- Email: [email protected]
Shaking of pyroclastic cones and the formation of granular flows on their flanks: Results from laboratory experiments
Journal of Volcanology and Geothermal Research 306 (2015) 83–89
Contents lists available at ScienceDirect
Journal of Volcanology and Geothermal Research journal homepage: www.elsevier.com/locate/jvolgeores
Shaking of pyroclastic cones and the formation of granular flows on their flanks: Results from laboratory experiments B. Cagnoli a,⁎, G.P. Romano b, G. Ventura c,d a
Istituto Nazionale di Geofisica e Vulcanologia, Via Donato Creti 12, 40128 Bologna, Italy Department of Mechanical and Aerospace Engineering, Università La Sapienza, Via Eudossiana 18, 00184 Rome, Italy Istituto Nazionale di Geofisica e Vulcanologia, Via di Vigna Murata 605, 00143 Rome, Italy d Istituto per l'Ambiente Marino Costiero, Consiglio Nazionale delle Ricerche, 80133 Naples, Italy b c
a r t i c l e
i n f o
Article history: Received 27 May 2015 Accepted 5 October 2015 Available online 13 October 2015 Keywords: Pyroclastic cones Volcanic tremor Earthquakes Slope instabilities Granular flows
a b s t r a c t We have carried out laboratory experiments to study the generation of granular flows on the slopes of pyroclastic cones that are experiencing volcanic tremor or tectonic earthquakes. These experiments are inspired by the occurrence of granular flows on the flanks of Mount Vesuvius during its 1944 eruption. Our laboratory model consists of sand cones built around a vibrating tube which represents a volcanic conduit with erupting magma inside. A video camera allows the study of the granular flow inception, movement and deposition. Although the collapse of the entire cone is obtained at a specific resonance frequency, single granular flows can be generated by all the vibration frequencies (1–16 Hz) and all the vibration amplitudes (0.5–1.5 mm) that our experimental apparatus has allowed us to adopt. We believe that this is due to the fact that the energy threshold to trigger the flows is small in value. Therefore, if this is true in nature as well, shaken pyroclastic cones are always potentially dangerous because they can easily generate flows that can strike the surrounding areas. © 2015 Elsevier B.V. All rights reserved.
1. Introduction Volcanoes are well known to be structures that are inherently unstable because they are built by rapid accumulation of rock material which can develop into rock avalanches and debris flows (e.g., Siebert et al., 1987; Scott et al., 2001) as well as small-scale grain flows (e.g., Sohn and Chough, 1993). The instability of volcanoes is increased by the fact that they are usually chemically altered and dissected by faults (Merle et al., 2008). In this paper, however, we show that also unaltered pyroclastic cones which are not weakened by faults or fractures can generate instabilities when shaken by volcanic tremor or tectonic earthquakes. These instabilities consist of dry granular flows (whose mechanics is therefore dominated by particle interactions of collisional and frictional nature) which travel on the surfaces of the cones. These flows are not due to volcanic dome explosions or eruptive column collapses such as the pyroclastic flows (e.g., Cas and Wright, 1988) which draw most of the attention in volcanology. Dry granular flows have been generated, for example, on the flanks of Mount Vesuvius (Italy) during the cone formation in 1944 as a result of syn-eruptive seismicity (Hazlett et al., 1991). Syn-eruptive ground shaking is caused by magma activity within the volcanic conduit and it is usually characterized by a relatively narrow band of frequencies with values smaller than 10 Hz and amplitudes that can span seven ⁎ Corresponding author. E-mail address: [email protected] (B. Cagnoli).
http://dx.doi.org/10.1016/j.jvolgeores.2015.10.003 0377-0273/© 2015 Elsevier B.V. All rights reserved.
orders of magnitude (McNutt, 1992). On Vesuvius, in 1944, the tremor is known to have increased during the lava fountain phase of the eruption with the increase of the gas discharge rate (Pappalardo et al., 2014). Usually, as the violence of the eruption increases, the amplitude of the vibrations becomes larger (McNutt, 1992). The Vesuvius cone is ~ 300 m high and it is located at the top of the much taller Mount Somma, Italy (Santacroce and Sbrana, 2003). Fig. 1 shows that virtually all flanks were affected by granular flows. These flows are not deep sector collapses because they travelled on the cone surface without dissecting the internal structures of the cone. They have left long fingerlike deposits at the base of Vesuvius and erosive scarps at the top (Fig. 1). Hazlett et al. (1991) describe the two types of flows on Vesuvius that are shown in Fig. 1: they are ash flows which consist of ash with comminuted scoria and block-and-ash flows with up to ~ 20% of blocks dispersed in ash. By definition, ash is b2 mm in grain size and blocks are N64 mm in grain size. In Fig. 1, the longest flow deposit is 1.3 km in length, but most are less than half this long. Lobe widths range from a few tens of metres to several hundred metres. The thickness is larger at the front of the deposits with typical values between 20 and 30 m. Levees are present. Fig. 2 shows the front of a flow deposit where block-size rock fragments are clearly visible. Similar granular flows are not a rare occurrence in nature. For instance, a flank failure originated a hot granular flow (resembling a pyroclastic flow) at the top of Mount Etna (Italy) on February 11, 2014 (www.ct. ingv.it). This phenomenon took place during a Strombolian activity with
84
B. Cagnoli et al. / Journal of Volcanology and Geothermal Research 306 (2015) 83–89
Fig. 1. Granular flows on Mount Vesuvius: (a) after Hazlett et al. (1991) and (b) as shown by a digital elevation model. The dashed ellipse highlights the hourglass shape of the combined outlines of an erosive scarp and the associated flow deposit.
ongoing volcanic tremor and it was due to the instability of accumulated scoria fragments. In 1930, also Stromboli volcano (Aeolian Islands, Italy) generated hot density currents that were the consequence of syneruptive failures of rapidly accumulated pyroclasts (Di Roberto et al., 2014). This example is noteworthy for future hazard assessments on Stromboli because the 1930 flows travelled within valleys distinct from the Sciara del Fuoco collapse scar that is the usual uninhabited path (e.g., Tommasi et al., 2008) for landslides and lava flows on the island. Flows on an island can also trigger tsunamis when they enter the sea. In this study we explore experimentally the effect of vibrations on the generation of superficial granular flows on the flanks of laboratory sand cones that model natural pyroclastic cones. We are not aware of other experiments in the literature which focus on the geological meaning of similar granular flows. For example, Tibaldi (1995), in his laboratory experiments, was interested in pyroclastic cone formation and the interaction of the cones with faults. Merle and Borgia (1996) carried out experiments with sand cones where their instability is induced by the deformation of the substratum. Acocella (2005) studied deep sector collapses of laboratory sand cones due to faults, erosion, rapid accumulation of material and volcanic intrusions. The stability of vibrated granular slopes in the laboratory has been investigated, for example,
Fig. 2. Front of the flow deposit highlighted by the ellipse in Fig. 1. Block-size fragments can be seen protruding from the deposit.
by Katz and Aharonov (2006), but the geometry of their slopes (planar) and that of their instabilities (deep slope deformations) are significantly different from those in our system. Our laboratory model consists of sand cones built around a vibrating tube which is located along the cone vertical axis. The sand cone represents a pyroclastic cone whereas the vibrating tube represents the volcanic conduit with pulsating magma inside. Here, we explore the inception, movement and deposition mechanisms of granular flows that travel on the surfaces of granular cones and that are generated by shaking. These flows of granular material, thus, do not travel within pre-existent channels. 2. Method 2.1. Experimental apparatus The experimental apparatus consists of a sand cone placed on a large table top (Fig. 3). The cone is 20 cm tall and it is built around an aluminium tube (3 cm in diameter) located along its axis. The tube is rigidly connected to the top surface of a vibrating table that is positioned underneath the large table. The tube does not touch the large table because its diameter is smaller than that of the hole through which the tube protrudes (Fig. 3). The vibrating table is rigidly joined to a
Fig. 3. Diagram of the experimental apparatus and position of the video camera.
B. Cagnoli et al. / Journal of Volcanology and Geothermal Research 306 (2015) 83–89 Table 1 Grain size analysis of the sand. Grain size mm
Weight %
1.000–0.500 0.500–0.250 0.250–0.125 0.125–0.063 b0.063
4.47 65.18 27.11 2.95 0.29
small table that is massive enough to prevent the disturbing effects of spurious vibrations. The tube is wrapped in sand paper to improve the coupling of tube and cone (the sands of sand paper and cone have similar grain size). Table 1 shows the results of the grain size analysis of the sand used in the experiments. The mean grain size is 332 μm. The sand has been dried in an oven before the experiments. Frequency and amplitude of the vibrations are computer-controlled. Our system allows frequencies in steps of 1 Hz from 1 to 16 Hz when the amplitude is 0.5 mm, from 1 to 12 Hz when the amplitude is 1 mm and from 1 to 8 Hz when the amplitude is 1.5 mm. These vibrations occur in one horizontal direction only. Therefore, here we focus on the effects of the horizontal component of a movement which, in nature, can have other components as well. The top of the vibrating table is as close as possible to the lower surface of the large table. This affects the amplitude of the vibrations experienced by the sand because these amplitudes are expected to increase upward along the tube. Each experiment with the same combination of vibration frequency and vibration amplitude has been repeated three times. Accelerometers have confirmed that the nominal and the real values of frequencies and amplitudes are very similar. After each experimental run, the sand cone has been rebuilt manually around the tube. 2.2. Particle Image Velocimetry analysis A digital video camera located above the vibrating tube has filmed the sand cones during the experiments (Fig. 3). The camera has a frame rate equal to 50 fps. The resolution of the images is 1920 × 1080 pixels. The movies have been analysed by means of Particle Image Velocimetry (PIV) technique, which is an optical method to obtain instantaneous velocity fields from moving tracers (Raffel et al., 2007). PIV analysis is routinely used to study granular flows (e.g., Pudasaini and Hutter, 2006; Cagnoli and Romano, 2012b). This technique allows here the measurement of the instantaneous speed of the sand grains on the surfaces of the cones. We have performed our PIV analysis with interrogation windows equal to 16 × 16 pixels, three iteration steps and 50% overlapping. The PIV algorithm uses advanced image processing technique such as window-offset, sub-pixel accuracy and image deformation (Di Florio et al., 2002). 2.3. Scaling We use, for our 20 cm tall cone, an axial conduit (i.e., a tube) that is 3 cm in diameter. This provides a correct geometric scaling, because the ratio of these dimensions is similar to that of the Vesuvius cone whose height is ~ 300 m and whose conduit diameter is ~ 50 m. The angle of repose of the sand in the laboratory (33°) is also similar in value to
Fig. 4. Two sand cones whose video camera images have been analysed by means of the PIV technique: (a) the cone shown in the left column of frames experiences no resonance frequency whereas (b) the cone shown in the right column does experience a resonance frequency. The time, which increases downward, is from the application of the vibrations. The white dash-dot curves indicate the position of the base of the cones as seen from above. The base of the cone on the right loses its circular shape after the application of the resonance frequency. The colour code refers to the speed of the sand on the surface of the cones. The values in millimetres are the vibration amplitude and the values in hertz are the vibration frequencies.
85
86
B. Cagnoli et al. / Journal of Volcanology and Geothermal Research 306 (2015) 83–89
Fig. 5. A sand cone (a) before and (b) after the application of a resonance frequency. The vibration frequency is equal to 10 Hz and the vibration amplitude is equal to 1 mm. The base of the cone loses its circular shape after the application of the resonance frequency. The arrow indicates an example of a scarp that has migrated backward.
the slope inclination of Vesuvius. Since the ratio N among natural and laboratory geometric dimensions is approximately 1500, 100 μm sand grains in the experiments represent clasts in nature that are 15 cm in diameter. Fig. 2 shows that some of the Vesuvius flow deposits contain also block-size fragments. Concerning the dynamic similarity, we take here into consideration the nondimensional parameter a/A f 2, where a is the acceleration, A is the amplitude and f is the frequency of the oscillations (e.g., Katz and Aharonov, 2006; Anastasopoulos et al., 2010). Assuming that the amplitude of the vibrations scales as the other geometric dimensions, it follows that: 1) the frequencies of the vibrations in the model and in nature are similar in value, if the accelerations are proportional to the amplitude of the oscillations (i.e., if they scale as N); 2) the frequencies in the model represent very small frequencies in nature (around N−1/2 ≈ 40 times smaller than those in the experiments), if the accelerations are the same in the model and in nature. In nature, the frequency of volcanic tremor is expected to have values below 10 Hz (including values smaller
Fig. 7. Speed measurements versus time obtained by the PIV technique in six fixed positions along the flow path (the inset shows these positions on the cone surface). The value in millimetres is the vibration amplitude and the value in hertz is the vibration frequency. This is not a resonance frequency. The arrows indicate that there is an early movement of the sand in positions 1 and 2 (blue and red curves) before the actual flow release in these same positions.
than 1 Hz), whereas the amplitudes can span seven orders of magnitudes (McNutt, 1992). 3. Results Single granular flows can be generated by all the vibration frequencies and all the vibration amplitudes that our apparatus has allowed us to adopt (even when there is no overall collapse). With some frequencies, it takes much longer for a flow to be generated, but, sooner or
Fig. 6. Two granular flows and their deposits generated in two experiments with non-resonance frequencies (i.e., without the collapse of the entire sand cones). The values in millimetres are the vibration amplitudes and the values in hertz are the vibration frequencies. The combined outlines of scar and deposit form an hourglass profile.
B. Cagnoli et al. / Journal of Volcanology and Geothermal Research 306 (2015) 83–89
87
Fig. 8. Simultaneous speed measurements obtained by means of the PIV technique versus distance along the flow path. This plot shows that particle speed decreases gradually toward the rear along the longitudinal axis of the flow so that the flow front is faster than the rear. The value in millimetres is the vibration amplitude and the value in hertz is the vibration frequency. This is not a resonance frequency.
later, at least a flow is released at all frequency values. These flows travel on the conical surfaces and they are relatively thin (less than 1 cm thick) when compared to the size of the 20 cm tall cones. However, the complete collapse of the entire cone is obtained only at a critical value of frequency. For example, Fig. 4 shows a PIV analysis of two experiments with 1 mm amplitude vibrations and a frequency equal to 7 Hz and 10 Hz, respectively. In this case, the critical frequency (i.e., the resonance frequency) is 10 Hz because this is the frequency that causes the entire cone collapse. By contrast, with 7 Hz, only a single flow is generated at ~24 s from the start of shaking. The granular flows are generated in unpredictable places (not necessarily in the direction of the vibrations) and at unpredictable times (even if most of them, but not all of them, occur not long after the application of the vibrations). This unpredictability (in terms of size and shape of the flows as well) occurs also in different experimental runs with the same vibration frequency and the same vibration amplitude. This is so because, even if the sand cones have been rebuilt after each test with the same identical size, shape and position, some
Fig. 10. Velocity field obtained by PIV analysis of a flow showing that the erosive scarp (on the left) migrates upward during the downflow movement of the flow front (on the right). The values in seconds are the time distance from the first frame. The value in millimetres is the vibration amplitude and the value in hertz is the vibration frequency. This is not a resonance frequency.
Fig. 9. Speed and acceleration of the front of a flow versus time as measured by the PIV technique from flow release to final deposition. The value in millimetres is the vibration amplitude and the value in hertz is the vibration frequency. This is not a resonance frequency.
differences are unavoidable. For example, each cone has its own different and unique spatial distribution of the sand particles. In the case of the critical frequency, all the slopes of the sand cone are simultaneously set in motion (as visible, for example, in the 3 s frame of Fig. 4b) so that the entire cone collapses and its final basal shape is not anymore circular as it was before the application of the vibrations (Figs. 4b and 5). The collapsed cones acquire a final smaller slope inclination which is not affected by further vibrations (as visible, for
88
B. Cagnoli et al. / Journal of Volcanology and Geothermal Research 306 (2015) 83–89
example, in the 24 s frame of Fig. 4b, where no more surface activity is detected). During the cone collapse, flows are generated with the erosive scarp behind them which migrates backward (an example of such a scarp is indicated by the arrow in Fig. 5b). In this case, the flows can be larger than those due to non-critical frequencies. With vibration amplitudes equal to 0.5 and 1 mm, the critical frequency is ~ 10 Hz. Unfortunately, when the amplitude of the vibrations is equal to 1.5 mm, our apparatus does not allow frequencies larger than 8 Hz, so that the resonance frequency is not obtained. Non-critical frequencies generate single flows without the collapse of the entire cone. The movies show that these granular flows are originated in most of the cases at the top of the slope (Fig. 6) where the sand starts moving before the flow proper is released. These early movements are pointed out by arrows in Fig. 7 which illustrates speed measurements versus time in fixed spots along the flow path as obtained by the PIV technique. In Fig. 7, the flow is released in positions 1 and 2 (red and blue curves) where the early movements have occurred. After the flow generation, the arrival of the flow front in spots 3 to 6 is marked by a sudden increase of speed (Fig. 7). A noteworthy characteristic of these flows is that their front is faster than their rear and that the speed of the sand particles decreases gradually toward the rear along the longitudinal axis of the flows. This is shown in Fig. 8, which illustrates simultaneous speed measurements obtained by means of the PIV technique in different positions along the flow path. The dynamics of the front of the flows presents a complicated behaviour. Fig. 9, for example, shows that the speed of the flow front first increases to reach a maximum value that stays constant for a while, then it decreases and increases again to reach a second peak and finally the speed decreases and the flow stops. During the downward motion of the flow, the sand along the scarp rim keeps moving into the depression left behind by the flow since the scarp has too steep and unstable surfaces. This results into a lateral spreading and a backward migration of the scarp in a direction opposite with respect to that of the flow front (Fig. 10). The transversal width of the scarp rim can thus become larger at the top than more downslope. This, with the relatively large width of the final deposit at the base of the cone, generates an hourglass outline of the combined profiles of scarp and deposit (Fig. 6). The flows stop when their front reaches the horizontal table top because of the abrupt change of slope. During their formation, the final deposits accrete rearward because the granular material accumulates at their back (Fig. 6). However there are also instances when the flow does not reach the base but it stops before. 4. Discussion Some of the pyroclasts deposited during the 1944 eruption on the Vesuvius cone were loose (i.e., unwelded) as suggested by the hourglass outline of the combined profiles of the scarp and deposit of the flow within the ellipse in Fig. 1b (as in the dry sand flow case of Hungr et al., 2014). This outline has been obtained in our laboratory cones (Fig. 6), but also, for example, in laboratory sand dunes (Sutton et al., 2013), where, in both cases, the sand is non-cohesive. Hourglass profiles are compatible with loose granular material where the avalanches are fed through scarp recession. Scarp recession (visible, for example, in Fig. 10) is a mechanism that is likely to have occurred on the Vesuvius cone as well. Scarp recession is known also in nonvolcanic settings (e.g., Ventura et al., 2011). A resonance frequency which causes the instantaneous collapse of the entire cone can be important in nature too (McNutt, 1992). We wonder whether the Vesuvius cone during its 1944 eruption has been shaken by its resonance frequency. It is true that there are flows on all the flanks of Vesuvius (Fig. 1), but they may simply be the summation of separate flows generated over a relatively long period of time by frequencies different from the critical one. In case a comparison with natural values is attempted, it is important to realise that the critical frequency obtained in the laboratory is the resonance frequency of not
just the sand cone alone, but that of the cone, tables and tube together (the tube transmits the vibrations to the sand cone and the sand cone transmits the vibrations to the large table). In any case, the noteworthy feature is that single granular flows can be generated by all the vibration frequencies and all the vibration amplitudes adopted in our experiments (i.e., a relatively large range of values). The occurrence of flows at all frequencies and amplitudes can be due to the fact that the energy threshold to trigger this type of instabilities is relatively small in value. This point is important because, if this is true in nature as well, a pyroclastic cone is always potentially dangerous when shaken. In our laboratory experiments, the compaction of the sand decreases toward the cone surface because this surface is manually reworked after each experimental run to rebuild the slope, whereas the inner part of the cone is left untouched. This decrease is expected also in natural volcanic cones as a consequence of the rapid accumulation of loose pyroclasts during an eruption and the natural increase of compaction with depth. The uncompacted superficial granular material is more likely to fail when the ground is shaken by volcanic tremor or tectonic earthquakes. The gradual decrease of the sand speed toward the rear along the flow axis (Fig. 8) is due to the fact that, since all portions of a flow are affected by a downslope acceleration, those that have left earlier (the frontal ones) have experienced the acceleration for a longer time and they have acquired, at the same moment in time, a speed larger than that of those that have left later (the rear ones). Thus, a trend such as that in Fig. 8 suggests that a flow is fed through scarp recession where the granular material is released gradually. On the other hand, multiple accelerations and decelerations of the flow fronts during the flow descent (Fig. 9) can be due to slope irregularities whose small local departures from the angle of repose affect the flows because they are relatively thin (b1 cm thick). Therefore, interestingly, the inclination of the cone slope can differ locally (albeit by small amounts) from the nominal value of the angle of repose. These departures can be the result of grain heterogeneities, differences of granular compaction as well as the manual reworking of the sand cones at the end of each experimental run. Furthermore, the change of slope at the base of the sand cone is more abrupt than on Vesuvius, so that the laboratory flows are unable to propagate behind the base of the cone and the flow forms a deposit as soon as its front reaches the horizontal table (Fig. 6). This does not happen on Vesuvius where the flows can travel further than the base of its cone on the flanks of Mount Somma that are not horizontal (as visible in the southern sector of Fig. 1). Therefore, the mobility of these flows depends also on the slope inclination of the ground surrounding the cone as well as on their grain size and on their volume (Cagnoli and Romano, 2012a).
5. Conclusions We conclude that pyroclastic cones can be hazardous when experiencing ground shaking because of the generation of granular flows on their slopes. This is true during an eruption because of volcanic tremor, but it can be true also after the volcanic activity has ceased as a consequence of distant tectonic earthquakes. Ground shaking can make a cone unstable even if its flanks have an inclination equal to the angle of repose of the granular material (as observed on Vesuvius by Hazlett et al., 1991, and in our laboratory experiments by us), but its instability certainly increases as the volume of rapidly accumulated material that exceeds the angle of repose becomes larger. Importantly, our experiments suggest that there is a wide range of frequencies (and not just a single value) that can generate granular flows. We believe that this is due to the fact that the energy threshold that triggers these instabilities is relatively small in value and, if this is true in nature as well, pyroclastic cones are always potentially dangerous when shaken.
B. Cagnoli et al. / Journal of Volcanology and Geothermal Research 306 (2015) 83–89
Acknowledgements We thank Andreas Lardon for the laboratory data collection at the Università La Sapienza in Rome where he worked for his DiplomIngenieur thesis (Technische Universität, Dresden). We are also grateful to Catello Acerra and Thorossian William from the INGV in Rome. Catello Acerra has designed and built the vibrating table with its mechanical parts, electronics and controlling software. Thorossian William has designed and manufactured the circuit board. We thank also Domenico Pietrogiacomi and Lucilla Monteleone for their assistance in the laboratory and we thank the reviewers for their comments. References Acocella, V., 2005. Modes of sector collapse of volcanic cones: insights from analogue experiments. J. Geophys. Res. 110, B02205. http://dx.doi.org/10.1029/2004JB003166. Anastasopoulos, I., Georgarakos, T., Georgiannou, V., Drosos, V., Kourkoulis, R., 2010. Seismic performance of bar-mat reinforced-soil retaining wall: shaking table testing versus numerical analysis with modified kinematic hardening constitutive model. Soil Dyn. Earthq. Eng. 30, 1089–1105. Cagnoli, B., Romano, G.P., 2012a. Effects of flow volume and grain size on mobility of dry granular flows of angular rock fragments: a functional relationship of scaling parameters. J. Geophys. Res. 117, B02207. http://dx.doi.org/10.1029/2011JB008926. Cagnoli, B., Romano, G.P., 2012b. Granular pressure at the base of dry flows of angular rock fragments as a function of grain size and flow volume: a relationship from laboratory experiments. J. Geophys. Res. 117, B10202. http://dx.doi.org/10.1029/ 2012JB009374. Cas, R.A.F., Wright, J.V., 1988. Volcanic Successions. Allen and Unwin, London. Di Florio, D., Di Felice, F., Romano, G.P., 2002. Windowing, re-shaping and re-orientation interrogation windows in particle image velocimetry for the investigation of shear flows. Meas. Sci. Technol. 13 (7), 953–962. Di Roberto, A., Bertagnini, A., Pompilio, M., Bisson, M., 2014. Pyroclastic density currents at Stromboli volcano (Aeolian Islands, Italy): a case study of the 1930 eruption. Bull. Volcanol. 76, 827. Hazlett, R.W., Buesch, D., Anderson, J.L., Elan, R., Scandone, R., 1991. Geology, failure conditions, and implications of seismogenic avalanches of the 1944 eruption at Vesuvius, Italy. J. Volcanol. Geotherm. Res. 47, 249–264.
89
Hungr, O., Leroueil, S., Picarelli, L., 2014. The Varnes classification of landslide types, un update. Landslides 11, 167–194. Katz, O., Aharonov, E., 2006. Landslides in vibrating sand box: what controls types of slope failure and frequency magnitude relations? Earth Planet. Sci. Lett. 247, 280–294. McNutt, S.R., 1992. Volcanic Tremor. Encyclopedia of Earth System Science. Academic Press, pp. 417–425. Merle, O., Borgia, A., 1996. Scaled experiments of volcanic spreading. J. Geophys. Res. 101 (B6), 13805–13817. Merle, O., Michon, L., Bachèlery, P., 2008. Caldera rim collapse: a hidden volcanic hazard. J. Volcanol. Geotherm. Res. 177, 525–530. Pappalardo, L., D'Auria, L., Cavallo, A., Fiore, S., 2014. Petrological and seismic precursors of the paroxismal phase of the last Vesuvius eruption on March 1944. Sci. Rep. 4, 6297. http://dx.doi.org/10.1038/srep06297. Pudasaini, S.P., Hutter, K., 2006. Avalanche Dynamics. Springer, Berlin. Raffel, M., Willert, C.E., Wereley, S.T., Kompenhans, J., 2007. Particle Image Velocimetry. Springer, Berlin. Santacroce, R, Sbrana, A. (Eds.), 2003. Geological Map of Vesuvius. S.EL.CA, Firenze. Scott, K.M., Macías, J.L., Naranjo, J.A., Rodrigues, S., McGeehin, J.P., 2001. Catastrophic debris flows transformed from landslides in volcanic terrains: mobility, hazard assessment and mitigation strategies. US Geological Survey Professional Paper 1630 (59 pp.). Siebert, L., Glicken, H., Ui, T., 1987. Volcanic hazards from Bezymianny- and Bandai-type eruptions. Bull. Volcanol. 49, 435–459. Sohn, Y.K., Chough, S.K., 1993. The Udo tuff cone, Cheju Island, South Korea: transformation of pyroclastic fall into debris fall and grain flow on a steep volcanic cone slope. Sedimentology 40, 769–786. Sutton, S.L.F., McKenna Neuman, C., Nickling, W., 2013. Avalanche grainflow on a simulated aeolian dune. J. Geophys. Res. Earth Surf. 118, 1–10. http://dx.doi.org/10.1002/jgrf. 20130. Tibaldi, A., 1995. Morphology of pyroclastic cones and tectonics. J. Geophys. Res. 100, 24521–24535. Tommasi, P., Baldi, P., Chiocci, F.L., Coltelli, M., Marsella, M., Romagnoli, C., 2008. Slope failures induced by the December 2002 eruption at Stromboli Volcano. In: Calvari, S., Inguaggiato, S., Puglisi, G., Ripepe, M., Rosi, M. (Eds.), The Stromboli Volcano: An integrated Study of the 2002–2003 Eruption. American Geophysical Union, Washington, D.C., pp. 129–145. Ventura, G., Vilardo, G., Terranova, C., Sessa, E.B., 2011. Tracking and evolution of complex active landslides by multi-temporal airborne LiDAR data: the Montaguto landslide (Southern Italy). Remote Sens. Environ. 115, 3237–3248.
Contents lists available at ScienceDirect
Journal of Volcanology and Geothermal Research journal homepage: www.elsevier.com/locate/jvolgeores
Shaking of pyroclastic cones and the formation of granular flows on their flanks: Results from laboratory experiments B. Cagnoli a,⁎, G.P. Romano b, G. Ventura c,d a
Istituto Nazionale di Geofisica e Vulcanologia, Via Donato Creti 12, 40128 Bologna, Italy Department of Mechanical and Aerospace Engineering, Università La Sapienza, Via Eudossiana 18, 00184 Rome, Italy Istituto Nazionale di Geofisica e Vulcanologia, Via di Vigna Murata 605, 00143 Rome, Italy d Istituto per l'Ambiente Marino Costiero, Consiglio Nazionale delle Ricerche, 80133 Naples, Italy b c
a r t i c l e
i n f o
Article history: Received 27 May 2015 Accepted 5 October 2015 Available online 13 October 2015 Keywords: Pyroclastic cones Volcanic tremor Earthquakes Slope instabilities Granular flows
a b s t r a c t We have carried out laboratory experiments to study the generation of granular flows on the slopes of pyroclastic cones that are experiencing volcanic tremor or tectonic earthquakes. These experiments are inspired by the occurrence of granular flows on the flanks of Mount Vesuvius during its 1944 eruption. Our laboratory model consists of sand cones built around a vibrating tube which represents a volcanic conduit with erupting magma inside. A video camera allows the study of the granular flow inception, movement and deposition. Although the collapse of the entire cone is obtained at a specific resonance frequency, single granular flows can be generated by all the vibration frequencies (1–16 Hz) and all the vibration amplitudes (0.5–1.5 mm) that our experimental apparatus has allowed us to adopt. We believe that this is due to the fact that the energy threshold to trigger the flows is small in value. Therefore, if this is true in nature as well, shaken pyroclastic cones are always potentially dangerous because they can easily generate flows that can strike the surrounding areas. © 2015 Elsevier B.V. All rights reserved.
1. Introduction Volcanoes are well known to be structures that are inherently unstable because they are built by rapid accumulation of rock material which can develop into rock avalanches and debris flows (e.g., Siebert et al., 1987; Scott et al., 2001) as well as small-scale grain flows (e.g., Sohn and Chough, 1993). The instability of volcanoes is increased by the fact that they are usually chemically altered and dissected by faults (Merle et al., 2008). In this paper, however, we show that also unaltered pyroclastic cones which are not weakened by faults or fractures can generate instabilities when shaken by volcanic tremor or tectonic earthquakes. These instabilities consist of dry granular flows (whose mechanics is therefore dominated by particle interactions of collisional and frictional nature) which travel on the surfaces of the cones. These flows are not due to volcanic dome explosions or eruptive column collapses such as the pyroclastic flows (e.g., Cas and Wright, 1988) which draw most of the attention in volcanology. Dry granular flows have been generated, for example, on the flanks of Mount Vesuvius (Italy) during the cone formation in 1944 as a result of syn-eruptive seismicity (Hazlett et al., 1991). Syn-eruptive ground shaking is caused by magma activity within the volcanic conduit and it is usually characterized by a relatively narrow band of frequencies with values smaller than 10 Hz and amplitudes that can span seven ⁎ Corresponding author. E-mail address: [email protected] (B. Cagnoli).
http://dx.doi.org/10.1016/j.jvolgeores.2015.10.003 0377-0273/© 2015 Elsevier B.V. All rights reserved.
orders of magnitude (McNutt, 1992). On Vesuvius, in 1944, the tremor is known to have increased during the lava fountain phase of the eruption with the increase of the gas discharge rate (Pappalardo et al., 2014). Usually, as the violence of the eruption increases, the amplitude of the vibrations becomes larger (McNutt, 1992). The Vesuvius cone is ~ 300 m high and it is located at the top of the much taller Mount Somma, Italy (Santacroce and Sbrana, 2003). Fig. 1 shows that virtually all flanks were affected by granular flows. These flows are not deep sector collapses because they travelled on the cone surface without dissecting the internal structures of the cone. They have left long fingerlike deposits at the base of Vesuvius and erosive scarps at the top (Fig. 1). Hazlett et al. (1991) describe the two types of flows on Vesuvius that are shown in Fig. 1: they are ash flows which consist of ash with comminuted scoria and block-and-ash flows with up to ~ 20% of blocks dispersed in ash. By definition, ash is b2 mm in grain size and blocks are N64 mm in grain size. In Fig. 1, the longest flow deposit is 1.3 km in length, but most are less than half this long. Lobe widths range from a few tens of metres to several hundred metres. The thickness is larger at the front of the deposits with typical values between 20 and 30 m. Levees are present. Fig. 2 shows the front of a flow deposit where block-size rock fragments are clearly visible. Similar granular flows are not a rare occurrence in nature. For instance, a flank failure originated a hot granular flow (resembling a pyroclastic flow) at the top of Mount Etna (Italy) on February 11, 2014 (www.ct. ingv.it). This phenomenon took place during a Strombolian activity with
84
B. Cagnoli et al. / Journal of Volcanology and Geothermal Research 306 (2015) 83–89
Fig. 1. Granular flows on Mount Vesuvius: (a) after Hazlett et al. (1991) and (b) as shown by a digital elevation model. The dashed ellipse highlights the hourglass shape of the combined outlines of an erosive scarp and the associated flow deposit.
ongoing volcanic tremor and it was due to the instability of accumulated scoria fragments. In 1930, also Stromboli volcano (Aeolian Islands, Italy) generated hot density currents that were the consequence of syneruptive failures of rapidly accumulated pyroclasts (Di Roberto et al., 2014). This example is noteworthy for future hazard assessments on Stromboli because the 1930 flows travelled within valleys distinct from the Sciara del Fuoco collapse scar that is the usual uninhabited path (e.g., Tommasi et al., 2008) for landslides and lava flows on the island. Flows on an island can also trigger tsunamis when they enter the sea. In this study we explore experimentally the effect of vibrations on the generation of superficial granular flows on the flanks of laboratory sand cones that model natural pyroclastic cones. We are not aware of other experiments in the literature which focus on the geological meaning of similar granular flows. For example, Tibaldi (1995), in his laboratory experiments, was interested in pyroclastic cone formation and the interaction of the cones with faults. Merle and Borgia (1996) carried out experiments with sand cones where their instability is induced by the deformation of the substratum. Acocella (2005) studied deep sector collapses of laboratory sand cones due to faults, erosion, rapid accumulation of material and volcanic intrusions. The stability of vibrated granular slopes in the laboratory has been investigated, for example,
Fig. 2. Front of the flow deposit highlighted by the ellipse in Fig. 1. Block-size fragments can be seen protruding from the deposit.
by Katz and Aharonov (2006), but the geometry of their slopes (planar) and that of their instabilities (deep slope deformations) are significantly different from those in our system. Our laboratory model consists of sand cones built around a vibrating tube which is located along the cone vertical axis. The sand cone represents a pyroclastic cone whereas the vibrating tube represents the volcanic conduit with pulsating magma inside. Here, we explore the inception, movement and deposition mechanisms of granular flows that travel on the surfaces of granular cones and that are generated by shaking. These flows of granular material, thus, do not travel within pre-existent channels. 2. Method 2.1. Experimental apparatus The experimental apparatus consists of a sand cone placed on a large table top (Fig. 3). The cone is 20 cm tall and it is built around an aluminium tube (3 cm in diameter) located along its axis. The tube is rigidly connected to the top surface of a vibrating table that is positioned underneath the large table. The tube does not touch the large table because its diameter is smaller than that of the hole through which the tube protrudes (Fig. 3). The vibrating table is rigidly joined to a
Fig. 3. Diagram of the experimental apparatus and position of the video camera.
B. Cagnoli et al. / Journal of Volcanology and Geothermal Research 306 (2015) 83–89 Table 1 Grain size analysis of the sand. Grain size mm
Weight %
1.000–0.500 0.500–0.250 0.250–0.125 0.125–0.063 b0.063
4.47 65.18 27.11 2.95 0.29
small table that is massive enough to prevent the disturbing effects of spurious vibrations. The tube is wrapped in sand paper to improve the coupling of tube and cone (the sands of sand paper and cone have similar grain size). Table 1 shows the results of the grain size analysis of the sand used in the experiments. The mean grain size is 332 μm. The sand has been dried in an oven before the experiments. Frequency and amplitude of the vibrations are computer-controlled. Our system allows frequencies in steps of 1 Hz from 1 to 16 Hz when the amplitude is 0.5 mm, from 1 to 12 Hz when the amplitude is 1 mm and from 1 to 8 Hz when the amplitude is 1.5 mm. These vibrations occur in one horizontal direction only. Therefore, here we focus on the effects of the horizontal component of a movement which, in nature, can have other components as well. The top of the vibrating table is as close as possible to the lower surface of the large table. This affects the amplitude of the vibrations experienced by the sand because these amplitudes are expected to increase upward along the tube. Each experiment with the same combination of vibration frequency and vibration amplitude has been repeated three times. Accelerometers have confirmed that the nominal and the real values of frequencies and amplitudes are very similar. After each experimental run, the sand cone has been rebuilt manually around the tube. 2.2. Particle Image Velocimetry analysis A digital video camera located above the vibrating tube has filmed the sand cones during the experiments (Fig. 3). The camera has a frame rate equal to 50 fps. The resolution of the images is 1920 × 1080 pixels. The movies have been analysed by means of Particle Image Velocimetry (PIV) technique, which is an optical method to obtain instantaneous velocity fields from moving tracers (Raffel et al., 2007). PIV analysis is routinely used to study granular flows (e.g., Pudasaini and Hutter, 2006; Cagnoli and Romano, 2012b). This technique allows here the measurement of the instantaneous speed of the sand grains on the surfaces of the cones. We have performed our PIV analysis with interrogation windows equal to 16 × 16 pixels, three iteration steps and 50% overlapping. The PIV algorithm uses advanced image processing technique such as window-offset, sub-pixel accuracy and image deformation (Di Florio et al., 2002). 2.3. Scaling We use, for our 20 cm tall cone, an axial conduit (i.e., a tube) that is 3 cm in diameter. This provides a correct geometric scaling, because the ratio of these dimensions is similar to that of the Vesuvius cone whose height is ~ 300 m and whose conduit diameter is ~ 50 m. The angle of repose of the sand in the laboratory (33°) is also similar in value to
Fig. 4. Two sand cones whose video camera images have been analysed by means of the PIV technique: (a) the cone shown in the left column of frames experiences no resonance frequency whereas (b) the cone shown in the right column does experience a resonance frequency. The time, which increases downward, is from the application of the vibrations. The white dash-dot curves indicate the position of the base of the cones as seen from above. The base of the cone on the right loses its circular shape after the application of the resonance frequency. The colour code refers to the speed of the sand on the surface of the cones. The values in millimetres are the vibration amplitude and the values in hertz are the vibration frequencies.
85
86
B. Cagnoli et al. / Journal of Volcanology and Geothermal Research 306 (2015) 83–89
Fig. 5. A sand cone (a) before and (b) after the application of a resonance frequency. The vibration frequency is equal to 10 Hz and the vibration amplitude is equal to 1 mm. The base of the cone loses its circular shape after the application of the resonance frequency. The arrow indicates an example of a scarp that has migrated backward.
the slope inclination of Vesuvius. Since the ratio N among natural and laboratory geometric dimensions is approximately 1500, 100 μm sand grains in the experiments represent clasts in nature that are 15 cm in diameter. Fig. 2 shows that some of the Vesuvius flow deposits contain also block-size fragments. Concerning the dynamic similarity, we take here into consideration the nondimensional parameter a/A f 2, where a is the acceleration, A is the amplitude and f is the frequency of the oscillations (e.g., Katz and Aharonov, 2006; Anastasopoulos et al., 2010). Assuming that the amplitude of the vibrations scales as the other geometric dimensions, it follows that: 1) the frequencies of the vibrations in the model and in nature are similar in value, if the accelerations are proportional to the amplitude of the oscillations (i.e., if they scale as N); 2) the frequencies in the model represent very small frequencies in nature (around N−1/2 ≈ 40 times smaller than those in the experiments), if the accelerations are the same in the model and in nature. In nature, the frequency of volcanic tremor is expected to have values below 10 Hz (including values smaller
Fig. 7. Speed measurements versus time obtained by the PIV technique in six fixed positions along the flow path (the inset shows these positions on the cone surface). The value in millimetres is the vibration amplitude and the value in hertz is the vibration frequency. This is not a resonance frequency. The arrows indicate that there is an early movement of the sand in positions 1 and 2 (blue and red curves) before the actual flow release in these same positions.
than 1 Hz), whereas the amplitudes can span seven orders of magnitudes (McNutt, 1992). 3. Results Single granular flows can be generated by all the vibration frequencies and all the vibration amplitudes that our apparatus has allowed us to adopt (even when there is no overall collapse). With some frequencies, it takes much longer for a flow to be generated, but, sooner or
Fig. 6. Two granular flows and their deposits generated in two experiments with non-resonance frequencies (i.e., without the collapse of the entire sand cones). The values in millimetres are the vibration amplitudes and the values in hertz are the vibration frequencies. The combined outlines of scar and deposit form an hourglass profile.
B. Cagnoli et al. / Journal of Volcanology and Geothermal Research 306 (2015) 83–89
87
Fig. 8. Simultaneous speed measurements obtained by means of the PIV technique versus distance along the flow path. This plot shows that particle speed decreases gradually toward the rear along the longitudinal axis of the flow so that the flow front is faster than the rear. The value in millimetres is the vibration amplitude and the value in hertz is the vibration frequency. This is not a resonance frequency.
later, at least a flow is released at all frequency values. These flows travel on the conical surfaces and they are relatively thin (less than 1 cm thick) when compared to the size of the 20 cm tall cones. However, the complete collapse of the entire cone is obtained only at a critical value of frequency. For example, Fig. 4 shows a PIV analysis of two experiments with 1 mm amplitude vibrations and a frequency equal to 7 Hz and 10 Hz, respectively. In this case, the critical frequency (i.e., the resonance frequency) is 10 Hz because this is the frequency that causes the entire cone collapse. By contrast, with 7 Hz, only a single flow is generated at ~24 s from the start of shaking. The granular flows are generated in unpredictable places (not necessarily in the direction of the vibrations) and at unpredictable times (even if most of them, but not all of them, occur not long after the application of the vibrations). This unpredictability (in terms of size and shape of the flows as well) occurs also in different experimental runs with the same vibration frequency and the same vibration amplitude. This is so because, even if the sand cones have been rebuilt after each test with the same identical size, shape and position, some
Fig. 10. Velocity field obtained by PIV analysis of a flow showing that the erosive scarp (on the left) migrates upward during the downflow movement of the flow front (on the right). The values in seconds are the time distance from the first frame. The value in millimetres is the vibration amplitude and the value in hertz is the vibration frequency. This is not a resonance frequency.
Fig. 9. Speed and acceleration of the front of a flow versus time as measured by the PIV technique from flow release to final deposition. The value in millimetres is the vibration amplitude and the value in hertz is the vibration frequency. This is not a resonance frequency.
differences are unavoidable. For example, each cone has its own different and unique spatial distribution of the sand particles. In the case of the critical frequency, all the slopes of the sand cone are simultaneously set in motion (as visible, for example, in the 3 s frame of Fig. 4b) so that the entire cone collapses and its final basal shape is not anymore circular as it was before the application of the vibrations (Figs. 4b and 5). The collapsed cones acquire a final smaller slope inclination which is not affected by further vibrations (as visible, for
88
B. Cagnoli et al. / Journal of Volcanology and Geothermal Research 306 (2015) 83–89
example, in the 24 s frame of Fig. 4b, where no more surface activity is detected). During the cone collapse, flows are generated with the erosive scarp behind them which migrates backward (an example of such a scarp is indicated by the arrow in Fig. 5b). In this case, the flows can be larger than those due to non-critical frequencies. With vibration amplitudes equal to 0.5 and 1 mm, the critical frequency is ~ 10 Hz. Unfortunately, when the amplitude of the vibrations is equal to 1.5 mm, our apparatus does not allow frequencies larger than 8 Hz, so that the resonance frequency is not obtained. Non-critical frequencies generate single flows without the collapse of the entire cone. The movies show that these granular flows are originated in most of the cases at the top of the slope (Fig. 6) where the sand starts moving before the flow proper is released. These early movements are pointed out by arrows in Fig. 7 which illustrates speed measurements versus time in fixed spots along the flow path as obtained by the PIV technique. In Fig. 7, the flow is released in positions 1 and 2 (red and blue curves) where the early movements have occurred. After the flow generation, the arrival of the flow front in spots 3 to 6 is marked by a sudden increase of speed (Fig. 7). A noteworthy characteristic of these flows is that their front is faster than their rear and that the speed of the sand particles decreases gradually toward the rear along the longitudinal axis of the flows. This is shown in Fig. 8, which illustrates simultaneous speed measurements obtained by means of the PIV technique in different positions along the flow path. The dynamics of the front of the flows presents a complicated behaviour. Fig. 9, for example, shows that the speed of the flow front first increases to reach a maximum value that stays constant for a while, then it decreases and increases again to reach a second peak and finally the speed decreases and the flow stops. During the downward motion of the flow, the sand along the scarp rim keeps moving into the depression left behind by the flow since the scarp has too steep and unstable surfaces. This results into a lateral spreading and a backward migration of the scarp in a direction opposite with respect to that of the flow front (Fig. 10). The transversal width of the scarp rim can thus become larger at the top than more downslope. This, with the relatively large width of the final deposit at the base of the cone, generates an hourglass outline of the combined profiles of scarp and deposit (Fig. 6). The flows stop when their front reaches the horizontal table top because of the abrupt change of slope. During their formation, the final deposits accrete rearward because the granular material accumulates at their back (Fig. 6). However there are also instances when the flow does not reach the base but it stops before. 4. Discussion Some of the pyroclasts deposited during the 1944 eruption on the Vesuvius cone were loose (i.e., unwelded) as suggested by the hourglass outline of the combined profiles of the scarp and deposit of the flow within the ellipse in Fig. 1b (as in the dry sand flow case of Hungr et al., 2014). This outline has been obtained in our laboratory cones (Fig. 6), but also, for example, in laboratory sand dunes (Sutton et al., 2013), where, in both cases, the sand is non-cohesive. Hourglass profiles are compatible with loose granular material where the avalanches are fed through scarp recession. Scarp recession (visible, for example, in Fig. 10) is a mechanism that is likely to have occurred on the Vesuvius cone as well. Scarp recession is known also in nonvolcanic settings (e.g., Ventura et al., 2011). A resonance frequency which causes the instantaneous collapse of the entire cone can be important in nature too (McNutt, 1992). We wonder whether the Vesuvius cone during its 1944 eruption has been shaken by its resonance frequency. It is true that there are flows on all the flanks of Vesuvius (Fig. 1), but they may simply be the summation of separate flows generated over a relatively long period of time by frequencies different from the critical one. In case a comparison with natural values is attempted, it is important to realise that the critical frequency obtained in the laboratory is the resonance frequency of not
just the sand cone alone, but that of the cone, tables and tube together (the tube transmits the vibrations to the sand cone and the sand cone transmits the vibrations to the large table). In any case, the noteworthy feature is that single granular flows can be generated by all the vibration frequencies and all the vibration amplitudes adopted in our experiments (i.e., a relatively large range of values). The occurrence of flows at all frequencies and amplitudes can be due to the fact that the energy threshold to trigger this type of instabilities is relatively small in value. This point is important because, if this is true in nature as well, a pyroclastic cone is always potentially dangerous when shaken. In our laboratory experiments, the compaction of the sand decreases toward the cone surface because this surface is manually reworked after each experimental run to rebuild the slope, whereas the inner part of the cone is left untouched. This decrease is expected also in natural volcanic cones as a consequence of the rapid accumulation of loose pyroclasts during an eruption and the natural increase of compaction with depth. The uncompacted superficial granular material is more likely to fail when the ground is shaken by volcanic tremor or tectonic earthquakes. The gradual decrease of the sand speed toward the rear along the flow axis (Fig. 8) is due to the fact that, since all portions of a flow are affected by a downslope acceleration, those that have left earlier (the frontal ones) have experienced the acceleration for a longer time and they have acquired, at the same moment in time, a speed larger than that of those that have left later (the rear ones). Thus, a trend such as that in Fig. 8 suggests that a flow is fed through scarp recession where the granular material is released gradually. On the other hand, multiple accelerations and decelerations of the flow fronts during the flow descent (Fig. 9) can be due to slope irregularities whose small local departures from the angle of repose affect the flows because they are relatively thin (b1 cm thick). Therefore, interestingly, the inclination of the cone slope can differ locally (albeit by small amounts) from the nominal value of the angle of repose. These departures can be the result of grain heterogeneities, differences of granular compaction as well as the manual reworking of the sand cones at the end of each experimental run. Furthermore, the change of slope at the base of the sand cone is more abrupt than on Vesuvius, so that the laboratory flows are unable to propagate behind the base of the cone and the flow forms a deposit as soon as its front reaches the horizontal table (Fig. 6). This does not happen on Vesuvius where the flows can travel further than the base of its cone on the flanks of Mount Somma that are not horizontal (as visible in the southern sector of Fig. 1). Therefore, the mobility of these flows depends also on the slope inclination of the ground surrounding the cone as well as on their grain size and on their volume (Cagnoli and Romano, 2012a).
5. Conclusions We conclude that pyroclastic cones can be hazardous when experiencing ground shaking because of the generation of granular flows on their slopes. This is true during an eruption because of volcanic tremor, but it can be true also after the volcanic activity has ceased as a consequence of distant tectonic earthquakes. Ground shaking can make a cone unstable even if its flanks have an inclination equal to the angle of repose of the granular material (as observed on Vesuvius by Hazlett et al., 1991, and in our laboratory experiments by us), but its instability certainly increases as the volume of rapidly accumulated material that exceeds the angle of repose becomes larger. Importantly, our experiments suggest that there is a wide range of frequencies (and not just a single value) that can generate granular flows. We believe that this is due to the fact that the energy threshold that triggers these instabilities is relatively small in value and, if this is true in nature as well, pyroclastic cones are always potentially dangerous when shaken.
B. Cagnoli et al. / Journal of Volcanology and Geothermal Research 306 (2015) 83–89
Acknowledgements We thank Andreas Lardon for the laboratory data collection at the Università La Sapienza in Rome where he worked for his DiplomIngenieur thesis (Technische Universität, Dresden). We are also grateful to Catello Acerra and Thorossian William from the INGV in Rome. Catello Acerra has designed and built the vibrating table with its mechanical parts, electronics and controlling software. Thorossian William has designed and manufactured the circuit board. We thank also Domenico Pietrogiacomi and Lucilla Monteleone for their assistance in the laboratory and we thank the reviewers for their comments. References Acocella, V., 2005. Modes of sector collapse of volcanic cones: insights from analogue experiments. J. Geophys. Res. 110, B02205. http://dx.doi.org/10.1029/2004JB003166. Anastasopoulos, I., Georgarakos, T., Georgiannou, V., Drosos, V., Kourkoulis, R., 2010. Seismic performance of bar-mat reinforced-soil retaining wall: shaking table testing versus numerical analysis with modified kinematic hardening constitutive model. Soil Dyn. Earthq. Eng. 30, 1089–1105. Cagnoli, B., Romano, G.P., 2012a. Effects of flow volume and grain size on mobility of dry granular flows of angular rock fragments: a functional relationship of scaling parameters. J. Geophys. Res. 117, B02207. http://dx.doi.org/10.1029/2011JB008926. Cagnoli, B., Romano, G.P., 2012b. Granular pressure at the base of dry flows of angular rock fragments as a function of grain size and flow volume: a relationship from laboratory experiments. J. Geophys. Res. 117, B10202. http://dx.doi.org/10.1029/ 2012JB009374. Cas, R.A.F., Wright, J.V., 1988. Volcanic Successions. Allen and Unwin, London. Di Florio, D., Di Felice, F., Romano, G.P., 2002. Windowing, re-shaping and re-orientation interrogation windows in particle image velocimetry for the investigation of shear flows. Meas. Sci. Technol. 13 (7), 953–962. Di Roberto, A., Bertagnini, A., Pompilio, M., Bisson, M., 2014. Pyroclastic density currents at Stromboli volcano (Aeolian Islands, Italy): a case study of the 1930 eruption. Bull. Volcanol. 76, 827. Hazlett, R.W., Buesch, D., Anderson, J.L., Elan, R., Scandone, R., 1991. Geology, failure conditions, and implications of seismogenic avalanches of the 1944 eruption at Vesuvius, Italy. J. Volcanol. Geotherm. Res. 47, 249–264.
89
Hungr, O., Leroueil, S., Picarelli, L., 2014. The Varnes classification of landslide types, un update. Landslides 11, 167–194. Katz, O., Aharonov, E., 2006. Landslides in vibrating sand box: what controls types of slope failure and frequency magnitude relations? Earth Planet. Sci. Lett. 247, 280–294. McNutt, S.R., 1992. Volcanic Tremor. Encyclopedia of Earth System Science. Academic Press, pp. 417–425. Merle, O., Borgia, A., 1996. Scaled experiments of volcanic spreading. J. Geophys. Res. 101 (B6), 13805–13817. Merle, O., Michon, L., Bachèlery, P., 2008. Caldera rim collapse: a hidden volcanic hazard. J. Volcanol. Geotherm. Res. 177, 525–530. Pappalardo, L., D'Auria, L., Cavallo, A., Fiore, S., 2014. Petrological and seismic precursors of the paroxismal phase of the last Vesuvius eruption on March 1944. Sci. Rep. 4, 6297. http://dx.doi.org/10.1038/srep06297. Pudasaini, S.P., Hutter, K., 2006. Avalanche Dynamics. Springer, Berlin. Raffel, M., Willert, C.E., Wereley, S.T., Kompenhans, J., 2007. Particle Image Velocimetry. Springer, Berlin. Santacroce, R, Sbrana, A. (Eds.), 2003. Geological Map of Vesuvius. S.EL.CA, Firenze. Scott, K.M., Macías, J.L., Naranjo, J.A., Rodrigues, S., McGeehin, J.P., 2001. Catastrophic debris flows transformed from landslides in volcanic terrains: mobility, hazard assessment and mitigation strategies. US Geological Survey Professional Paper 1630 (59 pp.). Siebert, L., Glicken, H., Ui, T., 1987. Volcanic hazards from Bezymianny- and Bandai-type eruptions. Bull. Volcanol. 49, 435–459. Sohn, Y.K., Chough, S.K., 1993. The Udo tuff cone, Cheju Island, South Korea: transformation of pyroclastic fall into debris fall and grain flow on a steep volcanic cone slope. Sedimentology 40, 769–786. Sutton, S.L.F., McKenna Neuman, C., Nickling, W., 2013. Avalanche grainflow on a simulated aeolian dune. J. Geophys. Res. Earth Surf. 118, 1–10. http://dx.doi.org/10.1002/jgrf. 20130. Tibaldi, A., 1995. Morphology of pyroclastic cones and tectonics. J. Geophys. Res. 100, 24521–24535. Tommasi, P., Baldi, P., Chiocci, F.L., Coltelli, M., Marsella, M., Romagnoli, C., 2008. Slope failures induced by the December 2002 eruption at Stromboli Volcano. In: Calvari, S., Inguaggiato, S., Puglisi, G., Ripepe, M., Rosi, M. (Eds.), The Stromboli Volcano: An integrated Study of the 2002–2003 Eruption. American Geophysical Union, Washington, D.C., pp. 129–145. Ventura, G., Vilardo, G., Terranova, C., Sessa, E.B., 2011. Tracking and evolution of complex active landslides by multi-temporal airborne LiDAR data: the Montaguto landslide (Southern Italy). Remote Sens. Environ. 115, 3237–3248.