Italiano  IT - Inglese  EN
Bacino Idrografico del Torrente Moscardo
Video 4,7 MB

Geological and morphological characteristics of the catchment

The bedrock of the Moscardo basin is made of flysch, dating back to the Carboniferous, represented by shale, slates, siltstone, sandstone and breccia that continuously outcrop in the upper portion of the basin and locally along the torrent (Arattano et al., 1997). The poor mechanical properties of the rocks, the instability caused by the presence of a deep seated gravitational slope deformation and the steepness of the upper basin slopes facilitate frequent widespread falls and toppling of rocky masses that continuously feed the talus and cause a withdrawal of the watershed divide. The very large quantity of debris that reaches the drainage network by gravitational and erosion processes and by snow avalanches explains the strong tendency of the Moscardo Torrent to generate debris flows and hyperconcentrated flows. The succession of these impulsive manifestations led to the construction of a huge fan that progressively invaded the receiving valley bottom forcing the But River to flow on the west side of the valley (Arattano et al., 1996).

The main morphological parameters of the basin closed at the fan apex are listed in Table 1; the length of the channel reach on the fan is about 1000 m.

Basin area [Km2] Maximum elevation [m] Minimum elevation [m] Average elevation [m] Average basin slope [%] Average channel slope [%] Channel length [m]
4,1 2043 890 1488 63 37 2760


Hystory of the installations

The first installations of monitoring devices in the Moscardo Torrent date back to 1989 when two ultrasonic sensors were installed along a straight torrent reach in the mid-fan area at a distance of 300 meters from each other. The ultrasonic gauges were anchored to two concrete structures placed on the banks; the recording equipment was recovered in a cabin ten meters from the left channel bank, close to the upstream sensor. A raingauge was also installed in the upper portion of the basin slightly outside of this latter (Arattano et al., 1997).

The reasons that led to the choice of this catchment were the following:

  • high frequency of debris flows occurrence (the basin was known to produce at least one debris flow every year)
  • easy accessibility to the fan, with an available energy supply
  • a deeply cut and well defined channel on the fan, not subject to frequent diversions

Between 1989 and 1994 twelve debris flow events occurred that were recorded by these monitoring devices (Arattano et al., 1997). These first results, particularly the relevant number of phenomena that had occurred in five years, were encouraging for the continuation of the activities in this basin and the addition of more equipment. Thus in 1995, with some funding made available through a European Research Project (Contract n EV5V CT94 0453), a third ultrasonic sensor was placed 150 m upstream of the two that already existed, a fixed video camera was installed close to the intermediate of these three ultrasonic sensors and a network of seismic detectors was set up about 1 km upstream from the ultrasonic gaging stations. The monitoring activities are continuing in the Moscardo Torrent within the context of the European Research Project "Debris Flow Risk" (Contract n ENV4 CT96 0253).

Topographic surveys carried out in the basin

The cross sections at the gaging stations and the longitudinal profile of the channel are topographically surveyed after the events allowing the calculation of the maximum discharge and the total volumes involved (Arattano et al., 1996, 1997). Several boulders lying along the torrent were marked with paint in the middle and upper portion of the basin and on the fan. The travel distances of these boulders were surveyed. Some of them have been found to move several hundreds meters even on the fan, where the slope of the channel is not very high (10%). Usually these boulders were deposited on the same bank where they lay before the event, indicating the absence of transversal forces in the moving mass.

In some cases, however, the boulders were found deposited on the opposite bank, suggesting that such transversal forces may occasionally be present in the flowing mixture.

Grain size analyses

The Moscardo Torrent debris flows usually carry material of very heterogeneous dimensions, that sometimes may reach volumes of several cubic meters. Particle size analyses have been carried out on the matrix of several samples collected within debris flow deposits left by different events on the fan. The particle size distribution envelope, obtained excluding particles with a diameter larger than 32 mm indicates the absence of significant variations among the materials of different events (Arattano et al., 1997). The grain size distribution of a complete debris flow sample collected after a debris flow occurred on July 5, 1995 has also been determined (Coussot et al., 1996, 1998).

Rheometrical tests carried out in the Moscardo catchment

In 1995 some rheometrical tests were carried out on three samples collected from a deposit left by a debris flow occurred on July 5. From these tests the rheological properties of the complete material composing the debris flow have been obtained revealing a shear thinning behaviour with a yield stress (Coussot et al., 1996, 1998). In the picture above, rheometer used for some of the tests (Property of CEMAGREF, Grenoble).

The rainfalls

Debris flows in the Moscardo Torrent are triggered by precipitation events of different characteristics with regard to duration, total (cumulated) rainfall and average and maximum intensity (Arattano et al., 1997). However the rainfall intensities that trigger debris flows in the Moscardo Torrent are lower than those required to induce instability phenomena or erosion in the surrounding basins.

For a comparison with other geographical regions the well-known relationship proposed by Caine (1980) may be used:

I = 14.82 D -0.39


In this formula, based on a number of cases coming from all over the world, D represents the duration of the event in hours and I the rainfall intensity in millimetres per hour. Average rainfall intensities that caused debris flows in the Moscardo Torrent versus rainfall duration often fall below the threshold identified by Caine, 1980 (Figure below).This could be a consequence of local conditions (slope steepness and poor mechanical characteristics of materials). However the space scale of the investigation and the criteria for the inclusion of debris flows may also affect the identification of the threshold (Johnson and Sitar 1990). A rainfall intensity-duration relationship based on individual failures in a spatially-limited area (Wieczorek, 1987) fits well the threshold values observed in the Moscardo basin.

Within the basin and disposition along the reach

A debris flow is a moving source of ground vibrations, consequently two velocities are involved in the process: the velocity of the moving mixture of water and debris itself (Vd) and the propagation velocity of the vibrations (waves) that it induces in the ground (Vw). This must be taken into account as far as monitoring activities through seismic detectors are concerned and the values of these two velocities, in the particular basin chosen for the investigations, have to be previously estimated for a convenient disposition of the detectors along the torrent. One type of wave that a debris flow, as any other ground vibration source, can generate in the ground is the compressional or P-wave. The propagation velocity of compressional waves in rocks like those present in the Moscardo basin, that is slate, shale, sandstone, limestone and breccia (Arattano et al., 1997), ranges between 700 and 6000 m/s (Telford et al., 1976). These velocities are certainly much higher than those reachable by a debris flow wave propagating along a natural channel. However the banks of the Moscardo torrent, where the detectors are placed, mainly consist of alluvium and in this type of rock P-wave velocity may be much lower. The lowest possible value of this latter velocity must be known, at least its order of magnitude, to verify whether it is comparable or not with the propagation velocity of a debris flow wave. This latter occurrence might in fact impede a distinction, in the interpretation of the recordings, between these two velocities and the related phenomena. Some data on the propagation velocity of seismic waves in the type of alluvium present in the zone where the Moscardo torrent is located were already available. Seismic surveys had in fact been carried out in this latter zone after the earthquake that hit the Friuli Venezia Giulia region in 1976. These surveys showed that the lowest velocity of P-waves in this alluvium is of 350 m/s (Manfredini, 1977).

Another type of wave that can be produced by the passage of a debris flow in a torrent and detected by a seismometer is the transversal or S-wave. S-waves velocity, , is a fraction of the velocity of P-waves, , that depends on the Poissons ratio, , according to the following relationship (Telford et al., 1976):

Even assuming for the Poissons ratio, , a value of 0.45 that holds for soft, poorly consolidated materials like those present in alluvium, would result greater than 100 m/s. (Telford et al., 1976). There are other types of waves, besides S and P waves, that might be involved in the process, however their velocities should still range between 100 and 350 m/s (Telford et al., 1976). As previously stated also the propagation velocity of debris flow waves has to be known for comparison. Some velocity data were already available that had been measured in the Moscardo Torrent between 1989 and 1994 (Arattano et al., 1997) through ultrasonic gauges placed on the fan, where the channel slope is about 10%. Mean front velocities of twelve debris flows occurred during that period ranged between 1 and 10 m/s. These values were comparable with those already observed by other authors elsewhere (Pierson 1985, 1986; Suwa, 1989; Pierson et al., 1990; Takahashi, 1991). At the seismic site velocities might have been higher for larger events and they might have also been higher for the greater steepness of the channel at that position (about 15%). However mean flow velocity usually depends on the square root of channel slope (Pierson, 1986; Takahashi, 1991), thus only a difference of few meters per second at most could have occurred at the seismic site because of its greater steepness. A difference of one order of magnitude at least was therefore expected between the velocity of debris flows similar to those observed and the propagation velocity of the ground vibrations induced by them.

On the basis of these considerations four seismic detectors were placed at a distance of 100 meters from each other along the right bank of a straight reach of the torrent located in the lower basin few hundred meters upstream of the fan apex (Figure 3a). The distance between the detectors was chosen to guarantee a good approximation of the debris flow velocity estimations obtainable through the seismic signals. These estimations were thought possible identifying in the recordings one or more clearly recognizable features having constant time lags between their occurrence at consecutive sensors. Such features could have been ascribed to the passage of some peculiar portion of the debris flow, such as the main front or a secondary wave, in the vicinity of each sensor. A typical debris flows is in fact a large wave with a steep front, consisting mostly of large boulders, usually followed by superimposed, smaller waves having lower front heights (Johnson and Rodine, 1984) and debris flows showing such characteristics had been already observed in the Moscardo Torrent (Arattano et al., 1997). The presence of boulders in the main front of a debris flow and in its following surges was expected to generate particularly intense ground vibrations that could have been traced back in the seismographs as recognizable features of the recorded signal. The ratio of the distance between each couple of consecutive detectors to the time lag between the occurrence at those detectors of one of these particular features would have led to a velocity estimation for this latter and for the entire event. The value of this time lag would have had to be of several seconds, that is consistent with the time needed by the wave to travel the distance between two consecutive sensors. For a debris flow velocity as high as 10 m/s, a distance of 100 meters between each couple of sensors would have led to an average error of about 10% in its estimation.

Another reason that led to the choice of the distance between the sensors and their location within the basin (Figure 3a) was the attempt to record the moment of movement inception. Debris flows triggered in the upper basin by a landslide or by the collapse of a channel dam was in fact thought capable of producing ground vibrations strong enough to travel the distance between the point of their occurrence and the seismic network. A distance of one hundred meters between the sensors should have caused an almost simultaneous recording of such an occurrence at the four sensors, considering the propagation velocity of the ground vibrations previously estimated (> 100 m/s). This would have facilitated its recognition, impeding any confusion with recordings due to the passage of a debris flow wave in the vicinity of the sensors. Moreover positioning the network in the lower basin (Figure 3a) was thought a good compromise among different needs: an eventual location on the fan, close to the gaging stations, might have been too far from the potential inception point: the eventual vibrations induced by the triggering of a debris flow might have dissipated before reaching it. On the other side no safe place was available in the upper basin because of its very steep and unstable nature.

An easy accessibility was also needed for the installation in order to be able to carry the equipment in place, to install it and subsequently perform inspections and management. The chosen location in the lower basin provided both these requirements. The presence of a road used by the forest service to inspect the check dam indicated in figure 3b allowed in fact the accessibility, while the presence of a long, straight embankment that flanked the torrent provided a safe place for the installation of the four detectors.

The first installations

The first installations of monitoring devices in the Moscardo Torrent date back to 1989 when two ultrasonic sensors were installed along a straight torrent reach in the mid-fan area at a distance of 300 meters from each other. The ultrasonic gauges were anchored to two concrete structures placed on the banks; the recording equipment was recovered in a concrete cabin ten meters from the left channel bank, close to the upstream sensor.

A raingauge was also installed in the upper portion of the basin slightly outside of this latter (Arattano et al., 1997).

The new installations

Between 1989 and 1994 twelve debris flow events occurred that were recorded by the monitoring devices originally installed (Arattano et al., 1997). These first results, particularly the relevant number of phenomena that had occurred in five years, were encouraging for the continuation of the activities in this basin and the addition of more equipment. In 1995, with some funding made available through a European Research Project (Contract n EV5V CT94 0453), a third ultrasonic sensor was placed 150 m upstream of the two that already existed. In 1993 the right bank of the creek had been severely eroded in the vicinity of the downstream station; this caused an enlargement of the cross section and a diversion of the channel toward the right bank. The main effect of this diversion was to impede the monitoring of the center of the flow as previously happened, because the sensor, after the erosion, was hanging over the left side of the creek and the concrete structure that sustained it could not be easily modified to face such a morphological change.

Thus the concrete structures were abandoned and all the existing sensors were suspended over the channel through a rope sustained by two stakes dug into the ground at a certain distance from the banks: this solution allows to easily move the sensors to follow any possible diversion of the channel (Arattano et al., 1996).

Installation of the video-camera The positive results obtained during the first years of research activities in the Moscardo Basin (between 1990 and 1994 twelve events have been recorded, Arattano et al, 1997) have suggested to continue the researches through the installation of new equipment. A video-camera has been installed close to the intermediate gaging station to shoot the channel slantwise.

This camera has been installed in 1995 (Arattano e Marchi, 1998).

Events recorded in 1995 and 1996

On July 5 1995, June 22 and July 8 1996 three debris flows occurred in the Moscardo Torrent that were recorded by the previously described seismic network. The output obtained by the recording system is plotted with the voltage as the ordinate and the time as the abscissa. The voltages can be easily converted in ground vibration velocity data, vi, through an instrumental constant. As it can be easily noticed the passage of the debris flow wave is clearly visible in all three cases, its signal being way above the level of natural noise. The signal is generally more intense at the first sensor for it is close to the check dam previously mentioned: the jump from this check dam evidently generates more intense vibrations.

As it can be easily noticed the passage of the debris flow wave is clearly visible in all three cases, its signal being way above the level of natural noise. The signal is generally more intense at the first sensor for it is close to the check dam previously mentioned: the jump from this check dam evidently generates more intense vibrations. The recording seems to have only started when the debris flow reached the location of each sensor: in fact the time lag between the inception of the recording at two consecutive detectors is of several seconds, that is consistent with the time needed by the moving mixture to travel the distance between them. However the precise value of this time lag cannot be easily established from these outputs. The events had different duration that can be easily established from the recordings: about 27 minutes for the 1995 event, 18 minutes for the June 1996 event and 15 minutes for the July 1996 event.

All the recorded signals present a peak of intensity, few tens of seconds after the inception of the recordings. This peak can be ascribed to the passage of the debris flow front, a feature of debris flows that has been mentioned earlier. As it happened for the moment of recording inception, the exact time of occurrence of this peak cannot be easily established. Further peaks appear in the seismographs behind the first peak, probably due to the presence of further surges, another characteristic feature of debris flows observed in several circumstances (Arattano et al., 1997). However it is not always easy to pick out the presence of these surges in the three recorded seismographs. A debris flow surge should produce a peak followed by a regular decrease of ground vibration behind it. Three of such surges are clearly observable for the June 22, 1996 event, but the presence of them is less evident in the July 8, 1996 seismograph and almost absent in the July 5, 1995 one. As we will see again later this is probably due to the fact that debris flows need to travel a certain distance from the inception point before clearly identifiable wave forms can develop. At the seismic site some debris flows might still have been collecting material eroded from the bed or provided by the slopes and thus have not yet developed neat wave forms. However this hypothesis needs further studies and data collection.

Measured velocities and volumes

Event Date hmax1 [m] hmax2 [m] v [m/s] Qmax [m3/s] V [m3]
17.08.90 0.69 deposition 1.0 - -
13.08.91 2.13 2.20 5.0 88 19000
30.09.91 1.67 2.40 1.90 24 3250
01.09.92 a 2.24 1.90 2.5 46 5800
01.09.92 b 2.35 1.04 10.0 - -
11.07.93 2.12 1.90 3.0 14 5600
19.07.93 0.74 0.94 0.9 3 730
20.07.93 2.30 1.52 4.3 16 6500
14.09.93 1.47 1.41 2.5 10 3800
18.07.94 2.09 1.51 4.0 - -


Table 1 - Summary of data for the events recorded from 1990 to 1994.. hmax1 is the front height recorded at the upstream station; hmax2 is the front height recorded at the downstream station; v is the mean debris flow velocity in the monitored reach; Qmax: is the maximum discharge; V total volume of the trasported material (Arattano et al. 1996).

Event Date Sensor Mean Velocity [m/s] Discharge [m3/s] Discharged Volumes [m3/s]
22.06.96 1 132 16.800
3.2
2 136 15.600
3.8
3 151 16.000
08.07.96 1 255 65.800
4.0
3 134 49.800
References
  • Arattano, M., Deganutti, A.M., Marchi, L. & Mortara G. (1996) - "Esperienze di monitoraggio delle colate detritiche nel Torrente Moscardo (Alpi Carniche)". In "Studi sui debris flow" suppl. a GEAM - Geoingegneria Ambientale e Mineraria, Anno XXXIII, n 2-3, 33-43 (in Italian).

  • Arattano, M., Deganutti, A.M. & Marchi, L. Chiarle M., Fioraso G., Mortara G., Turitto O. (1996) "Final report" U.E. contract n. EV5V-CT94-0453.

  • Arattano, M., Deganutti, A.M. & Marchi, L. (1997): Debris Flow Monitoring Activities in an Instrumented Watershed of the Italian Alps. In: Debris-flow Hazard Mitigation: Mechanics, Prediction, and Assessment (ed. by Cheng-lung Chen) (Proc. of ASCE - First International Conference, San Francisco, California, Aug. 7-9, 1997), pp. 506-515.

  • Arattano, M., & Marchi L. (1998) - "Il monitoraggio delle colate detritiche nel Torrente Moscardo (alpi Carniche): gli eventi dei mesi di giugno e luglio 1996". International Conference on the Prevention of hydrogeological hazards: the role of scientific research. Alba (Cn), 5-7 Novembre 1996, 175-185.

  • Caine, N. (1980). "The rainfall intensity-duration control of shallow landslides and debris flows". Geografiska Annaler, 62A, 23-27.

  • Coussot P., Laigle D., Arattano M., Deganutti A.M. & Marchi L. (1996) - "Determinazione delle caratteristiche reologiche di un debris flow". XXV Convegno di Idraulica e Costruzioni Idrauliche, Torino, 16-18 settembre 1996, Vol I, p. 124-135 (in Italian).

  • Coussot P., Laigle D., Arattano M., Deganutti A.M. & Marchi L. (1998) - "Direct determination of rheological characteristics of a debris flow". Journal of Hydraulic Engineering, ASCE Vol. 124, No. 8, p.865- 868.

  • Govi M., Maraga F. and Moia F. (1993): Seismic detectors for continuous bed load monitoring in a gravel stream, Journal des Sciences Hydrologiques 38, 2, 4/1993, 123-132.

  • Johnson, K.A. and Sitar, N. (1990). "Hydrologic conditions leading to debris-flow initiation". Can. Geotech. J., 27, 789-801.

  • Manfredini U. (1977): Prospezioni geofisiche, in B. Martinis (ed.), Geology of the Friuli area primarily involved in the 1976 earthquake, Rivista italiana di Paleontologia v. 83, No. 2, Milan, pp. 317-320.

  • Telford W.M., Geldart L.P., Sheriff R.E. and Keys D.A. (1976): Applied Geophysics, Cambridges University Press, Cambridge, 860 p.

  • Wieczorek, G.F. (1987). "Effect of rainfall intensity and duration on debris flows in central Santa Cruz Mountains, California". Geol. Soc. of America, Rev. in Engineering Geology, 7, 93-104.