Debris-flow in the Dolomites: Experimental data from a monitoring system

Debris flows in Dolomites: experimental data from a monitoring system

Genevois R., Tecca P.R., Berti M., Simoni A.

ABSTRACT: Field measurements of some dynamic properties of a channelised debris flow were obtained from an automated monitoring system at Acquabona, Italian Alps. Ground vibrations, videoimage analyses, and ultrasonic logs, were used to obtain front velocity, instantaneous urface velocity and depth for three debris flows.All debris flows moved as a series of surges characterised by a steep coarser front having a curbing effect, followed by a more fluid muddy tail. Front velocities were always lower than instantaneous surface velocities and appeared to be affected by frictional resistance encountered by the front, and by slope and depth. Development of rigid plugs is not constant and appears related to changes in concentration of coarser particles during a single surge as well as between different surges of a same event. Also estimated Newtonian viscosities are indicative of variations in flow characteristics during different stages of a debris flow and over a single surge.

1 INTRODUCTION The flow of debris-water mixtures implies the interaction of complex processes including grain-grain, grain-fluid and fluid-fluid interactions, whose relative importance is ruled by several factors as grainsize distribution, solid concentration and strain rate. Some attempts have been made to describe all the physical processes involved (Iverson, 1997). The application of rheological models, mainly derived from laboratory experiments (Bagnold, 1954; Coussot, 1997; Coussot and Piau, 1994; Major and Pierson, 1992), to real cases is critical to the definition of their field of applicability. In order to apply any rheological model to the study of debris flows, the parameters that govern the chosen constitutive equation need to be determined. One of the major problems related to the definition of rheological parameters suitable for actual debris

flow materials is that significant changes in solid concentration, and therefore, in rheological behaviour, can occur within a single debris flow wave as well as in different debris flow events. Debris flow surges usually show a steep front, where the coarsest material accumulates due to dynamic segregation processes (Suwa and Okuda, 1985; Suwa, 1988), followed by the body of the flow and by a long tail where flow depth and solid concentrations decrease progressively. Although this kind of behaviour has been widely observed in the field (Pierson, 1986) as well as in laboratory experiments, data on the variability of the rheological characteristics of debris during the flow process are scanty. Field monitoring of actual debris flows is of the outmost importance to validate the descriptive capabilities of rheological models. Important parameters such as density of the flowing debris and velocity distribution inside the flowing mass are quite difficult to measure in the field, whilst velocity and flow can be easily measured and can be utilised to define approximately the shear stress acting on the flowing debris and the resulting strain rates (Phillips and Davies, 1991). Debris flow depth and velocity have been measured in several monitoring field sites devoted to specific research purposes around the world: Japan (Suwa and Okuda, 1985); Philippines (Marcial et al., 1996); Indonesia, Ecuador and Columbia (LaHusen, 1996); Italy (Arattano et al., 1997). In 1997 the Dept. of Geology of Padova University, in the context of the E.U. Debris Flow Risk Project (Genevois et al., 1999), began monitoring activities on Acquabona Creek (Eastern Alps, Italy), where debris flows occur yearly, triggered by intense localised thunderstorms in summer and early autumn. In this paper, some aspects of the rheological behaviour of channelised debris flows have been investigated through the analysis of data on propagation front velocities, instantaneous surface velocities, and flow depth, measured in the lower part of the channel during summer 1998.

Figure 1. Location of Acquabona monitoring site.

2.2 Physical characteristics of debris

2 STUDY SITE

2.1 Geological and geomorphological setting The Acquabona catchment (Berti et al., 1999) is located on the left side of Boite River valley (Eastern Dolomites) (Fig. 1). The upper drainage basin is made of massive dolomite and limestone cliffs. The channel is deeply incised, mostly into the thick talus that covers the slope from the base of the rock cliff; in a 150 m long reach, it is cut into red marl and limestone bedrock. Talus includes heterogeneous scree, alluvium and debris-flow deposits, and consists of poorly sorted debris containing boulders up to 3-4 m in diameter. The main morphometric parameters of the Acquabona upper drainage basin and flow channel are listed in table 1.

Acquabona debris flows transport poorly-sorted gravelly-sandy material, and may include angular dolomite boulders exceeding 2 m in diameter. Grainsize analyses were carried out on the fraction finer than 20 mm of samples collected in the initiation area, along the flow channel and in the deposition area (Fig. 2). This matrix exhibits very low plasticity (PI = 3 to 4%). Particle-size distributions from these areas display similar values of average grain size (2.5 to 3.0 mm) but can have significant differences in the silt and clay fractions: the fine fraction does not exceed 10% upstream of the marl outcrop, but increases to about 30% downstream of the marls and in the deposition area.

Table 1. Main morphometric parameters of Acquabona upper drainage basin and channel. Upper drainage basin area (km2) Basin maximum elevation (m a.s.l.) Basin outlet elevation (m a.s.l.) Average rock basin slope (°) Channel length (m) Channel slope in the initiation area (°) Channel slope in the deposition area (°) Average channel slope (°)

0.30 2667 1650 43 1632 30-45 7 18

b a

c

Figure 2. Grain size distribution envelopes of Acquabona debris flows matrix. a: source area and upstream marls flow channel; b: downstream marls flow channel; c: deposition area.

Particle size distribution of the fraction coarser than 20 mm was carried out in the field on samples collected in the upper initiation area, both by "transect line sampling" (by measuring the diameter of the clasts) and by in situ-sieving and weighing. Coarse debris ranges from 22 to 28% by weight of the total size distribution; however, at fronts and tops of deposits the coarsest fraction increases up to 75% by weight, demonstrating the action of segregation processes (Suwa, 1988). 2.3 Monitoring System The automated monitoring system, installed in summer 1997, consists of three field stations and one off-site master collection station (Berti et al., 2000) and is equipped for measuring rainfall, wind, pore pressures (both in the talus and in the channel bed deposits), flow depth, flow-induced ground vibrations, pore pressure and total pressure at the base of the flowing mass and for recording visual data in the initiation area and in the lower channel. Figure 1 shows the location of the monitoring stations. In preevent mode, data are recorded and radio transmitted every 5 minutes to the off-site base station; in event mode, activated when threshold values of both ground vibrations and rainfall intensity are exceeded, data are recorded every 0.2 sec and stored in a memory support. The station n. 3 (Fig. 3) has been equipped with three geophones installed along a reach of the lower channel at a distance of 100 m from each other; a pressure cell and a shallow pore-water sensor inserted in a steel box within the channel bed material; an ultrasonic sensor and a video system, composed of a camera and VCR, suspended perpendicularly to the flow surface (Fig. 3), framing the left half-width of the channel. Channel gradient in the instrumented reach at station n. 3 is 6°. ultrasonic sensor

camera

a

d

s

geophone

pressure cell and transducer

Figure 3. Measurement devices at monitoring station n. 3. a: antenna; s: solar panel; d: data logger.

cles on the flow surface, geophone recordings and ultrasonic logs. The video system recorded images of flows to observe the overall appearance of flow, the qualitative size of the front, and to measure flow front and flow surface velocities. By tracking the motion, scaled to actual distance, of single particles among successive video images recorded during the passage of debris flows, surface velocities were estimated at the front and along the channel centreline. The time interval between video frames is very short (0.4 sec) and therefore, the surface velocity can be considered as an instantaneous velocity. In the same way, horizontal velocity profiles were determined. The average velocity of a surge front was obtained from the time interval between the passage of a single debris surge (signalled by a distinct peak in the ground acceleration) and the distance between two neighbour geophones. The ultrasonic sensor was used to measure the flow depths and to obtain depth hydrographs of the flows. 4 RESULTS Data analysed in this study are related to three debris flows that occurred in summer 1998, and whose basic characteristics are summarised in table 2. The largest event of August 17 was related to the highest rainfall peak intensity (16 mm / 10 min) and total precipitation; the July events were triggered by rainfalls characterised by lower intensity (6 mm / 10 min) and total precipitation (11-12 mm). 4.1 Description of flow processes Observations of some physical features common to the recorded debris flows were based on video evidence. Each flow consisted of a highly concentrated mixture of granular solids (boulders to clay) and water which propagated downslope as a series of surges. Each debris surge was characterised by a steep coarser front, displaying the maximum flow dept. Behind the surge front, flow depth progressively decreased, and videos showed an apparent decline of sediment concentration, causing the final tail to appear as a thin flow of muddy water. Videos of flows showed that these changes in sediment concentration occurred between different surges of a same event as well as during a single debris flow surge. 4.2 Flow velocity and flow depth

3 METHODS Data analysis has been carried on utilising video recordings of the flow, timing of single coarse parti-

Debris flow surface velocity and front velocity of the flow were both measured. Front velocity values measured from geophone logs, including different events and different surges of a same flow, have

Total volume (m3) Date 25.7.98 27.7.98 17.8.98

600-700 400-500 8000-9000

Rainfall 10 min intensity (mm) 6 6 16

Total (mm) 11.4 12.4 29

Duration (min) 90 40 58

Mean front velocity range (m/sec) 0.47 – 0.83 0.77 – 1.17 1.82 – 7.69

Table 2: Main characteristics of summer 1998 debris flows.

been plotted versus front velocity values estimated by video-image analysis: whatever determined, similar values have been obtained and plot well along the 1:1 line (see dots in Fig. 4). In the cases where peak surface velocity (by video-image analysis) and front velocity were measured for the same surge, peak surface velocity values were compared to front velocity values (see cross marks in Fig. 4): front velocity was always lower than surface velocity.

Fig. 4 – Front velocity from geophones logs vs. front velocity from image analysis and surface peak flow velocity. dots: front velocity; cross marks: peak surface velocity from image analysis.

During the July 25 event, two debris flow surges were observed at station n. 3 (Fig. 5). The first surge had a maximum height of 150 cm and was followed by a smaller (85 cm) surge; the overall duration of the debris flow was about 8 minutes. The first surge had an average front velocity, determined by geophones, as much as 0.5 m/s; a similar value (0.55 m/s) was obtained by video-images analysis (see Fig. 4). Peak surface velocity, determined along the channel centre-line, was as much as 2.2 m/s and was achieved just behind the debris flow front. On July 27, a single debris flow surge, with a maximum height of 130 cm (Fig. 6), passed station n. 3. Front

velocity, however measured (by geophones = 0.97 m/s; by image analysis = 0.72 m/s), was lower than surface velocities measured after the passage of the bouldery snout (peak value of 2.36 m/s). On August 17, 1998 the ultrasonic sensor at station n. 3 recorded the passage of more than 10 debris flow surges over 30 minutes. Unfortunately, due to the oncoming darkness, only three main debris flow surges were clearly observable in the recorded videos, allowing the surface velocity measurement only for the initial stages of the event (Fig. 7 a, b). Although velocities, depths, and hence, total volume of this event were significantly higher than those characterising the July flows, the trend in variation of the instantaneous surface velocity was similar to the July debris flows. Higher surface velocities were always achieved in correspondence with the debris flow front or just behind it. This differential between surface centre-line velocity and front velocity has been observed at other sites (Pierson, 1986; Suwa et al., 1993) and is mainly due to the frictional resistance encountered by the debris flow front. Half-width horizontal velocity profiles were obtained by video-image analysis. Distinct particles were tracked and velocities calculated for points from the channel centre-line to its edge. All three debris flows exhibited rigid plugs, represented by constant velocity zones in the central half of the channel, but the presence of such plugs was not constant during a same flow. During the July 25 debris flow, while flow depth was 139 cm, shear was distributed throughout the cross section (Fig. 8 a) but 25 seconds later, after decreasing 18 cm in depth, the same flow displayed a rigid plug that occupied approximately 60 percent of the total flow width (Fig. 8 b), with shear concentrated along a zone of about 90 cm at the flow margin. The horizontal velocity distribution for the July 27 debris flow is shown in figure 9. The rigid plug occupied about 60 percent of the total width, when the flow depth was 121 cm. The half-width horizontal velocity profile during peak flow of a surge (140 cm) of the August 17 debris flow, shows a rigid plug occupied approximately 63 percent of the total width, with shear concentrated along a band about 72 cm wide at the flow edge (Fig. 10).

Figure 5 – Variations in Newtonian viscosity, instantaneous surface velocity and flow depth during July 26 debris flow. Arrows on instantaneous surface velocity plot indicate front velocity by video-image analysis.

Figure 6 – Variations in Newtonian viscosity, instantaneous surface velocity and flow depth during July 27 debris flow. Arrow on instantaneous surface velocity plot indicate front velocity by video-image analysis.

a

b

Figure 7 (a), (b) – Variations in Newtonian viscosity, instantaneous surface velocity and flow depth during different surges of August 17 debris flow. Arrows on instantaneous surface velocity plot indicate front velocity by video-image analysis.

a

b

Figure 8 – Horizontal velocity distribution during peak flow of the July 25 debris flow. (a): 344 sec after debris flow beginning; (b): 369 sec after debris flow beginning.

Figure 9 – Half-width horizontal velocity profile during July 27 debris flow.

Figure 10 – Half width horizontal velocity profile during August 17 debris flow.

5 DISCUSSION Investigations on the variability of the rheological properties can be made through the analysis of debris flow depth and velocity data. Debris flow velocity is a function of independent variables, such as channel slope, channel-bed roughness, flow depth and sediment concentration, this last variable being responsible for the internal resistance of the flow to deformation. Since channel slope is constant (6°) in the lower instrumented channel reach, and roughness, although variable for scouring and deposition effects, can be considered of the same order, velocity should be largely controlled by sediment concentration and flow depth. Maximum surface velocities for each surge were observed in correspondence with the highest flow depths, approximately equal to the height of the front, or just behind it. After the passage of the bouldery front, surface velocity progressively decreased, as well as flow depth. Surface velocity then increased in the more dilute tail.

Total shear strength and coefficient of Bingham viscosity of moving flows have been determined by measuring, wherever possible, the velocity distribution across the surface of the debris flows, based on the method developed by Johnson (1970). Assuming a semi-elliptical channel, shear strength can be computed (Pierson, 1986): K=

(Wp / 2)γ d sinS (W / 2d ) 2 + 1

where K is shear strength (Pa), Wp is depth of plug (m), W is width of flow (m), d is flow depth (m), γd is unit weight (density times gravity) of slurry (N/m3) and S is slope angle (deg). If density of slurry is approximately 2.0 g/cm3 (a common value by Pierson, 1986), shear strength values range from 611 to 850 Pa. The estimated strength values for Acquabona debris flows are similar to shear strength values estimated by Pierson (1986) for Shoestring, Washington debris flows (780 to 840 Pa), and slightly higher than values measured by Johnson (1970) for debris flows at Wrightwood, California (600 Pa), mainly due to larger channel and plug widths observed for Acquabona debris flows. As visual data and horizontal velocity distributions suggest, a range of shear strengths and viscosities may occur in different debris flows at the same site and even during the same debris flow. Bingham viscosity of moving flows has been computed, wherever possible, using the expression

derived from Johnson (1970) and Johnson and Rodine (1984):

[

]

µ b = ( KW p / 4v m ) (W / W p ) − 1

2

where µb is Bingham viscosity ( Pa ⋅ s ) and νm is velocity of the plug which, according to Figs. 8, 9, 10, at specific points of the hydrographs, was about 1.2, 1.5 and 8.5 m/s for different flows (other variables are defined as above). The estimated µb values for Acquabona July debris flows were 127 to 178 Pa ⋅ s ; this range is similar to Bingham viscosity values computed by Johnson and Rodine (1984) for Wrightwood flows (130 Pa ⋅ s ). The estimated Bingham viscosity for the larger debris flow of August 17, 1998, was about one order of magnitude lower (21 Pa ⋅ s ) than the smaller July debris flows. This value is comparable with Bingham viscosity estimated Pierson (1986) for Shoestring debris flows (8 – 12 Pa ⋅ s ). Ignoring the strength, the Newtonian viscosity, or apparent viscosity, has been calculated by the equation: γ d sinSd 2 µn = 2v s where µn is Newtonian viscosity ( Pa ⋅ s ), d is flow depth (m), νs is surface velocity (m/s) and other variables are as defined above (Costa, 1984). Typical values of apparent Newtonian viscosity for Acquabona debris flows range between 60 to 1700 Pa ⋅ s . Although the assumptions related to apparent viscosity (ideal Newtonian fluids, no slip on the base or shear stress on the surface, and laminar flow parallel to the base) are not valid for debris flows, calculated apparent viscosities can give relative values for comparisons to other flow parameters. Newtonian viscosity has been determined at several points of the hydrographs of the three debris flows (Figs. 57). Along a typical surge, the gradual decline in apparent viscosity with time suggests that a progressive decrease of overall solid concentration (increase in water content and/or reduction of the coarser fraction) must take place within the surges. At the flow tail, as video images show, the flow appears as a hyperconcentrated flow. The observed trend (gradual decline in apparent viscosity with time) is valid for any values of flow velocity (1 – 8 m/s) and flow depth (0.8 – 1.8 m). Both surface velocity and Newtonian viscosity trends confirm that flow properties can not be considered as constant even over a single pulse.

6 CONCLUSIONS Data on propagation front velocity, surface velocity and flow depth of three debris flows have been recorded at a monitoring site in the Eastern Italian Alps during 1998. The research is continuing, but based on the data so far collected, some preliminary considerations about debris flow processes can be drawn. 1 Average front velocity of single surges was always significantly lower than the instantaneous surface velocity (front velocity = 30-90% of maximum surface velocity). The observed curbing effect of the debris flow front is related to frictional resistance due to the higher concentration of coarse sediment at the flow front, as shown by video-recording. Surface velocity, as well, appears to be affected by sediment concentration: more dilute tails are often, but not always, faster-moving. 2 Observed horizontal velocity distributions suggest that a range of shear strengths and viscosity may occur in different debris flows and even in single surges of the same flow. 3 The Newtonian viscosity trends for all the monitored flows indicate relative changes of the flow characteristics during different debris flows and within individual surges. 4 Common rheological models with constant physical and rheological parameters can not describe sufficiently flow behaviour and characteristics. 5 Measurements of flow velocities and depths of some real debris flow events in the Alps show that flow resistance varies widely within the same surges. Relatively small variations of the water content in grain-clay-water mixtures can influence dramatically their viscosity, as has been already shown in some laboratory experiments (Coussot, 1997). 6 More comprehensive and reliable experimental data are needed to better understand debris flow processes so that rheological models can be tested and improved. These applications are essential for the development of analytical or empirical debris flow models aimed to countermeasures design and warning systems development, in order to reduce the societal risk in debris flow prone areas.

Berti M., Genevois R., Simoni A., Tecca P.R. 1999. Field observations of a debris flow event in the Dolomites. Geomorphology, 29: 265-274. Berti M., Genevois R., LaHusen R., Simoni A., Tecca P.R. 2000. Debris flow monitoring in the Acquabona watershed (Dolomites, Italian Alps). Physics and Chemistry of the Earth, special issue of Proc. of EGS XXIV General Assembly, 19-23 April 1999 The Hague, Netherlands, NH6.4: monitoring, modelling and mapping, in press. Costa J. E. 1984. Physical Geomorphology of Debris Flows. In E. Costa & P. J. Fleischer (eds), Developments and Applications of Geomorphology: 257-361. Berlin, Heidelberg: J. Springer-Verlag. Coussot P. 1997. Mudflow rheology and dynamics. IAHR Monograph, Rotterdam: Balkema. Coussot P. & Piau J.M. 1994. On the behaviour of fine mud suspensions. Rheol. Acta, 33: 175-184. Genevois R., Berti M., Simoni A., Tecca P.R. 1999. Debris flow monitoring and analysis in the Dolomitic Region (Upper Boite Valley, Italian Alps). In Debris flow management and risk assessment in the Alpine region, unpub. final report of E.U. Debris Flow Risk project (ENV4.CT96.0253), 2: 21.1-21.58. Iverson R.M. 1997. The Physics of debris flows. Reviews of Geophysics, 35(3): 245-296. Johnson A.M. 1970. Physical Processes in Geology. San Francisco: Freeman, Cooper. Johnson A.M. & Rodine J.R. 1984. Debris flow. In D. Brunsden & D.B. Prior (eds), Slope Instability: 257-361. Chichester: Wiley. LaHusen R.G. 1996. Detecting debris flows using ground vibrations. USGS Fact Sheet: 236-96. Major J.J. & Pierson T.C. 1992. Debris flow rheology: experimental analysis of fine-grained slurries. Water Res. Research, 28(3): 841-857. Marcial S., Melosantos A.A., Hadley K.C., LaHusen R.G. and Marso N. 1996. Instrumental lahar monitoring at Mount Pinatubo. In: Newhall C.G. and Punongbayan R.S. (eds), Fire and mud: eruptions and lahars of Mount Pinatubo, Philippines: 1015-1022. Seattle: University of Washington Press. Phillips C.J. & Davies T.R.H. 1991. Determining rheological parameters of debris flow material. Geomorphology 4: 101110. Pierson. T.C. 1986. Flow behavior of channelized debris flows, Mount St. Helens, Washington. In A. D. Abrahams (ed), Hillslope Processes: 269-296. Boston: Allen and Unwin. Suwa H. & Okuda S. 1985. Measurements of debris flows in Japan. In Proc. IV Int. Conf. and Field Workshop on Landslides, Tokyo, August 1985: 391-400. Suwa H. 1988. Focusing mechanism of large boulders to a debris flow front. Transactions, Jap. Geomorph. Union 9(3): 151-178. Suwa H., Okunishi K. & Sakai M. 1993. Motion, debris size and scale of debris flows in a valley on Mount Yajkedake, Japan. In Sediment Problems: Strategies for Monitoring, Prediction and Control; Proc. of the Yokohama Symp., July, 1993: 239-248. IAHS Publ. 217.

REFERENCES Arattano, M., Deganutti, A.M. and Marchi, L. 1997. Debris Flow Monitoring Activities in and Instrumented Watershed of the Italian Alps. In Cheng-lung Chen (ed.), Debris-flow Hazard Mitigation: Mechanics, Prediction, and Assessment; Proc. First Int. Conf., San Francisco, California, August 7-9, 1997: 506-515. New York: ASCE. Bagnold R.A., 1954. Experiments on a gravity-free dispersion of large solid spheres in a Newtonian fluid under shear. Proc. of the Royal Soc. of London, 225: 49-63.

ACKNOWLEDGEMENTS The research activities in the Acquabona Creek have been carried out in the context of a Research Project funded by the European Union (Contract no. ENV4 CT96-0253).