is filled with unconsolidated gravels before debris-flow events and are then evacuated out.

0.049 m, respectively. In other words, the lower the roughness of the channel is before the event, the higher the probability is to observe erosion by debris flows. No significant difference was found between the post-event roughness distributions. However, the post-event scoured cells are significantly higher in roughness than the pre-event scoured cells, this signifies the significant increase of roughness after debrisflow events. Scour depths are correlated with the pre-event roughness that is distributed in 1-cm bins (Fig. 12). The distribution is plotted below the D95 of roughness (17 cm), which gives the upper-bound of roughness values representative of channel surfaces (excluding banks or in-channel steep boundaries). The roughness of 2–3 cm is the highest frequency for scoured cells. There is a linear relation between the scour depth and roughness, with the bins statistically different from one another. This trend extends to the 6-cm roughness, which indicates the extent of the sensitive gravel storage. With the 20-cm grid, the gravel deposits can be automatically mapped (Fig. 10B) according to these empirical observations. Gravel wedges are two classes, 0–0.35 and 0.35–0.6 m, which marks the median and extent of roughness relating to scouring. The next class, 0.6– 0.17 m, covers areas that might be scoured but with little contribution, such as large cobbles and boulders that are embedded in a matrix (debris-flow deposits). Roughness of N 0.17 m covers areas corresponding mostly to permanent structures such as check-dams and steep channel banks. The search window of 1 m is sometimes problematic for capturing the roughness for entire boulders (their tops appear to have a smooth roughness). However, the more sensitive areas (the gravel wedges) are mapped accurately, which is of most concern in active debris-flow channels.

5. Discussion 5.1. Scour and fill by debris flows and bedload transport Field observations on two steep-slope channels based on intensive cross section resurveys after 14 flow events indicated that equilibrium between erosion and deposition is achieved during bedload transport whereas, during debris flows, erosion systematically exceeds deposition. These contrasting scour-and-fill signatures between flow processes reflect distinct sediment transfer processes along these channels. The net erosion of the channels during debris flows is not surprising given the very high transport capacity of such flows and the presence of loose debris in the investigated reaches. During bedload transport, although net deposition or net erosion can be observed at the scale of one specific flow event or along a single reach, the averaging of all the recorded flow events reveals a dynamic equilibrium from which a time-integrated uniform sediment flux can be inferred along the reach. Sediment recharge of the channel during bedload transport can occur locally, with the formation of large gravel deposits as attested by the TLS survey of the test reach, but these macroforms behave like sediment waves propagating downstream under the effect of scour-and-fill processes during bedload transport. They likely reflect pulses of sediment supply from upper catchments, which finally controls scourand-fill patterns of bedload transport along the main channel. These pulses can be induced by the seasonal variability of the sediment production from rockwalls (Loye et al., 2012) or by the episodic failures of colluvial slopes along headwaters. Channel deformations during debris flows significantly exceed those observed during bedload transport. Sediment budget analysis of the two

J.I. Theule et al. / Geomorphology 243 (2015) 92–105 Jul09: Aug09 Aug09: Nov09 (debris flow) (bedload)

Nov09: Jul10 (debris flow)

(A)

Jul09

Aug09

101

Nov09

Jul10

(B) A

A

A

A

A

A

A

Flow

elevation change (m)

A1

A1

A1

A1

0.2

A1

A1

A1

roughness (m)

0.2

% 0.1

%

0 -4

0

4 -4

0

4 -4

0

0.1

0 0

4

0.06

0.17

0.06

0.17 0

0.06

0.17 0

0.06

0.17

Roughness (m)

Elevation change (m)

(C) Elevation (masl)

770 A 765 760 755 750

Jul-09 Aug-09 Nov-09 Jul-10

745 740

A1

735 0

20

40

60

80

100

120

140

160

180

Relative distance (m) Fig. 10. Results of TLS scans in the test reach of the Manival torrent showing the relationship between debris-flow scour and roughness. (A) Elevation differences between survey dates with frequency distribution of scour and fill. (B) Roughness derived from the TLS for each survey date with their frequency distributions in the reach below the D95 (roughness above the D95 is considered noise from steep features such as banks and check-dams). The classes are defined according to the roughness/scour relationship: dark green (median), light green and rose (defined in Fig. 12), and red (above the roughness D95). (C) Channel profile of each TLS survey with July 2010 revealing channel bottom.

investigated torrents previously revealed that debris flows produce 2– 10 times more sediment than bedload transport (Theule, 2012). In the Italian Alps, event-based sediment yield comparisons were also made for equivalent return periods between a debris-flow catchment (Moscardo) and a bedload-dominated catchment (Rio Cordon) (Mao et al., 2009). Their long-term field data also revealed that debris-flow volumes were much larger (2 to 3 orders of magnitude) than bedload transport volumes. We observed that the most important debris-flow scouring (maximum of 4.3 m and D95 of 2.0 m) occurs along reaches where gravel wedges are exposed. Just within the test reach of the Manival, scoured gravels amount to 31–41% of the total debris-flow volume. According to multidate cross section and sediment trap surveys (Theule et al., 2012), ~90% of the debris-flow volumes were entrained where gravel

storage was present. Maximum scouring from bedload transport (maximum of 2.9 m and D95 of 0.85 m) is much less, even if it is also associated with the presence of the deep accumulations of loose debris in the channel before the flow event. Extreme bedload-induced erosion was always associated with the downstream progradation of gravel wedges. These differences, in terms of channel responses, reflect contrasting sediment transport rates and also contrasting flow energy between debris flows and bedload transport. 5.2. Slope effect on channel scour and fill During bedload transport, no significant slope effect on channel scour and fill was detected, whereas a substantial slope effect was observed for debris flows. These contrasting responses suggest distinct

102

J.I. Theule et al. / Geomorphology 243 (2015) 92–105

0.16

(A)

(B)

0.14

Roughness (m)

0.12

0.1

0.08

0.06 (0.049)

(0.046)

(0.044)

0.04 (0.032)

0.02

0

p-value <0.001

Scoured cells

p-value =0.414

Nonscoured cells

Scoured cells

Nonscoured cells

Fig. 11. Notched box plots for precondition surfaces (A) show that debris-flow scouring occurs on a smoother roughness with a 95% confidence level. Post-condition surfaces (B) are rougher because of the eroded gravel wedges and the deposition of debris-flow levees. One hundred random samples were taken in each group from over 207 × 103 samples for the analysis.

Gravel storage

(A) 3

Scour (m)

2.5 2 1.5 1 0.5 0 0-0.01

0.02-0.03 0.04-0.05 0.06-0.07 0.08-0.9 0.10-0.11 0.12-0.13 0.14-0.15 0.16-0.17

0

0.02

(B) 16000 14000

Frequency

12000 10000 8000 6000 4000 2000 0

0.04

0.06

0.08

0.1

0.12

0.14

0.16

Roughness (m) Fig. 12. (A) The roughness relation with scour reveals a linear trend ending at 0.06 m roughness indicating the extent of the gravel storage. (B) Roughness is binned at 1 cm with the highest frequency.

J.I. Theule et al. / Geomorphology 243 (2015) 92–105

primary controls on channel behavior during the two types of flows. The lack of slope effect during bedload transport shows that the transport capacity of the reach does not have a strong influence on bed-level fluctuations, which seems to be much more controlled by fluctuations of sediment supply. This is not the case for debris flows, where the transport capacity of the flow is the critical factor controlling bed-level fluctuations. One way to explain these observations is to consider that sediment concentration during bedload transport and debris flows presents a kind of saturation threshold that cannot be surpassed unless the water-sediment mixture transforms into another form of flow and that therefore influence the capacity of the flow to erode the channel (Egashira et al., 2001). It is also clear that this saturation level is much higher for debris flows, which can commonly reach 70% of sediment concentration by weight and even more. This higher saturation level can explain why sediment supply is not a limiting factor of channel erosion by debris flows, as long as the flow can still incorporate some sediment. Things are different for bedload transport, which may saturate more rapidly once the sediment supply increases. Given the relatively short steps traveled by bedload particles during flow events, even in steep-slope channels where the maximum traveled distances generally do not exceed hundreds of meters (Schneider et al., 2014), the sediment supply of a given reach must be considered as primarily influenced by local conditions. Therefore, channel scour and fill during bedload presents a strong stochastic pattern associated with the variability of channel conditions in the vicinity of the reach before the flow event. This is not the case during debris flows, where conditions far upstream of a given reach may control the sediment supply of this reach. This is a consequence of the long travel distances of particles during debris flows (typically several kilometers), as recently documented using RFID tracers (McCoy et al., 2011). During a debris flow, once a particle has been entrained from the bed, the probability that this particle gets deposited in the channel is low until the flow stops. The databases of debris-flow scouring in the Manival and Réal show a sharp increase of scouring above a slope threshold comprised between 0.15 and 0.16 (Figs. 6 and 7). A similar critical slope of 0.15 (8.5°) was already mentioned in a technical review paper about natural hazards from debris flows, which indicated that below a channel slope of 0.15 (8.5°), the channel is very unlikely to observe initiation of debris flows (Rickenmann, 1995). This is in good agreement with the reported range of slopes (0.14–0.21; 8–12°) below which net deposition occurs for channelized debris flows of British Columbia (Hungr et al., 1984). A recent study proposed a little steeper slopes (0.21–0.26; 12–15°) for the onset of net deposition in this region (Guthrie et al., 2010). The critical slope threshold for normalized debris-flow scouring (0.19; 11°) is much less than the critical slope for which erosion starts to affect granular flow dynamics (Mangeney et al., 2010). The critical slope obtained from laboratory experiments in dry erodible beds (with a tilting flume) was found to be half the repose angle of the erodible material composing the channel. Loye et al. (2011) mapped geomorphic units in the Manival according to slope angle frequency distributions with the debris-flow fan averaging a slope of 0.2 (11°) and the talus slopes 0.78 (38°). If we consider the average talus slope angle as representative of the repose angle of loose debris found in the torrent channels, following Mangeney et al. (2010), a critical slope of 0.34 (19°) is obtained that is significantly larger than the critical slope above which we observed a strong increase of erosion. The disagreement between field and experimental findings can be explained by the influence of pore-water pressures. Mangeney's experiments were conducted on dry erodible beds, and Iverson et al. (2011) found that pore-water pressure significantly increases the erodibility of the bed which boosts the momentum of the debris flow. Investigated debris flows were triggered by high-intensity rainfalls that likely increased the bed sediment water content before the passage of the debris flows. Therefore it is likely that pore-water pressure explains a lower slope threshold for erosion.

103

The normalized scour depth during debris flows for the Manival and the Réal was shown to be controlled by the channel slope, with a change of regime for a threshold slope comprised between 0.17 and 0.21. Above this slope, an asymptotic behavior of the normalized scour depth is observed, suggesting a supply-limited effect related to the upstream decreasing depth of the erodible sediment. Below this slope threshold, the effect of the slope on channel erosion increase is much higher and channel erosion appears limited by the transport capacity of the flow. 5.3. Transition from supply-limited to transport-limited regime The transport- and supply-limited regimes defined by looking at the effect of channel slope on the normalized scour depth during debris flows can be examined by looking at the longitudinal fluctuation of channel slope, combined with channel roughness variations, which was shown to be a good proxy for the identification of loose gravel deposits prone to erosion. This was done for the Manival torrent (Fig. 13). Channel slope and roughness were derived from an ALS survey before the 2009 debris-flow event. The yield rates for the 2009 debris flows (derived from cross section resurveys) are also plotted to see if the combination of slope and roughness can be useful for the identification of the most sensitive reaches for debris-flow bulking. The supply-limited regime (slope N 0.21; 12°) corresponds to the upper limit of the 95% confidence interval determined from the piecewise regression (Fig. 8). It corresponds to the fan apex of the Manival (Fig. 13). This regime corresponds to the asymptotic trend of the normalized scour (Fig. 8), which likely reflects the decreasing thickness of the erodible sediment cover in upper steep-slope channels. The roughness profile shows an increasing trend going upstream supporting that the scour depth is limited by the available sediment cover. A similar pattern was already observed in a database from British Columbia (Guthrie et al., 2010). In the transport-limited regime (slope b 0.17; 10°), a rapid increase of erosion with slope is observed (Fig. 8). During debris flows, they are clearly seen to be transport-limited with an increasing stability of the channel bed with decreasing slope. This lower reach presents a high variation of roughness owing to the interactions of debris-flow transport/deposition and bedload transport. Smooth-surfaced gravel deposits can exist locally, but their erosion by debris flows is limited by the decrease of energy related to gentler slopes. When yield rates of the 2009 debris flows are compared with the slope and roughness profiles, a singular pattern emerges, indicating that the maximum yield rates correspond to the reach where the roughness indicates the presence of loose gravel deposits, and the slope is in the range corresponding to the transition between the supply- and transport-limited regimes. Downstream from this domain, channel erosion by debris flows is limited by the slope. Upstream from this critical reach, the slope is high enough for massive incision to occur, but the roughness profile does not indicate the presence of loose gravel deposits. A balanced sediment budget was found between the sediment trap and the entrained reach, which confirms that little erosion occurred upstream from the fan apex (Theule et al., 2012). This example illustrates how the combination of slope and roughness can be used to detect the most sensitive reach for debris-flow bulking. 6. Conclusion Channel deformations associated with debris flows and bedload transport were intensively surveyed along two active channels by cross section resurveys. Scour depths exceed fill depths for debris flows in the entrainment and transport zone, whereas equilibrium conditions are observed during bedload transport events in the same zones. We found that channel deformation is significantly larger during debris flows than during bedload transport. Channel deformations are influenced by slope for debris flows but not for bedload transport. For debris flows, a critical slope threshold of 0.19 (11°) — above which the

104

J.I. Theule et al. / Geomorphology 243 (2015) 92–105

(A)

(B)

(C)

Fig. 13. Long profile of the Manival from channel head to sediment trap with (A) the upstream slope (S) with the 95% confidence limit of the breakpoint defined in Fig. 8; (B) the averaged bed roughness (window of moving of 17 m, mean channel width) with the roughness threshold of 0.06 (defined in Figs. 10 and 12) where sensitive gravel beds are located; (C) yield rates for the 2009 debris flow measured from multidate cross section surveys, the sediment production from the transition and transport-limited regimes equals the deposition in the sediment trap (derived in Theule et al., 2012). The supply- and transport-limited regimes are separated by the outer limits of the 95% confidence interval of the slope threshold (0.17 to 0.21 m/m). Within this interval is the transition between the two regimes, which is the critical area of entrainment. The transition regime is where the roughness profile corresponds strongly with the yield rates.

normalized scour stabilizes around an asymptotic maximum value — was observed, which is interpreted as a supply-limited regime related to the decreasing thickness of the loose debris cover. The most susceptible materials for debris-flow erosion and most important contribution to the debris-flow volumes are the unconsolidated gravel deposits that are formed during bedload transport. This material has a smooth surface within the rugged channel that can be automatically mapped with a 20cm elevation model from either TLS or ALS by calculating roughness with a 1-m search window.

Acknowledgments This research was supported by the EU Alpine Space PARAmount and SedAlp projects, the Interreg Alcotra Risknat project, and the PGRN (Pôle Grenoblois d'étude et de recherche pour la prévention des Risques Naturels, Conseil Général de l'Isére). The ONF-RTM 38 is acknowledged for facilitating field access to the Manival torrent and the ONF-RTM 06 for the Réal torrent. We thank Bruno Bacq, Adeline Heymann, Sandrine Tacon, Mathieu Cassel, Michael Deschatres, Hugo Jantzi, Mathieu Labbé, Emilien Parisot, Nicolas Talaska, Greg Tucker, Philippe Chavignon, Luc Demirdjian, Eric Travaglini, and Nathalie Andreis for their assistance during the fieldwork. Many discussions with John Pitlick were also deeply appreciated. We finally thank the

two anonymous reviewers and the editor Richard Marston for their help and advice in improving the manuscript.

References Ashmore, P.E., Church, M.A., 1998. Sediment transport and river morphology: a paradigm for study. In: Klingeman, P.C., Beschta, R.L., Komar, P.D., Bradley, J.B. (Eds.), GravelBed Rivers in the Environment. Water Resources Publications, LLC, Highlands Ranch, Colorado, pp. 115–148. Berger, C., McArdell, B.W., Schlunegger, F., 2011. Direct measurement of channel erosion by debris flows, Illgraben, Switzerland. J. Geophys. Res. 116 (F01002). Berti, M., Simoni, A., 2005. Experimental evidences and numerical modelling of debris flow initiated by channel runoff. Landslides 2, 171–182. Besl, P.J., McKay, N.D., 1992. A method for registration of 3-D shapes. IEEE Trans. Pattern Anal. 14, 239–256. Beylich, A.A., Sandberg, O., 2005. Geomorphic effects of the extreme rainfall event of 20– 21 July, 2004 in the Latnjavagge catchment, northern Swedish Lapland. Geogr. Ann. 87A (3), 409–419. Bovis, M.J., Jakob, M., 1999. The role of debris supply conditions in predicting debris flow activity. Earth Surf. Process. Landf. 24 (11), 1039–1054. Cavalli, M., Tarolli, P., Marchi, L., Dalla Fontana, G., 2008. The effectiveness of airborne LiDAR data in the recognition of channel-bed morphology. Catena 73 (3), 249–260. Chambon, G., Richard, D., 2004. Diagnostic du fonctionnement torrentiel du bassin versant du Réal (Péone-Alpes Maritimes). CEMAGREF, Grenoble Unpublished technical report, Cemagref Grenoble, 51 pp. Coe, J.A., Kinner, D.A., Godt, J.W., 2008. Initiation conditions for debris flows generated by runoff at Chalk Cliffs, central Colorado. Geomorphology 96 (3–4), 270–297.

J.I. Theule et al. / Geomorphology 243 (2015) 92–105 Conway, S.J., Decaulne, A., Balme, M.R., Murray, J.B., Towner, M.C., 2010. A new approach to estimating hazard posed by debris flows in the Westfjords of Iceland. Geomorphology 114 (4), 556–572. Davies, R.B., 1987. Hypothesis testing when a nuisance parameter is present only under the alternative. Biometrika 74, 33–43. Decaulne, A., Sæmundsson, B., Pétursson, O., 2005. Debris flows triggered by rapid snowmelt: a case study in the Glei√arhjalli area, northwestern Iceland. Geogr. Ann. 87A (4), 487–500. Egashira, S., Honda, N., Itoh, T., 2001. Experimental study on the entrainment of bed material into debris flow. Phys. Chem. Earth C 26 (9), 645–650. Godt, J.W., Coe, J.A., 2007. Alpine debris flows triggered by a 28 July 1999 thunderstorm in the Central Front Range, Colorado. Geomorphology 84, 80–97. Gregoretti, C., Fontana, G.D., 2008. The triggering of debris flow due to channel-bed failure in some alpine headwater basins of the dolomites: analyses of critical runoff. Hydrol. Process. 22, 2248–2263. Guthrie, R.H., Hockin, A., Colquhoun, L., Nagy, T., Evans, S.G., Ayles, C., 2010. An examination of controls on debris flow mobility: evidence from coastal British Columbia. Geomorphology 114, 601–613. Hungr, O., Morgan, G.C., Kellerhals, R., 1984. Quantitative analysis of debris torrent hazards for design of remedial measures. Can. Geotech. J. 21, 663–677. Hungr, O., Evans, S.G., Bovis, M., Hutchinson, J.N., 2001. Review of classification of landslides of flow type. Environ. Eng. Geosci. VII, 221–238. Hungr, O., McDougall, S., Bovis, M., 2005. Entrainment of material by debris flows. In: Jakob, M., Hungr, O. (Eds.), Debris-flow Hazards and Related Phenomena. SpringerPraxis, Berlin, pp. 135–158. Hungr, O., McDougall, S., Wise, M., Cullen, M., 2008. Magnitude–frequency relationships of debris flows and debris avalanches in relation to slope relief. Geomorphology 96 (3–4), 355–365. Imaizumi, F., Sidle, R.C., Tsuchiya, S., Ohsaka, O., 2006. Hydrogeomorphic processes in steep debris flow initiation zone. Geophys. Res. Lett. 33, L10404. Innes, J., 1983. Debris flows. Prog. Phys. Geogr. 7, 469–501. Iverson, R.M., Reid, M.E., Logan, M., LaHusen, R.G., Godt, J.W., Griswold, J.P., 2011. Positive feedback and momentum growth during debris-flow entrainment of wet sediment. Nat. Geosci. 4, 116–121. Jakob, M., Jordan, P., 2001. Design flood estimates in mountain streams — the need for a geomorphic approach. Can. J. Civ. Eng. 28 (3), 425–439. Jakob, M., Bovis, M., Oden, M., 2005. The significance of channel recharge rates for estimating debris-flow magnitude and frequency. Earth Surf. Process. Landf. 30, 755–766. Jonasson, C., Nyberg, R., 1999. The rainstorm of August 1998 in the Abisko area, northern Sweden: preliminary report on observations of erosion and sediment transport. Geogr. Ann. 81A (3), 387–390. Kotarba, A., 1992. High-energy geomorphic events in the Polish Tatra Mountains. Geogr. Ann. 74A (2–3), 123–131. Loye, A., Jaboyedoff, M., Pedrazzini, A., Theule, J., Liébault, F., Metzger, R., 2011. Morphostructural analysis of an alpine debris flows catchment: implication for debris supply. In: Genevois, R., Hamilton, D.L., Prestininzi, A. (Eds.), 5th International Conference on Debris-Flow Hazards Mitigation: Mechanics, Prediction and Assessment. Italian Journal of Engineering Geology and Environment, Padua, Italy, 14–17 June 2011, pp. 115–125. Loye, A., Pedrazzini, A., Theule, J.I., Jaboyedoff, M., Liébault, F., Metzger, R., 2012. Influence of bedrock structures on the spatial pattern of erosional landforms in small alpine catchments. Earth Surf. Process. Landf. 37 (13), 1407–1423. Mangeney, A., Roche, O., Hungr, O., Mangold, N., Faccanoni, G., Lucas, A., 2010. Erosion and mobility in granular collapse over sloping beds. J. Geophys. Res. 115 (F03040). Mao, L., Cavalli, M., Comiti, F., Marchi, L., Lenzi, M.A., Arattano, M., 2009. Sediment transfer processes in two Alpine catchments of contrasting morphological settings. J. Hydrol. 364 (1–2), 88–98.

105

Marchi, L., D'Agostino, V., 2004. Estimation of debris-flow magnitude in the Eastern Italian Alps. Earth Surf. Process. Landf. 29, 207–220. McCoy, S.W., Coe, J.A., Kean, J.W., Tucker, G.E., Staley, D.M., Wasklewicz, T.A., 2011. 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. In: Genevois, R., Hamilton, D.L., Prestininzi, A. (Eds.), 5th International Conference on Debris-Flow Hazards Mitigation: Mechanics, Prediction and Assessment. Italian Journal of Engineering Geology and Environment, Padua, Italy, 14–17 June 2011, pp. 715–726. McCoy, S.W., Kean, J.W., Coe, J.A., Tucker, G.E., Staley, D.M., Wasklewcz, T.A., 2012. Sediment entrainment by debris flows: in situ measurements from the headwaters of a steep catchment. J. Geophys. Res. 117 (F03016), 25. Muggeo, V.M.R., 2008. Segmented: an R package to fit regression models with broken-line relationships. R News 8 (1). Navratil, O., Liébault, F., Bellot, H., Travaglini, E., Theule, J., Chambon, G., Laigle, D., 2013. High-frequency monitoring of debris-flow propagation along the Réal Torrent, Southern French Prealps. Geomorphology 201, 157–171. Peteuil, C., Maraval, C., Bertrand, C., Monier, G., 2008. Torrent du Manival: Schéma d'aménagement et de gestion du bassin versant contre les crues, Unpublished technical report; Office National des Forêts, Service de Restauration des Terrains en Montagne de l'Isère, Grenoble, France. Remaître, A., Malet, J.-P., Maquaire, O., 2005. Morphology and sedimentology of a complex debris flow in a clay-shale basin. Earth Surf. Process. Landf. 30, 339–348. Rickenmann, D., 1995. Beurteilung von murgängen. Schweiz. Ing. Archit. 48, 1104–1108. Rickenmann, D., Zimmermann, M., 1993. The 1987 debris flows in Switzerland: documentation and analysis. Geomorphology 8, 175–189. Rickenmann, D., Weber, D., Stepanov, B., 2003. Erosion by debris flows in field and laboratory experiments. In: Rickenmann, D., Chen, C.L. (Eds.), Debris-Flow Hazards Mitigation: Mechanics, Prediction, and Assessment. Millpress, Rotterdam, The Netherlands, pp. 883–894. Ryan, S.E., Porth, L.S., Troendle, C.A., 2002. Defining phases of bedload transport using piecewise regression. Earth Surf. Process. Landf. 27, 971–990. Ryan, S.E., Porth, L.S., Troendle, C.A., 2005. Coarse sediment transport in mountain streams in Colorado and Wyoming, USA. Earth Surf. Process. Landf. 30, 269–288. Schneider, J.M., Turowski, J.M., Rickenmann, D., Hegglin, R., Arrigo, S., Mao, L., Kirchner, J.W., 2014. Scaling relationships between bed load volumes, transport distances, and stream power in steep mountain channels. J. Geophys. Res. Earth Surf. 119, 533–549. Schürch, P., Densmore, A.L., Rosser, N.J., Lim, M., McArdell, B.W., 2011. Detection of surface change in complex topography using terrestrial laser scanning: application to the Illgraben debris-flow channel. Earth Surf. Process. Landf. 36 (14), 1847–1859. Theule, J.I., 2012. Geomorphic Study of Sediment Dynamics in Active Debris-flow Catchments (French Alps) (Ph.D. Dissertation). Université de Grenoble, Saint-Martind'Hères, France. Theule, J.I., Liébault, F., Loye, A., Laigle, D., Jaboyedoff, M., 2012. Sediment budget monitoring of debris-flow and bedload transport in the Manival Torrent, SE France. Nat. Hazards Earth Syst. Sci. 12, 731–749. Vedin, H., Eklund, A., Alexandersson, H., 1999. The rainstorm and flash flood at Mount Fulufjället in August 1997: the meteorological and hydrological situation. Geogr. Ann. 81A (3), 361–368. Veyrat-Charvillon, S., Memier, M., 2006. Stereophotogrammetry of archive data and topographic approaches to debris-flow torrent measurements: calculation of channelsediment states and a partial sediment budget for Manival torrent (Isère, France). Earth Surf. Process. Landf. 31, 201–219.